Immunolocalization of Hsp60 in Legionella pneumophila

Abstract

One of the most abundant proteins synthesized by Legionella pneumophila, particularly during growth in a variety of eukaryotic host cells, is Hsp60, a member of the GroEL family of molecular chaperones. The present study was initiated in response to a growing number of reports suggesting that for some bacteria, including L. pneumophila, Hsp60 may exist in extracytoplasmic locations. Immunolocalization techniques with Hsp60-specific monoclonal and polyclonal antibodies were used to define the subcellular location and distribution of Hsp60 in L. pneumophila grown in vitro, or in vivo inside of HeLa cells. For comparative purposes Escherichia coli, expressing recombinant L. pneumophila Hsp60, was employed. In contrast to E. coli, where Hsp60 was localized exclusively in the cytoplasm, in L. pneumophila Hsp60 was predominantly associated with the cell envelope, conforming to a distribution pattern typical of surface molecules that included the major outer membrane protein OmpS and lipopolysaccharide. Interestingly, heat-shocked L. pneumophila organisms exhibited decreased overall levels of cell-associated Hsp60 epitopes and increased relative levels of surface epitopes, suggesting that Hsp60 was released by stressed bacteria. Putative secretion of Hsp60 by L. pneumophila was further indicated by the accumulation of Hsp60 in the endosomal space, between replicating intracellular bacteria. These results are consistent with an extracytoplasmic location for Hsp60 in L. pneumophila and further suggest both the existence of a novel secretion mechanism (not present in E. coli) and a potential role in pathogenesis.

Chaperonins comprise two groups of related multifunctional proteins (20), the Hsp60s of bacteria, mitochondria, and plastids and the CCT-like proteins of the archaea and eucarya (31, 56). Chaperonins prevent aggregation and promote folding of nonnative proteins through an ATP-dependent process. They also assist in the assembly of multisubunit protein complexes and the targeting of proteins for membrane translocation (15, 31). Hsp60 chaperonins are also heat shock- or stress-induced proteins, since their cellular levels increase dramatically following thermal stress or other environmental insults, an essential protective response of all forms of life (9, 37).

Based largely on studies of GroEL in Escherichia coli and mitochondrial Hsp60 in Saccharomyces cerevisiae, Hsp60s are believed to reside in the cytoplasm (matrix or stroma in organelles) (9, 46). Furthermore, no member of the GroEL family possesses a leader sequence or other recognizable motifs that would suggest a secretory role. However, a growing number of reports indicate an extracytoplasmic location for chaperonins (26, 28–30, 34, 45–47, 54), raising the possibility of unique mobilization mechanisms specific for Hsp60 and perhaps novel biological functions for this highly conserved group of proteins. Indeed, the recent description of chaperonin filaments (52) and a novel membrane-stabilizing lipochaperonin activity (50) have expanded chaperonin function beyond protein folding and assembly.

The Hsp60 of the human pneumonic pathogen Legionella pneumophila, like the Hsp60s of many other human microbial pathogens, was initially investigated because of its immunodominant properties (44). However, subsequent studies revealed several fundamental differences between the L. pneumophila Hsp60 and those of other bacteria, particularly E. coli. These include the following: (i) steady-state high basal levels of Hsp60 that increase only twofold following heat shock, as compared to a 20-fold increase observed for GroEL in E. coli (1, 23, 33); (ii) a delay in the return of Hsp60 synthesis to baseline levels following the removal of heat-shock stress (33); (iii) early up-regulation of Hsp60 during association of L. pneumophila with eukaryotic host cells (13); (iv) abundant synthesis of Hsp60 throughout the course of intracellular infection (1, 13, 24); and (v) apparent association of Hsp60 with both membranes and the bacterial cell surface (13, 15, 22, 24, 33, 48).

Surface-exposed Hsp60 has been reported in Mycobacterium leprae (19), Salmonella typhimurium (11), and Helicobacter pylori (10, 40, 57). In Bordetella pertussis, Pseudomonas fluorescens, and Pseudomonas aeruginosa surface exposure has been inferred by experiments in which whole cells were used to remove cross-reactive Hsp60 antibodies from an L. pneumophila Hsp60 antiserum (41). Interestingly, in those mucosal pathogens for which Hsp60 is suggested to be surface exposed, the protein is also implicated in attachment and/or immune modulation activities (10, 11, 25, 40, 42).

To begin to understand the role of Hsp60 in pathogenesis, we first set out to define the subcellular location of Hsp60 in L. pneumophila. Here, we detail the specificity for monoclonal (MAb) and polyclonal antibodies (PAb) employed in this study and their subsequent use to distinguish fundamental differences between the subcellular locations of Hsp60 in L. pneumophila and in a strain of E. coli expressing recombinant Hsp60. Localization patterns of known surface components were compared to patterns of L. pneumophila Hsp60. Finally, immunolabeling of infected HeLa cells was used to demonstrate the accumulation of Hsp60 within L. pneumophila-laden endosomes. These studies suggest that L. pneumophila may possess a novel mechanism for transporting Hsp60 to the periplasm, as well as for facilitating the release of Hsp60 once the pathogen is within host cells.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

L. pneumophila Philadelphia 1 (SVir) and the clinical isolate Lp2064 have been previously described (13, 23). These strains were grown on ACES [N-(2-acetamido)-2-aminoethanesulfonic acid]-buffered charcoal yeast extract (BCYE) agar (38) at 37°C in a humid incubator. For liquid culture, BYE broth (48) was used. Heat shock was performed on L. pneumophila grown at 30°C, as described previously (23, 33). E. coli PSH16 (E. coli JM109 harboring the L. pneumophila htpAB operon), as well as its growth and overexpression of htpB, have also been described previously (23). B. pertussis F6321 was obtained from the Centers for Disease Control and Prevention (Atlanta, Ga.) and was grown on BCYE agar at 37°C.

Infection of HeLa cells.

Monolayers of HeLa cells were grown at 37°C under a 5% CO₂ atmosphere. Monolayers were established in 25-cm² cell culture flasks (Falcon) containing 7 ml of minimal essential medium (MEM) with Earle’s salts, 10% (vol/vol) newborn calf serum, and antibiotics (100 U of penicillin, 100 μg of streptomycin, and 0.25 μg of amphotericin B per ml) (all of these reagents were obtained from GIBCO). The cells were washed twice with 10 mM phosphate-buffered saline (PBS) (140 mM NaCl), pH 7.4, incubated for 1 h in complete MEM without antibiotics, and infected with a 1-ml suspension of Lp2064 in MEM adjusted to an optical density (OD) of 1.0 (7.5 × 10⁸ to 10 × 10⁸ bacteria per ml). OD was measured at 620 nm (OD₆₂₀) in microcuvettes with a light path of 1 cm. After an overnight incubation, infected cultures were washed with PBS and fresh medium without antibiotics was added. Two days after infection, the cells were detached with 0.2% (wt/vol) trypsin in PBS, harvested by centrifugation (300 × g, 10 min), and prepared for electron microscopy as described below.

Antibodies.

Preparation of rabbit hyperimmune serum against L. pneumophila Hsp60 (subsequently referred to simply as “PAb”) has been previously described (23), as has the preparation of the Hsp60 MAb GW2X4B8B2H6 (subsequently referred to simply as “MAb”) (22). Control antibodies included the following: (i) rabbit anti-L. pneumophila major outer membrane protein (OmpS) (6) and rabbit anti-L. pneumophila serogroup 1 (which recognizes lipopolysaccharide [8]) (Centers for Disease Control and Prevention), both used to display the distribution of surface molecules in L. pneumophila; (ii) anti-histone-like, sperm-specific protein Φ2B from mussels (a gift from J. Ausio, University of Victoria), used as an irrelevant polyclonal rabbit antiserum; and (iii) MAb 8-1, raised against the major outer capsid protein of porcine rotavirus (received as neat ascites fluid from Eric Nelson, South Dakota State University), used as an irrelevant antibody.

SDS-PAGE and immunobloting.

Recombinant Hsp60 was purified from PSH16 by a combination of (NH₄)₂SO₄ precipitation and ion exchange chromatography as described previously (39, 55). Approximately 30 μg of purified Hsp60, or cell pellets of L. pneumophila 2064 or E. coli PSH16 (from 1 ml of a suspension with an OD₆₂₀ of 1.0), was solubilized in 100 μl of sample buffer, and 30 μl per lane was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (32) in a 7.5 to 15% (wt/vol) polyacrylamide gradient gel. Immunoblotting procedures were performed as described by Towbin et al. (51) with a Bio-Rad electrotransfer apparatus. Blotted proteins were immunolabeled with PAb or MAb at 37°C. Briefly, membranes were blocked in TTBS (50 mM Tris, 5 mM EDTA, 150 mM NaCl, 0.05% [vol/vol] Tween 20 [pH 7.6]) containing 1% (wt/vol) skim milk and 1% (wt/vol) bovine serum albumin (BSA). Antibodies were diluted in TTBS–0.1% (wt/vol) BSA. PAb was diluted 1:500 and MAb was used as neat hybridoma cell culture supernatant. The secondary antibody (alkaline phosphatase conjugate of anti-rabbit or anti-mouse immunoglobulin G [IgG]) (Cedarlane Laboratories Ltd.) was diluted 1:2,000. Labeled proteins were developed in 10 ml of alkaline phosphatase buffer (100 mM Tris, 100 mM NaCl, 50 mM MgCl₂ [pH 9.5]) in the presence of 1.6 mg of 5-bromo-4-chloro-3-indolylphosphate and 3.3 mg of nitroblue tetrazolium.

Immunoprecipitation.

Pellets of L. pneumophila 2064 or E. coli PSH16 (from 5-ml broth cultures with an OD₆₂₀ of 0.7 to 0.8) were suspended in 1 ml of ice-cold RIPA buffer (50 mM Tris, 150 mM NaCl, 1% [vol/vol] Nonidet P-40, 0.5% [wt/vol] sodium deoxycholate, 0.1% [wt/vol] SDS, 0.1% [wt/vol] sodium azide, 1 mM phenylmethanesulfonyl fluoride [pH 7.5]) and sonicated on ice in three 1-min periods. Each bacterial cell lysate, or 1 ml of a solution of purified Hsp60 (33 μg/ml) in RIPA buffer, was mixed with 50 μl of preswollen protein A-agarose beads (Sigma) at 50% (vol/vol) in RIPA buffer and 20 μl of PAb. The mixture was agitated overnight at 4°C and then for 2 h at 37°C before the spent lysate was removed and the beads were thoroughly washed with ice-cold RIPA buffer. Immunoprecipitated proteins were subjected to SDS-PAGE and immunoblotting with MAb as described above. Alternatively, L. pneumophila suspended at an OD₆₂₀ of 0.5 in 10 ml of diluted BYE broth (1:10 in deionized water) was radiolabeled for 2 h at 37°C with 100 μCi of [³⁵S]methionine (NEN Life Science Products, Boston, Mass.). The labeled cells were then lysed, immunoprecipitated, and subjected to SDS-PAGE as described above. After electrophoresis the gel was fixed for 1 h in 10% (vol/vol) acetic acid, soaked in 1 M sodium salicylate for 30 min, and dried. An autoradiogram was obtained by exposing a sheet of Reflection autoradiography film (NEN Life Science Products) for 10 days.

Specimen preparation for electron microscopy.

In vitro-grown bacteria or infected HeLa cells were fixed in 4% (wt/vol) freshly depolymerized paraformaldehyde and 0.5% (vol/vol) glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.2, and postfixed in 0.25% (wt/vol) aqueous uranyl acetate to stabilize phospholipids and enhance membrane contrast (4). For in vitro-grown cells, we used an epoxy resin embedment (TAAB 812; Marivac Ltd.) previously shown to provide optimal specimen contrast and good antigen accessibility (12, 13). Infected HeLa cells were embedded in the acrylic resin LR White because it penetrated bacteria-laden endosomes better than TAAB 812. LR White was polymerized in gelatin capsules at 50 to 55°C for 24 h under vacuum, and TAAB 812 was polymerized at 60°C for 24 to 48 h. Ultrathin sections were picked up on 200-mesh nickel grids for subsequent immunolabeling.

Postembedding immunolabeling.

Immunolabeling was carried out at room temperature according to the basic method reported by Fernandez et al. (13). Briefly, the grids were blocked on drops of PBS containing 1% (wt/vol) BSA. Washing and antibody dilution were done in PBS containing 0.1% (wt/vol) BSA. PAb and rabbit anti-OmpS serum were diluted 1:400, whereas MAb was used as neat hybridoma cell culture supernatant. Freeze-dried anti-serogroup 1 L. pneumophila antibody was reconstituted as directed on the label and was used undiluted. The secondary antibody (anti-mouse IgG or anti-rabbit IgG conjugated to 10-nm colloidal gold spheres [Sigma Immunochemicals]) was routinely diluted 1:100. Unbound antibodies were washed off by sequentially floating and agitating the grids (in periods of 10 min) on 1 ml of PBS–0.1% BSA in series of three wells in 24-well plates. Extended labeling with the primary antibody (overnight mild agitation at 4°C, followed by 2 h at 37°C, in wells of 24-well plates containing ∼300 μl of the primary antibody) was particularly useful for labeling with MAb. After completion of the labeling procedure, the specimens were fixed by floating the grids on drops of 2.5% (vol/vol) glutaraldehyde in PBS for 10 min, repeatedly washed on drops of deionized water, and then stained with 2% (wt/vol) aqueous uranyl acetate and a modified Sato’s lead stain (21). Control experiments included mock labeling with gold conjugates in the absence of the primary antibody (incubation in the primary antibody was replaced by incubation in PBS–0.1% BSA) or labeling with the irrelevant antibodies diluted 1:400 in either PBS–0.1% BSA (anti-Φ2b) or complete MEM (MAb 8-1). Also, PAb premixed with purified recombinant Hsp60 was used (purified Hsp60 was added to the antibody used for routine labeling [already diluted 1:400 in PBS–0.1% BSA] to a final concentration of ∼50 μg/ml and incubated for 1 h with agitation before being used in the labeling experiment).

Labeling of whole intact cells mounted on Formvar-coated copper grids was performed by the basic method described by Fernandez et al. (13).

Relative analysis of gold-labeling patterns.

Grids were examined in a Philips EM300 transmission electron microscope at an accelerating voltage of 60 kV. Observations of HeLa cell-grown L. pneumophila were restricted to actively growing bacteria (contained in replicative endosomes), easily distinguished from nonreplicating, mature bacteria by well-defined morphological traits, e.g., the presence of a thick envelope layer and large inclusions in the mature bacteria (18). Micrographs of randomly selected fields from the labeled specimens were taken, and prints were made at a defined magnification (usually ×39,000) for further relative analysis. Unless otherwise stated, 30 bacterial sections were analyzed for each labeling condition in every experiment.

To facilitate a relative comparison of labeling patterns, the subcellular distribution of gold particles was standardized to the dimensions of a “typical” bacterial section, calculated for each bacterial species or Legionella strain included in our studies. Four measurements were taken on each bacterial section as indicated in Fig. 1, and the number of gold particles was counted for each bacterial section in the following compartments: cytoplasm, cytoplasmic membrane, periplasm, and outer membrane-cell surface. Gold particles touching the cytoplasmic membrane from the cytoplasm side were counted as belonging to the cytoplasmic membrane, whereas those touching the cytoplasmic membrane from the periplasm side were counted as belonging to the periplasm. Gold particles touching the outer membrane from the periplasm side were counted as belonging to the periplasm. Particles on the outer membrane, or touching it from the outside, were counted as belonging to the outer membrane-cell surface. Because the primary and secondary antibody-gold conjugate may span a distance of ∼20 nm, particles in the extracellular space that were not touching the outer membrane, but were separated from it by two gold particle diameters or less were still counted as belonging to the outer membrane-cell surface.

FIG. 1.

Schematic representation of a bacterial cell section indicating the four measurements (x_c, y_c, x_p, and y_p) taken on each section analyzed. The formulas used to calculate the areas and perimeters of cytoplasmic and periplasmic compartments are indicated below the drawing. A_c, area of the cytoplasm; L_c, length of the cytoplasmic membrane; A_p, area of the periplasmic space; L_p, length of the outer membrane.

The data generated by measuring the bacterial cell sections and counting gold particles in the different compartments was entered in a Microsoft Excel 4.0 worksheet (one worksheet was used per labeling condition) programmed to perform the following calculations: (i) estimate, per section analyzed, the apparent area occupied by the cytoplasm and the periplasm, as well as the length of the cytoplasmic and outer membranes, according to the formulas shown in Fig. 1; (ii) calculate the number of gold particles per square micrometer of cytoplasm or periplasm and the number of gold particles per micrometer of cytoplasmic or outer membrane; (iii) average the corresponding compartment sizes from all bacterial sections analyzed (usually 30 sections per labeling condition); and (iv) average the numbers of gold particles per unit of area (square micrometer, for the cytoplasm and periplasm) or unit of length (micrometer, for the cytoplasmic and outer membranes) from all the sections analyzed. The size of the typical section was calculated by averaging all the compartment sizes [from (iii) above and including all labeling conditions] for each bacterial species or Legionella strain included. Then, the number of gold particles per typical compartment was calculated for each labeling condition by multiplying the average number of gold particles per unit of area or length [from (iv) above] by the corresponding size of each typical compartment. Finally, the percent of gold particles in each typical compartment was calculated with respect to the total number of particles per typical section.

RESULTS

Specificity of antibodies.

A major concern with immunolocalization techniques, particularly those using polyclonal sera, is antibody specificity. The high specificity of MAb has been previously assessed in immunoblots of one- or two-dimensional gels (1, 22, 23). In contrast to MAb, our L. pneumophila Hsp60-specific PAb recognizes both epitopes exclusive to the Legionella Hsp60 and epitopes common to other bacterial Hsp60s (including GroEL) (23), as is the case with other L. pneumophila Hsp60-specific polyclonal antisera (41). PAb immunoprecipitated Hsp60 from crude cell lysates of E. coli PSH16 and L. pneumophila SVir, as detected by immunoblotting with MAb (Fig. 2a). Immunoprecipitates from E. coli PSH16 included a series of Hsp60 degradation products (Fig. 2a, lane 1) that were also labeled in immunoblots of cell lysates (see below). Immunoprecipitates from L. pneumophila (Fig. 2a, lane 2) contained Hsp60 as well as a smear of high-molecular-weight material that was also labeled in cell lysates (Fig. 2b, lanes 2 and 4). Autoradiography showed several protein bands, although Hsp60 itself was not efficiently immunoprecipitated from L. pneumophila cell lysates, as judged by the large amount of Hsp60 remaining in the spent lysate (not shown).

FIG. 2.

Immunospecificity of anti-Hsp60 reagents. (a) Immunoblots developed with MAb of the material immunoprecipitated by PAb from whole-cell lysates of E. coli PSH16 (Ec; lane 1) or L. pneumophila SVir (Lp; lane 2). (b) Immunoblots of whole-cell lysates of E. coli PSH16 or L. pneumophila SVir, developed with PAb (lanes 1 and 2) or MAb (lanes 3 and 4). The positions and molecular weights (in thousands) of broad-range, prestained protein markers (New England BioLabs, Beverly, Mass.) are indicated on the left side of each panel. The open arrowhead on the right side of each panel indicates the position at which purified recombinant Hsp60 migrated.

In immunoblots of whole-cell lysates of E. coli PSH16 or L. pneumophila SVir, PAb and MAb showed virtually identical patterns. Both antibodies strongly labeled a single band that migrated to the position of purified Hsp60 (Fig. 2b). In addition, a series of bands in immunoblots of E. coli PSH16 (Fig. 2b, lanes 1 and 3) likely represented proteolytically degraded Hsp60 and/or truncated recombinant Hsp60 (23), since they were not present in immunoblots of E. coli JM109. PAb did not cross-react with the 28- and 31-kDa OmpS subunits of L. pneumophila, as indicated in Fig. 2b, lane 2, or in immunoblots against purified OmpS (not shown), nor did it cross-react with lipopolysaccharide (Fig. 2b, lane 2).

Labeling patterns of in vitro-grown L. pneumophila. (i) PAb.

PAb labeling clearly showed that Hsp60 epitopes were predominantly found in association with the cell envelope and cell surface of L. pneumophila (Fig. 3a). This labeling pattern conformed to those obtained with rabbit antisera against OmpS or lipopolysaccharide, two well-characterized surface molecules of L. pneumophila (Fig. 4). Surface exposure of Hsp60 was further confirmed by the labeling of whole intact SVir cells (Fig. 5). Standardization of results indicated that the majority of epitopes (>70%) recognized by PAb were found in extracytoplasmic locations (Table 1). The results were standardized to the dimensions of the typical L. pneumophila section (averaged from 22 labeling experiments comprising a total of 720 bacterial cell sections), with a cytoplasmic area of 0.15 ± 0.02 μm², a cytoplasmic membrane length of 1.56 ± 0.13 μm, a periplasmic space area of 0.08 ± 0.01 μm², and an outer membrane length of 1.91 ± 0.14 μm. Mock-labeled and irrelevant antibody controls always showed ∼5% of the labeling obtained with PAb. Moreover, this background (nonspecific) labeling was largely restricted to the cytoplasmic area, and the chance for a random gold particle to be found in association with the cell envelope was estimated to be ∼1.3%. Since control labelings were run for every condition shown, background labeling was subtracted from specific PAb labeling to obtain the values presented in Table 1.

FIG. 3.

Comparative labeling of L. pneumophila and control bacteria with PAb. Representative electron micrographs show the labeling patterns of ultrathin sections cut from L. pneumophila SVir (a), E. coli PSH16 (b), and B. pertussis (c) grown at 37°C. Bars, 0.1 μm.

FIG. 4.

Comparative labeling of ultrathin sections of non-heat-shocked L. pneumophila SVir with different rabbit PAbs. Representative electron micrographs show the labeling patterns obtained after an overnight incubation with anti-Hsp60 (a), anti-OmpS (b), or anti-serogroup 1 lipopolysaccharide (c) rabbit sera. Specimens were not stained to facilitate visual recognition of the labeling patterns. Bars, 0.1 μm.

FIG. 5.

Surface expression of Hsp60 in L. pneumophila SVir. Representative electron micrograph showing profuse gold labeling on the surface of an intact, whole bacterial cell grown at 37°C. Bar, 0.1 μm.

TABLE 1.

Distribution of Hsp60 epitopes and OmpS epitopes in typical sections of L. pneumophila SVir as detected by immunoelectron microscopy with various polyclonal rabbit immunoreagents

Position	Distributiona of epitopes as labeled by:
	PAb		PAb + Hsp60		Anti-OmpS
	Non-HSb	HSc	HS	% Inhibitiond	Non-HS	HS
Cytoplasm	11.1 ± 0.6 (26.2)	6.2 ± 1.1 (22.4)	2.8 (30.8)	54.8	6.9 (14.6)	10.0 (21.3)
Cyt. membr.	6.6 ± 1.8 (15.6)	5.0 ± 2.3 (18.0)	1.5 (16.5)	70.0	4.4 (9.3)	4.7 (10.0)
Periplasm	11.6 ± 1.3 (27.4)	5.2 ± 2.1 (18.8)	1.1 (12.1)	78.9	16.6 (35.0)	13.7 (29.2)
OM-surface	13.0 ± 1.6 (30.7)	11.3 ± 0.4 (40.8)	3.7 (40.6)	67.3	19.5 (41.1)	18.5 (39.5)
Total	42.3 ± 0.6 (100)	27.7 ± 1.2 (100)	9.1 (100)	67.1	47.4 (100)	46.9 (100)

^aThe distribution of gold particles standardized to the dimensions of the typical L. pneumophila section is shown in number of particles per cell compartment and as relative percent (in parentheses) with respect to the total number (100%) of particles per section. Cyt. membr., cytoplasmic membrane; OM-surface, outer membrane and surface; Non-HS, non-heat shocked (30°C); HS, heat shocked (42°C for 1 h). The background values for nonspecific labeling (irrelevant antibody controls) have been subtracted from the values shown for each condition. 

^bResults are shown as means ± standard deviations of two independent labeling experiments. Each experiment included the analysis of 30 sections. 

^cResults are shown as means ± standard deviations of three independent labeling experiments. Each of the experiments included the analysis of 30 sections. 

^d% Inhibition, percent reduction of gold particles per compartment compared to that for HS labeled with PAb. 

Heat-shocked bacteria had a reduced total number of PAb epitopes compared to that of non-heat-stressed cells. This overall reduction in Hsp60 epitopes was associated with a relative increase in labeling of the outer membrane and surface. The ratio of gold particles on the surface to those in the cytoplasm in non-heat-shocked bacteria was about 1:1, and in heat-stressed bacteria, it was closer to 2:1. The specificity of PAb labeling was confirmed in heat-shocked specimens by the significant inhibition in labeling (67%) observed after PAb was incubated with purified recombinant Hsp60. It is important to note that this inhibition was not accompanied by a change in the distribution of gold particles, except for a slight increase in the relative amount of cytoplasmic label (Table 1).

The standardized labeling patterns of control and heat-stressed SVir with PAb were further compared to the standardized pattern obtained with the anti-OmpS rabbit serum. In contrast to Hsp60, the total number of OmpS epitopes did not change upon heat stress, and the changes observed in the subcellular distribution of OmpS were reversed with respect to those observed for Hsp60 (Table 1). However, the overall distribution of OmpS epitopes was very similar to that of Hsp60 epitopes in heat-stressed SVir; this was all the more remarkable considering that OmpS is a surface-exposed outer membrane protein (Fig. 4).

(ii) MAb.

Labeling with MAb was very inefficient, and only specimens subjected to extended labeling (overnight incubation with MAb) achieved levels higher than the mock-labeled and irrelevant antibody controls (Table 2). Like the results in Table 1, the values shown in Table 2 were standardized to the dimensions of the typical L. pneumophila cross section and the background values of nonspecific labeling have been subtracted for each condition. Although the relative labeling of the cytoplasm of L. pneumophila by MAb was more prominent than that by PAb, 50 to 78% of the total number of gold particles were still distributed in extracytoplasmic locations (Table 2). Furthermore, in contrast to PAb epitopes, total MAb epitopes slightly increased in heat-shocked bacteria, with an associated increase in the relative labeling of the cytoplasm but not a significant increase in outer membrane-cell surface epitopes (Table 2). It was thus clear that MAb recognized significantly fewer epitopes than PAb, and those recognized after heat shock were mainly located in the cytoplasm.

TABLE 2.

Distribution of Hsp60 epitopes in typical sections of L. pneumophila SVir as detected by immunoelectron microscopy with MAb^a

Position	Distribution of epitopesb
	Non-HS	HS
Cytoplasm	0.50 (22.2)	1.73 (49.4)
Cyt. membr.	0.38 (16.9)	0.30 (8.6)
Periplasm	0.75 (33.3)	0.38 (10.9)
OM-surface	0.62 (27.6)	1.09 (31.1)
Total	2.25 (100)	3.50 (100)

^aFor abbreviations and general organization, see footnote a to Table 1. 

^bOnly results from the protocol with an overnight labeling with the primary antibody are shown. 

Labeling patterns of in vitro-grown E. coli PSH16 and B. pertussis. (i) PAb.

Expression of the L. pneumophila Hsp60 in E. coli led to a predominantly cytoplasmic labeling (Fig. 3b), indicating that surface exposure of Hsp60 epitopes is a peculiarity of L. pneumophila and does not constitute an intrinsic property of the L. pneumophila Hsp60, nor is it a gross artifact of our labeling technique that would preferentially label the cell surface of any bacterial cell. On the other hand, labeling of B. pertussis with PAb showed that B. pertussis Hsp60 associates with the bacterial cell envelope (Fig. 3c), an expected result which is in agreement with previous evidence suggesting surface exposure of Hsp60 epitopes in B. pertussis (41).

The standardized results for E. coli PSH16 indicated that >90% of the total Hsp60 epitopes were confined to the cytoplasm and cytoplasmic membrane (Table 3). These results were standardized to a typical section (averaged from 10 labeling experiments, comprising 300 bacterial sections), with a cytoplasmic area of 0.39 ± 0.11 μm², a cytoplasmic membrane length of 2.84 ± 0.81 μm, a periplasmic space area of 0.12 ± 0.03 μm², and an outer membrane length of 3.17 ± 0.86 μm, and they have been corrected to account for the nonspecific labeling observed in mock-labeled and irrelevant controls. Heat stress led to a two- to threefold increase in total Hsp60 PAb epitopes, but the distribution of epitopes remained virtually unchanged. The standardized distribution of Hsp60 in B. pertussis was very similar to that of non-heat-shocked L. pneumophila: location at the cell envelope was predominant over a cytoplasmic location (Table 3). The results shown for B. pertussis were standardized to a typical section (averaged from 24 bacterial sections from a single experiment), with a cytoplasmic area of 0.20 ± 0.12 μm², a cytoplasmic membrane length of 1.77 ± 0.79 μm, a periplasmic space area of 0.07 ± 0.03 μm², and an outer membrane length of 2.03 ± 0.79 μm.

TABLE 3.

Distribution of Hsp60 epitopes in typical sections of the control species, E. coli PSH16, and B. pertussis F6321 as detected by immunoelectron microscopy with PAb or MAb^a

Position	Distribution of epitopes as labeled by:
	PAb			MAbb
	B. pertussis non-HS	E. coli PSH16		E. coli PSH16
		Non-HSc	HSc	Non-HS	HS
Cytoplasm	6.8 (37.4)	26.5 ± 3.0 (83.9)	62.9 ± 6.1 (85.8)	4.80 (82.2)	6.01 (90.4)
Cyt. membr.	2.0 (11.0)	2.5 ± 0.3 (7.9)	4.4 ± 0.3 (6.0)	0.30 (5.2)	0.26 (3.9)
Periplasm	4.8 (26.4)	1.7 ± 1.1 (5.4)	3.6 ± 0.9 (4.9)	0.27 (4.6)	0.16 (2.4)
OM-surface	4.6 (25.2)	0.9 ± 0.4 (2.8)	2.4 ± 0.8 (3.3)	0.47 (8.0)	0.22 (3.3)
Total	18.2 (100)	31.6 ± 4.8 (100)	73.3 ± 7.5 (100)	5.84 (100)	6.65 (100)

^aFor definitions and general organization, see footnote a to Table 1. 

^bOnly results from the protocol with an overnight labeling with the primary antibody are shown. 

^cResults are shown as the means ± standard deviations of two (Non-HS) or three (HS) independent labeling experiments. Each experiment included the analysis of 30 sections. 

(ii) MAb.

As was observed for the experiments with L. pneumophila, the labeling of PSH16 sections with MAb was very inefficient (Table 3), suggesting that the epitope recognized by MAb (the carboxyl terminus of Hsp60) is not readily available in sections of either L. pneumophila or E. coli. The increase in total MAb epitopes detected after heat shock was negligible, suggesting that the proportion of available MAb epitopes in the increasing population of recombinant Hsp60 decreased upon heat shock. Moreover, the changes in the relative distribution of MAb epitopes upon heat shock were modest and mainly restricted to an increase in cytoplasmic labeling.

Labeling patterns of in vivo-grown L. pneumophila.

As averaged from six labeling experiments representing 180 bacterial sections, the typical section of intracellular Lp2064 had a cytoplasmic area of 0.25 ± 0.09 μm², a cytoplasmic membrane length of 1.88 ± 0.40 μm, a periplasmic space area of 0.10 ± 0.02 μm², and an outer membrane length of 2.20 ± 0.41 μm. The standardized distribution of PAb and MAb epitopes in this typical section of in vivo-grown L. pneumophila (Table 4), was very similar to that of in vitro-grown SVir bacteria (Tables 1 and 2), suggesting that, in spite of the changes observed in the size of the cell compartments, the localization pattern of Hsp60 was conserved. Interestingly, it was apparent that the endosomal space between bacteria displayed many more gold particles than the space between in vitro-grown bacteria. In particular, the endosomal space was strongly labeled after an overnight incubation with PAb (Fig. 6b), as were the surfaces of the replicating intracellular bacteria. We concluded that this dense labeling was specific and did not include HeLa cell Hsp60, because nonendosomal HeLa cell material (including mitochondria), endosomes devoid of bacteria, or the inter-HeLa cell space was not labeled. Strong labeling was exclusively associated with bacteria-laden endosomes. Furthermore, preincubation of PAb with purified recombinant Hsp60 caused a 77% overall reduction in the labeling of replicating intracellular bacteria (Table 4) and an estimated 60% reduction in the labeling of the interbacterial space (not shown). Thus, these results indicated that Hsp60 was released by L. pneumophila organisms growing in the intracellular space.

TABLE 4.

Distribution of Hsp60 epitopes in typical sections of replicating L. pneumophila 2064 contained in endosomes of infected HeLa cells as detected by immunoelectron microscopy with PAb or MAb^a

Position	Distribution of epitopes as labeled by:
	PAbb	PAb + Hsp60	% Inhibitionc	MAbd
Cytoplasm	4.2 ± 2.4 (18.5)	1.2 (23.1)	71.4	8.7 (50.0)
Cyt. membr.	3.5 ± 1.3 (15.4)	0.7 (13.5)	80.0	3.3 (19.0)
Periplasm	6.2 ± 0.1 (27.3)	1.4 (26.9)	77.4	1.9 (11.0)
OM-surface	8.8 ± 2.0 (38.8)	1.9 (36.5)	78.4	3.5 (20.0)
Total	22.7 ± 5.8 (100)	5.2 (100)	77.1	17.4 (100)

^aFor definitions and general organization, see footnote to Table 1. 

^bResults are shown as means ± standard deviations of two independent labeling experiments. Each experiment included the analysis of 30 sections. 

^c% Inhibition, percent reduction of gold particles per compartment labeled with PAb + Hsp60 compared to those labeled with PAb. 

^dOnly results from the protocol with an overnight labeling with the primary antibody are shown. 

FIG. 6.

Distribution of Hsp60 epitopes in ultrathin sections of L. pneumophila 2064 grown in vivo at 37°C. (a) Low-magnification view to show a complex endosome containing replicating bacteria. A few mature forms still contained in a degraded cell are present (arrow), as well as some free mature forms, likely released from a lysed HeLa cell (top right corner). (b) Heavy labeling of the endosomal space between intracellular bacteria, and of the surfaces of replicating intracellular bacteria, after an overnight incubation with PAb. Bars, 1 μm (a) and 0.1 μm (b).

DISCUSSION

Our immunolocalization studies showed that ∼75% of the Hsp60 epitopes detected in sections of L. pneumophila were extracytoplasmic and conformed to a localization pattern typical of surface-expressed molecules. In contrast, nearly 90% of the recombinant Hsp60 expressed by E. coli PSH16 was located in the cytoplasm. Because release of Hsp60 was suggested by the labeling patterns of both heat-shocked L. pneumophila and bacteria-laden endosomes of infected HeLa cells, we have concluded that L. pneumophila, but not E. coli, may possess novel mechanisms for mobilizing Hsp60 to extracytoplasmic locations and the surrounding milieu. It is unlikely that our results were derived from the action of nonspecific or contaminating antibodies because (i) immunoprecipitation and immunoblotting showed that our antibodies possessed similar high specificities for Hsp60, (ii) labeling in mock and irrelevant antibody controls was always minimal, and (iii) a significant reduction in labeling (with no change in relative distribution) was obtained with Hsp60-treated PAb.

The differential efficiency at which Hsp60 was immunoprecipitated from lysates of L. pneumophila (low efficiency) or E. coli PSH16 and solutions of pure protein (high efficiency) could be explained by the propensity of Hsp60 to associate with L. pneumophila membrane fractions (16, 33) and peptidoglycan-OmpS complexes (7). Vestiges of these Hsp60 complexes appeared as high-molecular-weight smears, exclusively labeled in immunoblots of L. pneumophila. This may also explain why Susa et al. (49), using immunoprecipitation, were unable to identify Hsp60 as a prominent antigen expressed during intracellular growth, in spite of the fact that Hsp60 is abundantly synthesized by intracellular L. pneumophila, as shown by autoradiography (1, 13).

The reduced level of cell-associated Hsp60 in heat-shocked bacteria may have resulted from restrained immunorecognition due to increased association of Hsp60 with membranes, mobilization of Hsp60 to the cell surface and secretion, or a combination of the two. Restrained immunorecognition is supported by previous observations reporting a twofold increase of Hsp60 levels in heat-shocked L. pneumophila (23, 33) (i.e., Hsp60 epitopes are abundant but not recognized), as well as increased levels of membrane-associated Hsp60 in heat-stressed Borrelia burgdorferi (45), Synechocystis sp. (29), and isolated chloroplasts (30). Recent experimental evidence has indicated that GroEL interacts with lipid layers through its carboxy terminus, increasing the molecular order of the layers (50). Thus, it has been proposed that GroEL possesses a lipochaperonin activity that transiently stabilizes membranes under stress (50). As the major cytoplasmic membrane protein of L. pneumophila (5, 16), Hsp60 may perform a lipochaperonin function and interact with membranes through its carboxy terminus, accounting for the “blindness” of MAb for membrane-associated Hsp60 (MAb recognizes the carboxy terminus of Hsp60 [23]). Alternatively, MAb-specific epitopes could have been selectively altered in extracytoplasmic Hsp60 during the fixation and processing of specimens for electron microscopy.

The total amount of cell-associated Hsp60 was also low in intracellular L. pneumophila, but in this case, the profuse labeling of the endosomal space with PAb strongly suggested secretion. The following points argue against “altruistic autolysis” (10, 40) as the mechanism of release and surface expression of L. pneumophila Hsp60. First, all our observations were restricted to young, actively growing bacteria. Second, contrary to the case of H. pylori (40), we did not commonly observe labeled cell debris in preparations of L. pneumophila. Third, to conform with altruistic autolysis, surface labeling of L. pneumophila should have been irregular (40); instead it showed quite consistent patterns. Fourth, free Hsp60 did not bind to the surface of L. pneumophila, as determined by immunolabeling of whole, non-heat-shocked cells (results not shown). Whereas our results strongly support secretion of Hsp60 by L. pneumophila, this needs to be unequivocally proved by a combination of biochemical and genetic approaches.

With respect to the functions fulfilled by extracytoplasmic Hsp60, a traditional protein-folding role could be ruled out because, as discussed by Missiakas and Raina (35), ATP, which is essential for this function, is absent from the periplasm and extracellular locations. It has been reported that surface-located Hsp60 mediates adherence of H. pylori (25) and S. typhimurium (11) to host cells, and preliminary results from our laboratory have indicated that Hsp60 also mediates adherence of L. pneumophila to HeLa cells (17). Novel immunomodulatory functions have recently been reported for several Hsp60s (42), but the fact that L. pneumophila Hsp60 modulates macrophage function through a mechanism that involves surface interactions in the absence of Hsp60 internalization (43) suggests that surface-exposed Hsp60 may play an important role in the pathogenesis of Legionnaires’ disease. In this respect it has been observed that during the early infection of macrophages and L929 cells by L. pneumophila, Hsp60 appears to be released into newly formed phagosomes (13). The notion that Hsp60 is secreted by entering and intracellular L. pneumophila is consistent with the observation that Hsp60 is a dominant antigen recognized early by the cellular immune system of patients with Legionnaires’ disease (55). Although cell-associated Hsp60 has been implicated in stress relief in L. pneumophila (1, 2), the functions of the secreted Hsp60 remain largely undefined. Obligate bacterial endosymbionts of aphids constantly overproduce and release an Hsp60 homolog known as symbionin (27, 36, 53), which is believed to play a key role in establishing and maintaining the endosymbiosis rather than in alleviating stress (3). Thus, we propose that, besides its potential stress-alleviating functions, Hsp60 fulfills an important role in supporting the intracellular lifestyle of L. pneumophila. This function, which presumably evolved within the intracellular environment of freshwater protozoa (14), must require dominant synthesis of Hsp60 and its mobilization to extracytoplasmic locations.