INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Legionella pneumophila is the causative agent of Legionnaires' disease, a potentially lethal pneumonia (20), and of Pontiac fever, a self-limiting influenza-like syndrome (28). Growth conditions affect a number of factors that are thought to be related to the virulence of L. pneumophila (1, 2, 8, 9, 33, 39). Although L. pneumophila has been shown to enter and replicate within many different eukaryotic cell types, it is found primarily within monocytes during disease in humans (10, 55) and guinea pigs (11, 15). L. pneumophila grown under different conditions differs in cytotoxicity for, trafficking within, and entry into monocytes (8, 13). Since the ability of L. pneumophila to gain access to and survive within monocytes is an essential pathogenic trait (15, 55), it is likely that growth conditions will affect the ability of this organism to cause disease.

Since growth in protozoa, such as Acanthamoeba castellanii, is thought to reflect the environment growth conditions for L. pneumophila (23, 48), we chose to examine the effects of growth in these amoebae on virulence. Previous studies in our laboratory have demonstrated that growth in environmental amoebae enhances the ability of L. pneumophila to enter epithelial cells (100- to 1,000-fold), murine macrophages (10- to 100-fold), human monocytes (100- to 1,000-fold), and A. castellanii (10- to 100-fold) (13). However, the effects of this growth condition on the mechanism of entry of L. pneumophila into host cells and its role in virulence are not well understood. It has been shown that L. pneumophila enters monocytes through an unusual mechanism, an asymmetric phagocytic event termed coiling phagocytosis (25). Phagocytosis of L. pneumophila is thought to involve the complement receptors CR1 and CR3 and is enhanced by the presence of complement (32, 40). Entry via this mechanism may occur via binding of the major outer membrane protein to complement receptors either directly (30) or through complement (3). Since both conventional and coiling phagocytic events have been observed in L. pneumophila (25, 42) and the frequencies of each have not been quantitated, it is difficult to ascertain the effects of these mechanisms of entry on subsequent intracellular survival. However, recent studies demonstrating that L. pneumophila has the ability to affect very early trafficking events from 5 min (49) to 30 min (52) after entry suggest that entry or other very early events play a role in intracellular survival.

To determine the effects of growing L. pneumophila in amoebae on entry mechanisms, we examined the role of complement opsonization, frequencies of coiling and conventional phagocytosis, and role of cytoskeletal components in entry into monocytes of these bacteria compared to BCYE-grown bacteria. Effects of growth in amoebae on virulence were evaluated by the ability to replicate intracellularly in resting and activated monocytes as well as virulence in the three mouse strains A/J, C57BL/6J Beige, and C57BL/6J. From these data, we conclude that growth in amoebae affects the entry mechanism used for monocytes, enhances intracellular replication in monocytes, and increases virulence in mice.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and culture conditions. The L. pneumophila strain used for these studies was the streptomycin-resistant variant (34) of L. pneumophila serogroup 1 (130b) (18). This L. pneumophila strain is virulent in both in vitro and in vivo models of infection (34) and was passaged no more than twice in the laboratory before use in these studies, to prevent loss of virulence. L. pneumophila was grown either on BCYE agar (16) for 3 days or in the amoeba A. castellanii as described previously (13). The preparation of amoeba-grown bacteria has been improved to increase the number of bacteria isolated and reduce contamination with amoebal debris. These improvements were accomplished by using a multiplicity of infection (MOI) of 100 in the amoebae and by allowing intracellular growth to continue for 48 h at 37°C in 5% CO₂. Potential remaining amoebae and amoebal debris were removed from the preparation by filtration through a 5-µm-pore-size filter (Costar) after centrifugation. Samples prepared in this manner were examined by light microscopy to confirm the absence of intact amoebae. The Escherichia coli K-12 strain HB101 (ara-14 leuB6 proA2 lacY1 glnV44 glaK2 recA13 rpsL20 xyl-5 mtl-1 thi-1 hsdS29) (Promega) was grown to stationary phase in Lennox broth (GIBCO). Bacterial viability was determined when necessary by using the LIVE-DEAD assay (Molecular Probes, Eugene, Oreg.). All bacterial preparations used in these studies were found to be greater than 99% viable.

Cell lines and culture conditions. The amoebae used in these experiments were A. castellanii ATCC 30234. They were grown in PYG broth in 75-cm² tissue culture flasks, and the number of viable cells was determined as described previously (13, 35).

Entry assays. Entry assays were carried out essentially as described previously (13). Assays with human PBMs and activated THP-1 cells were carried out in 1 ml of RPMI in 24-well microtiter dishes (Falcon), using 10⁶ cells/well. The bacteria to be assayed were mixed with an equal volume of RPMI, complete nonimmune human serum (titer of anti-L. pneumophila antibodies, <1:100 [as measured by enzyme-linked immunosorbent assay]), or heat-inactivated human serum, incubated for 10 min at 37°C, and used to infect the cells for 30 min at 37°C with an MOI of 10 to 100 bacteria/cell. The relative differences in entry between amoeba- and BCYE-grown bacteria remained approximately the same (100- to 1,000-fold) over 5- to 120-min infection periods in entry assays (data not shown). Assays with THP-1 cells that had not been activated were carried out in suspension in 1.5-ml microcentrifuge tubes (Applied Scientific) containing 10⁶ cells in 1 ml of RPMI at 37°C. After infection with bacteria, the cells were washed with PBS once (or three times if the assay was to be carried out in the absence of gentamicin). Treatment with 100 µg of gentamicin per ml for 2 h was used to kill extracellular bacteria when necessary. After the initial washes or gentamicin treatment, the cells were washed once more in PBS and then lysed in 1 ml of sterile distilled water for 10 min at room temperature. For each change of solution for assays in suspension, the cells were pelleted by centrifugation for 1 min at 100 × g. After incubation of the cells with water, they were passed through a 27-gauge syringe three times to ensure lysis and dilutions were plated on BCYE (L. pneumophila) or Lennox broth (E. coli) agar to determine bacterial counts (expressed in CFU). Entry levels were determined by calculating the percentage of the inoculum that became gentamicin resistant over the course of the assay [i.e., % entry = 100 × (CFU of gentamicin-resistant cells/CFU of inoculum)]. To correct for variation in levels of uptake between experiments, entry is reported relative to that of L. pneumophila 130b (AA100) (i.e., relative entry = % entry of test strain/% entry of AA100). Similar to previous studies (13), entry levels of amoeba-grown Legionella were approximately 0.4 and 4% of the inoculum in resting and activated monocytes, respectively, under these conditions.

Inhibitor studies. To examine the effect of cellular inhibitors on the entry process, the THP-1 cells were incubated with 2 µM cytochalasin D (Sigma) for 1 h, 10 µM colchicine (Sigma) for 30 min, or 20 µM nocodazole (Sigma) for 30 min to 2 h prior to addition of the bacteria. These inhibitors were left in the assay mixture during the 30-min incubation for entry. The subsequent steps in the entry assay were carried out as described above without the addition of gentamicin. Each inhibitor, as well as the solution in which it was solubilized, was tested for cytotoxicity for the bacteria, as assayed by acridine orange staining (58), and for the host cell, as assayed by uptake of 0.1% eosin Y (Sigma). No significant effects of the inhibitors or solubilization buffer on bacterial or host cell viability were observed (data not shown).

Intracellular viability assays. The THP-1 cells used for viability assays were seeded into 24-well tissue culture dishes (Falcon) containing RPMI plus 10% serum, 5 µg of lipopolysaccharide (E. coli O127:B8; Difco) per ml, and 40 U of human gamma interferon (Boehringer Mannheim) per ml. The cells were incubated for 2 days with this medium, and fresh medium was added for an additional 24 h to allow activation of the monocytes prior to entry. The bacteria were added to the cells and incubated at 37°C for 5 min, washed three times with warm PBS, and suspended in fresh medium with gamma interferon and lipopolysaccharide for various times before lysis with water. Dilutions of the resulting lysates were plated to determined the CFU at each time point. For nonactivated THP-1 cells, the assays were carried out in 1.5-ml microcentrifuge tubes in the same manner as described for entry assays.

Microscopy techniques. Coiling phagocytosis was evaluated as described previously (13, 25). The bacteria were incubated in the presence of RPMI, heat-inactivated serum, or complete serum in the same manner as for the entry assays prior to addition to the cells. For fluorescence microscopy, the samples were fixed at room temperature for 10 min in PBS containing 3.7% formaldehyde. They were then washed twice with PBS, suspended in ice-cold acetone for 3 min, and washed twice more with PBS. At this point, the distribution of tubulin in the cells was examined by using an anti--tubulin antibody (1:200 in PBS [Amersham]) and a secondary anti-mouse immunoglobulin G-fluorescein isothiocyanate-conjugated antibody (1:200 in PBS [Sigma]). Both antibodies were sequentially incubated with the sample for 1 h at room temperature. The cells were then washed three times with PBS and examined by fluorescence microscopy.

Mouse infections. To examine the virulence of amoeba-grown and BCYE-grown L. pneumophila in mice, we used methods described previously (5, 7). C57BL/6J, C57BL/6Jbg^J/bg^J, and A/J mice were infected by intratracheal inoculation with 10⁶ bacteria, and after 2 h and 1, 2, and 4 days the mouse lungs were harvested and the bacteria in the lungs were quantitated as described previously (5, 7). Data presented represent the means and standard deviations of bacterial counts (CFU/gram of lung) for seven mice in each experimental group. Four different bacterial preparations were used: (i) bacteria grown on BCYE agar for 3 days under standard laboratory conditions, (ii) BCYE-grown bacteria mixed with a lysate of A. castellanii, (iii) A. castellanii that had been infected with L. pneumophila for 2 days, and (iv) L. pneumophila that had been previously grown in amoebae for 2 days. Each preparation was equilibrated to 10⁶ bacteria per sample. Preparations i and iv were prepared in the same manner as described for entry assays. Separation of bacteria from amoebae in preparation iv was ensured by passage of the preparation through a 5-µm-pore-size filter (Costar) prior to inoculation. Preparation ii was made by mixing 10⁶ bacteria with a lysate of 10⁶ amoebae. The amoebal lysate was produced by washing 10⁶ amoebae in 1 ml of distilled water and resuspending them in 1 ml of water for 10 min at room temperature. The cells were then vortexed vigorously 10 times for 3 s each. All preparations were suspended in PBS prior to inoculation.

Statistical analyses. All in vitro experiments were carried out in triplicate and repeated three times. The experiments in vivo were carried out with seven mice per experimental group. The significance of the results was analyzed by analysis of variance. P values of < 0.05 were considered significant.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effects of complement on entry of amoeba-grown L. pneumophila. Although L. pneumophila organisms grown in amoebae have been previously shown to enter monocytes at higher levels than do those grown on BCYE (13), the effects of complement on entry of bacteria grown in amoebae were not known. We examined entry of amoeba-grown and BCYE-grown bacteria into PBMs and THP-1 cells by using L. pneumophila that had been preincubated with RPMI, complete human serum, or heat-inactivated serum (Fig. 1). A significant increase in entry into both THP-1 cells (a twofold increase) and PBMs (an eightfold increase) was observed with BCYE-grown bacteria in the presence of complete serum, comparable to results from previous studies (40). In contrast, entry of amoeba-grown bacteria was not affected by the presence of complement. These data suggest that complement opsonization does not play a role in the entry mechanism of amoeba-grown L. pneumophila into monocytes.

View larger version (27K): [in this window] [in a new window]   FIG. 1.   Effects of human serum on entry of BCYE- and amoeba-grown L. pneumophila into THP-1 cells (A) and human PBMs (B). Entry is expressed relative to BCYE-grown AA100 in the absence of serum and is calculated as described in Materials and Methods. Serum was replaced by RPMI in the no-serum control. Error bars represent the standard deviations for triplicate samples in each experiment. The results shown are from a single representative experiment. All experiments were repeated at least three times.

Quantitation of L. pneumophila entry via coiling phagocytosis. Although both conventional and coiling phagocytic events have been observed previously in BCYE-grown L. pneumophila (25, 42), they have not been quantitated. Thus, to determine the frequencies of coiling phagocytosis and evaluate the role of complement in this event, we compared the structural mechanism of entry by L. pneumophila grown in amoebae and BCYE in the presence and absence of complement. We examined 100 entry events and characterized them as conventional or coiling phagocytosis (Fig. 2). BCYE-grown L. pneumophila enters THP-1 cells by coiling phagocytosis between 40 and 50% of the time, but in the presence of complete serum the frequency of coiling phagocytosis decreases to less than 20%. Amoeba-grown L. pneumophila, however, enters by coiling phagocytosis 90 to 98% of the time in both the presence and absence of complete serum. Apparently, complement has a significant effect upon the mechanism of entry into monocytes by BCYE-grown bacteria, but amoeba-grown bacteria can either overcome the effects of or prevent opsonization.

View larger version (31K): [in this window] [in a new window]   FIG. 2.   Frequency of coiling phagocytic entry events triggered by BCYE- and amoeba-grown L. pneumophila in THP-1 cells. The percent coiling phagocytosis is the number of coiling phagocytic events in every 100 entry events observed. E. coli HB101 is included as a conventional phagocytosis control. Serum was replaced with RPMI in the no-serum control. The results shown are from a single representative experiment of three independent experiments.

Cellular inhibitors affect the entry of amoeba-grown and BCYE-grown L. pneumophila differently. Based on the above results, we now have the ability to compare conditions where 90% of L. pneumophila bacteria enter via conventional phagocytosis (BCYE-grown bacteria in the presence of complement) with conditions where 98% enter via coiling phagocytosis (amoeba-grown bacteria in the absence of complement). In addition, amoeba-grown bacteria enter monocytes at a much higher frequency than do BCYE-grown bacteria. Comparison of the effects of different cytoskeletal inhibitors on the entry of each of these bacterial populations should allow us to dissect the cellular mechanisms of the phagocytic events involved. Entry into and growth in monocytes of BCYE-grown L. pneumophila in the absence of complement are microfilament dependent (17, 29). Thus, we focused our studies on the microfilament polymerization inhibitor cytochalasin D and the microtubule polymerization inhibitors colchicine and nocodazole. THP-1 cells were pretreated with each of these inhibitors, and the subsequent entry of L. pneumophila was examined by using the entry assay (Fig. 3). While cytochalasin D completely inhibited HB101 uptake, it had only a modest effect upon entry of BCYE-grown L. pneumophila and less of an effect upon entry of the amoeba-grown bacteria (Fig. 3A). By contrast, colchicine appeared to inhibit the entry of all bacterial strains to the same level (Fig. 3C). We initially tested nocodazole to confirm the results of the colchicine inhibition studies. Surprisingly, we found that nocodazole pretreatment for 30 min at 37°C affected the entry of HB101 and BCYE-grown L. pneumophila in approximately the same manner as did colchicine but, significantly, that entry of amoeba-grown bacteria was not affected (Fig. 3B). In contrast, when the cells were pretreated for 2 h with nocodazole in the same manner, entry of all bacteria was significantly affected, as had been seen for colchicine (Fig. 3D). The effects of these inhibitors on the distribution of tubulin in THP-1 cells were examined by fluorescence microscopy (Fig. 4). After colchicine (30-min) and nocodazole (2-h) treatment, an anti--tubulin antibody gave diffuse labeling. When the cells were treated with nocodazole for only 30 min, however, stable microtubules radiating from the centriole were detected. Thus, microtubule and microfilament polymerization appears to be involved in conventional phagocytic events whereas only stable microtubules and microfilament polymerization are involved in coiling phagocytosis.

View larger version (71K): [in this window] [in a new window]   FIG. 3.   Comparison of the ability of cytochalasin D (A), 30-min preincubation with nocodazole (B), colchicine (C), and 2-h preincubation with nocodazole (D) to inhibit the entry of BCYE-grown (BG) and amoeba-grown (AG) L. pneumophila into THP-1 cells. The ability of each of these pharmacological agents to inhibit entry is reported as the percentage of CFU that enter in the presence of the inhibitor compared to a control in the absence of the inhibitor. E. coli HB101 was used as a conventional phagocytosis control. Serum was replaced with RPMI in the no-serum controls (BG and AG). The abbreviation +H indicates that heat-inactivated serum was incubated with the bacteria prior to the assay, and +C indicates that complete human serum was used. % Inhibition = 100  [100 × (cell-associated CFU in the presence of the inhibitor/cell-associated CFU in the absence of the inhibitor)]. Error bars represent the standard deviations of triplicate wells for each experiment. The results shown are from a single representative experiment of at least three independent experiments.

View larger version (72K): [in this window] [in a new window]   FIG. 4.   Fluorescence microscopy showing the distribution of microtubules in THP-1 cells either untreated (A) or treated with nocodazole for 30 min (B), colchicine for 30 min (C), or nocodazole for 2 h (D) at 37°C.

Growth in amoebae affects intracellular replication. The ability of amoeba-grown and BCYE-grown L. pneumophila to survive and replicate in resting and activated THP-1 cells was examined during the first 8.5 h following entry (Fig. 5). BCYE-grown L. pneumophila had the ability to persist in both resting and activated monocytes. However, the presence of complement decreased intracellular survival in both cell types. Amoeba-grown L. pneumophila displayed intracellular replication during the first 8.5 h in resting and activated monocytes. The presence of complement decreased the intracellular survival of amoeba-grown L. pneumophila initially, and this was followed by slightly slower intracellular growth than in the absence of complement. Thus, growth of L. pneumophila in amoebae enhanced intracellular survival and replication in both resting and activated monocytes compared to growth in BCYE.

View larger version (22K): [in this window] [in a new window]   FIG. 5.   Viability of BCYE-grown (A and B) and amoeba-grown (C and D) L. pneumophila after entry into resting (A and C) and activated (B and D) macrophages in the presence of RPMI (no serum), heat-inactivated serum, or complete serum. The number of CFU present at the first time point (30 min after entry) was considered 100% survival. Error bars represent the standard deviations of triplicate wells for each experiment. The results shown are from a single representative experiment of at least three independent experiments.

Growth of L. pneumophila in amoebae affects virulence. Although it is clear that growth in amoebae affects replication in monocytes, this phenotype may not affect virulence in vivo. There are several animal models for L. pneumophila infection, including mice (7, 57), rats (47, 54), and guinea pigs (4, 15, 54). We used the mouse model of infection based on its cost-effectiveness and the availability of a number of different strains that vary in their susceptibility to L. pneumophila. A/J mice are highly susceptible to infection by BCYE-grown L. pneumophila (7). Since infections in a highly susceptible animal model may be very different from those in humans, we also compared the virulence of amoeba- and BCYE-grown bacteria in two other mouse strains, C57BL/6, which is thought to be resistant to infection with L. pneumophila (57), and C57BL/6 Beige, which is commonly used as a model for mycobacterial infection due to its higher susceptibility (5, 21). When each of these mouse strains was infected by the intratracheal route, there was an increased rate of replication in the lungs for amoeba-grown L. pneumophila compared to BCYE-grown L. pneumophila (Table 1). Inoculation of the unpurified bacterium-amoeba 2-day culture resulted in the same bacterial numbers in the lungs as did inoculation of the amoeba-grown bacteria alone. Addition of a lysate of the amoebae to BCYE-grown bacteria prior to inoculation did not affect bacterial replication in the mouse lungs compared to BCYE-grown bacteria alone. Although an increase in bacterial growth of amoeba-grown L. pneumophila was observed in the lungs of all three mouse strains, the infections appeared to be resolving after 4 days in the C57BL/6 and C57BL/6 Beige mouse strains. Based on these data, the C57BL/6 Beige mouse strain is intermediate between A/J and C57BL/6 in its susceptibility to L. pneumophila infection. Overall, these data indicate that amoeba-grown L. pneumophila is more infectious than BCYE-grown bacteria in both susceptible and resistant mice.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have found that growth of L. pneumophila in amoebae affects entry mechanisms into monocytes. Previous studies performed with BCYE-grown L. pneumophila demonstrated that the presence of complete serum enhances entry and that entry can be blocked by specific antibodies to the complement receptors (32, 40). Our experiments confirmed these studies but also indicated that complement does not affect the frequencies of entry by amoeba-grown L. pneumophila. This result may be explained by the fact that the outer membrane protein profile of bacteria grown under these conditions is very different (13), suggesting that different ligands and host cell receptors may be used. Alternatively, amoeba-grown L. pneumophila may bind to the same receptors but trigger uptake more efficiently. In either case, this is the first demonstration that amoeba-grown L. pneumophila enters monocytes by a complement-independent mechanism. This observation suggests that other growth conditions may also affect the frequencies of entry via the previously described complement-dependent mechanism.

The higher levels of entry and absence of complement involvement suggest that amoeba-grown L. pneumophila enters monocytes via a different mechanism from that of BCYE-grown L. pneumophila. Previous studies have demonstrated that both conventional and coiling phagocytic events occur in amoeba-grown (13) and BCYE-grown (13, 25) bacteria. Quantitation of the frequencies of these phagocytic events was necessary to determine the role of complement and the growth environment. Since the frequencies of coiling phagocytosis are significantly lower in the presence of complete serum, it is likely that complement opsonization enhances conventional phagocytic uptake of L. pneumophila by monocytes. Thus, the increase in the frequency of entry of BCYE-grown bacteria in the presence of complement may be due to an increase in the frequency of conventional phagocytic events. This conclusion is supported by previous observations that complement-opsonized bacterial strains (14, 25) and liposomes (3) enter via conventional phagocytosis. The role of complement receptors in coiling and conventional phagocytic events was not examined in our studies. In future studies it should be possible to determine the role of complement receptors in each of these entry mechanisms by using monoclonal antibodies to block the receptor binding sites followed by quantitation of the structural mechanisms of entry used.

The coiling phagocytic mechanism used by L. pneumophila is unusual and is only rarely seen in other bacteria, with the exception of spirochetes (12, 45, 46). A number of bacteria enter cells via mechanisms that are microfilament dependent (inhibited by cytochalasin D); these include Chlamydia (51), Salmonella (19), Shigella (6, 19), Streptococcus (31), Bordetella (22), Citrobacter (38), Escherichia (37), Ehrlichia (44), and Yersinia (19). Of these, the majority (Escherichia, Streptococcus, Ehrlichia, Bordetella, and Chlamydia) are also microtubule dependent (inhibited by colchicine or nocodazole). In our study, BCYE-grown L. pneumophila entered monocytes by a mechanism involving both microfilaments and microtubules, similar to previous observations for conventional phagocytic mechanisms (41). Our data suggest that microfilaments also play a role in uptake of amoeba-grown L. pneumophila but that this mechanism is not as sensitive to cytochalasin D as is entry by BCYE-grown bacteria. Inhibition of entry of amoeba-grown bacteria in the presence of colchicine and nocodazole indicates that microtubules are required. Interestingly, the 30-min preincubation with nocodazole did not inhibit the entry of amoeba-grown L. pneumophila. This is probably because colchicine causes a rapid breakdown of tubulin polymers to monomers at 37°C (53) whereas nocodazole has little effect on stable microtubules over short periods at this temperature (24). Although breakdown of preformed microtubules by nocodazole is inefficient at 37°C, extended treatment (2 h) at this temperature results in complete breakdown of the microtubular network (27). These data are consistent with the distribution of microtubules in THP-1 cells after treatment with nocodazole observed by fluorescence microscopy in our studies. Taken together, these results suggest that the de novo synthesis of microtubules is required for entry into monocytes by BCYE-grown L. pneumophila but that only stable microtubules are required for entry by amoeba-grown bacteria.

We have found that amoeba-grown L. pneumophila displays increased intracellular survival and replication in macrophages compared to BCYE-grown L. pneumophila. It is possible that this enhanced intracellular survival is related to the different entry mechanisms used by these bacteria, but it may also be due to unrelated factors that are influenced by growth in amoebae. We examined survival at very early time points after entry into monocytes with the assumption that if entry plays a role in intracellular survival, it is likely to affect events soon after entry. As shown in previously published studies, BCYE-grown bacteria will go on to replicate in monocytes even in the presence of complement (26, 32, 40). The conventional phagocytic mechanism is consistent with a complement-mediated mechanism of entry, since complement receptor-mediated uptake involves both microfilaments and microtubules (36). It has been suggested that entry of L. pneumophila into monocytes via the complement-mediated mechanism is advantageous since it prevents activation of the oxidative burst (56). Our data indicate, however, that the complement-mediated mechanism of uptake does not provide a selective advantage in monocytes. In addition, BCYE-grown bacteria were not as successful at replicating in vivo as were amoeba-grown bacteria in all three mouse strains tested. These observations suggest that the differences between the abilities of L. pneumophila grown in amoebae and of that grown in BCYE to replicate in monocytes are significant enough to affect virulence in vivo. Alternatively, other factors unrelated to replication in monocytes could be responsible for this phenotype. Due to the large number of differences between BCYE- and amoeba-grown L. pneumophila, a detailed analysis of the genes involved is necessary to differentiate between these possibilities.

The fact that amoeba-grown L. pneumophila enters monocytes via a somewhat altered mechanism and displays enhanced growth in the lungs of mice suggests that bacteria grown in protozoan hosts present in domestic water supplies are responsible for the production of Legionnaires' disease in humans. The differences in the ability of L. pneumophila to enter, survive, and replicate in host cells when grown under different conditions underscore the importance of taking bacterial growth conditions into account when carrying out pathogenesis studies. Further examination of the genes induced by growth in amoebae should lead to a better understanding of the bacterial factors required for the production of Legionnaires' disease.