1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Legionella pneumophila is the cause of Legionnaires' pneumonia. After internalization by macrophages, it bypasses the normal endocytic pathway and occupies a replicative phagosome bound by endoplasmic reticulum. Here, we show that lysis of macrophages and red blood cells by L. pneumophila was dependent on dotA and other loci known to be required for proper targeting of the phagosome and replication within the host cell. Cytotoxicity occurred rapidly during a high-multiplicity infection, required close association of the bacteria with the eukaryotic cell and was a form of necrotic cell death accompanied by osmotic lysis. The differential cytoprotective ability of high-molecular-weight polyethylene glycols suggested that osmotic lysis resulted from insertion of a pore less than 3 nm in diameter into the plasma membrane. Results concerning the uptake of membrane-impermeant fluorescent compounds of various sizes are consistent with the osmoprotection analysis. Therefore, kinetic and genetic evidence suggested that the apparent ability of L. pneumophila to insert a pore into eukaryotic membranes on initial contact may play a role in altering endocytic trafficking events within the host cell and in the establishment of a replicative vacuole.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Legionella pneumophila (Lp) causes a severe pneumonia known as Legionnaires' disease (Winn and Myerowitz, 1981; Fraser et al., 1977). After inhalation by a susceptible host, the organism grows intracellularly within resident or recruited macrophages. Amenable to in vitro genetic manipulation, it has proved to be a useful model for gaining insight into how organisms are able to grow within a phagocytic compartment and cause disease (reviewed in Isberg, 1994). Horwitz and co-workers have shown that it subverts the usual endocytic pathway, avoiding phagolysosome fusion and eventually establishing a replicative vacuole (Horwitz, 1983a,b; 1987; Clemens and Horwitz, 1992; 1995). The replicative vacuole is bound by endoplasmic reticulum (ER) and appears to be similar to autophagic vacuoles used to digest organelles during periods of amino acid starvation (Swanson and Isberg, 1995).

In an effort to define the virulence mechanisms of this organism, cytotoxic activities have been studied. Only those activities relevant to the present study will be reviewed here. Two non-cell-associated cytotoxins capable of causing zones of haemolysis on blood plates have been described. Keen and Hoffman (1989) and Szeto and Shuman (1990) showed that mutants unable to produce the Zn2+ metalloprotease, a major secreted protein of L. pneumophila, were unable to cause zones of haemolysis when streaked on canine or guinea pig erythrocyte plates respectively. Wintermeyer et al. (1991) described an activity named legiolysin that differs from the protease. However, neither secreted activity is necessary for growth within macrophages, although there is a report of a mild attenuation of virulence for a mutant lacking the protease in a guinea pig model (Moffat et al., 1994; Wintermeyer et al., 1994).

Husmann and Johnson (1994) reported that L. pneumophila serogroup 1 killed guinea pig macrophages and the J774 macrophage-like cell line within a few hours after a high-multiplicity infection. Bacterial strains isolated after passage on supplemented Mueller–Hinton agar, a procedure known to select for mutants unable to grow intracellularly, were incapable of killing. The activity described by Husmann and Johnson (1994) was potentiated by opsonization and appeared to be associated with the bacterial cell, because cytotoxicity was prevented by insertion of a mixed cellulose ester membrane between bacteria and target cells.

Muller et al. (1996) described apoptotic death of the HL60 macrophage-like cell line after a 24 h or longer infection with L. pneumophila, presumably near to the time that bacteria had completed a full round of replication within a host cell. As the time course of cell death was very different from that observed by Husmann and Johnson (1994), the mechanism of cell death described in these two studies may be different. In summary, L. pneumophila produces a cell-associated cytotoxin, one or more non-cell-associated toxins capable of causing β-haemolysis and an undefined activity that causes apoptosis later in infection.

Recently, genetic loci necessary or presumed necessary for the inhibition of phagolysosome fusion and targeting to an endoplasmic reticulum-bound compartment have been defined. These include dotA, which codes for an inner membrane protein (Berger and Isberg, 1993; Berger et al., 1994; Roy and Isberg, 1997; C. R. Roy and R. R. Isberg, unpublished data), icmWXYZ (Marra et al., 1992; Brand et al., 1994) and at least eight other loci identified by transposon mutagenesis (Sadosky et al., 1993), chemical mutagenesis (H. L. Andrews and R. R. Isberg, unpublished data), and selection on high concentrations of salt (Vogel et al., 1996; J. P. Vogel and R. R. Isberg, unpublished data).

In this report, we describe and analyse for the first time a rapid cytotoxicity activity that depends directly on known L. pneumophila genes required for virulence. We found that cytotoxicity occurred by osmotic lysis and was consistent with the insertion of a pore into the eukaroytic cell membrane.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Legionella pneumophila is cytotoxic for macrophages

Primary bone marrow macrophages were infected at high multiplicity with L. pneumophila strains (Lp02 or Lp01) competent for intracellular growth or with isogenic mutants that fail to target properly and grow within primary macrophages or U937 cells (Berger et al., 1994; Marra et al., 1992; J. P. Vogel, H. L. Andrews and R. R. Isberg, unpublished data). After a 1 h incubation, cell death was assessed after staining with the fluorescent dyes, ethidium bromide (EtdBr) and acridine orange (AO; McGahon et al., 1995). Cells with red nuclei were scored as dead in this assay (EtdBr permeable) and those with green nuclei as alive (EtdBr impermeable). Representative fields of stained macrophages infected with the intracellular replication-competent strains, Lp02 or Lp01, and mutant strains are shown in Fig. 1. Lp02- or Lp01-infected macrophages showed almost uniformly red nuclei (insets) in contrast to the almost uniformly green nuclei of macrophages infected with mutants that fail to target properly within macrophages (dotA, dotB, dotE, dotG, dotH, dotI and dotO ; dotI data not shown). Introduction of a plasmid containing a dotA+ gene, but not vector alone, into a dotA strain [Fig. 1; JV382 (pdotA+), JV383 (vector alone)] restored cytotoxic activity. As the dotA mutant, Lp03, has been the most extensively characterized of the group, it was chosen as the representative mutant for the experiments described below.


Figure 1. . Cytoxicity by L. pneumophila requires factors necessary for intracellular targeting and growth. Bone marrow-derived macrophages were infected at an MOI = 500:1 for 1 h at 37°C with strains competent for intracellular growth (Lp01 or Lp02) or dotA, dotB, dotE, dotG, dotH, dotI and dotN mutants, followed by staining with ethidium bromide/acridine orange (Experimental Procedures). JV382 is a dotA mutant complemented with a plasmid containing a dotA+ gene; JV383 is a dotA mutant containing vector alone (pKB5; Experimental procedures). Note that the nuclei of JV383 are green, not orange, and that the orange-tinged cytoplasm is from the staining of acidic compartments by acridine orange. Photomicrographs were taken with a FITC filter and 10 × objective. Insets show the corresponding isogenic Lp01 or Lp02 infections from the same experiment as the mutant in the larger panel.

Download figure to PowerPoint

Cell death was quantified using this microscopic assay (Fig. 2A) or, alternatively, by measuring lactate dehydrogenase (LDH) released owing to loss of cell membrane integrity (Fig. 2B). Over 90% of Lp02 (dotA+)-infected macrophages were EtdBr permeable after a 1 h incubation with replication-competent L. pneumophila, compared with 5% of the Lp03 (dotA)-infected macrophages or uninfected controls. Lp02-infected macrophages similarly released more than 65% of their LDH stores compared with less than 25% released by the dotA mutant or uninfected control cells. The lower fractional LDH release compared with fractional cell death estimated by the EtdBr-AO assay may reflect trapping of LDH within membrane-bound compartments or a more rapid influx of EtdBr (molecular weight 394) compared with the efflux of the much larger LDH molecule (135 kDa).


Figure 2. . Quantitative evaluation of cell death caused by L. pneumophila.A. Bone marrow-derived macrophages were infected at an MOI = 500:1 for 1 h at 37°C. Percentage of macrophages permeable to EtdBr was quantified after staining with ethidium bromide/acridine orange by counting video images of fluorescent cells observed with a rhodamine filter and dividing by the total number of cells (FITC filter). Each plotted value includes an average and standard deviation determined from four coverslips (average of 930 cells counted per coverslip). Control values are from parallel incubations without added bacteria. B. Percentage LDH release from the same infected macrophages was calculated by dividing the LDH released from samples by the LDH released from Triton X-100-permeabilized macrophages incubated under the same conditions (Experimental procedures). Data points represent the averages and standard deviations of six parallel assays.

Download figure to PowerPoint

In a time course experiment, more than 50% LDH release (t1/2; Fig. 3A) occurred at approximately 20 min. Ethidium bromide staining appeared to be a more sensitive indicator of early cell death than LDH release, although it was not practical to follow a rapid time course with this technique, as additional unfixed cells became permeable to EtdBr during microscopy. Detectable cell death was observed with a multiplicity of infection (MOI) as low as 5:1 with the EtdBr-AO assay (Fig. 3B; with 50% maximal haemolysis occurring with MOI = 40:1) and an MOI as low as 25:1 using the LDH assay (= 0.01; data not shown).


Figure 3. . Time course of cytotoxicity and dependence of cytotoxicity on multiplicity of infection. A. LDH determinations at indicated time points were determined for macrophages infected at an MOI of 500:1 with either Lp02 (dotA+) or Lp03 (dotA). The zero time point started at the beginning of the 37°C incubation period. Values and error bars represent the means and standard deviations of assays performed in triplicate. B. Percentage EtdBr-permeable cells at indicated multiplicities of infection were determined after staining with ethidium bromide and acridine orange after a 1 h infection (Experimental procedures). Plotted values indicate the means and standard deviations of percentage EtdBr-permeable cells determined from four coverslips, one representative field counted per coverslip except for the MOI of 5:1 and 1:1, where two fields per coverslip were counted (average of 500 cells counted per field).

Download figure to PowerPoint

Macrophages infected with replication-competent L. pneumophila were examined morphologically by phase contrast, fluorescence and transmission electron microscopy to determine whether cell death occurred by apoptosis or necrosis. Apoptotic morphology is characterized by nuclear condensation and fragmentation in the presence of an intact cell membrane at early stages of progression to death (Kerr et al., 1995; McGahon et al., 1995). Using the EtdBr-AO stain, apoptotic cells are therefore characterized by fragmented, green-staining nuclei, as EtdBr is excluded at early time points. As can be observed in Fig. 1, however, the nuclei of Lp02-infected macrophages were neither condensed nor fragmented and almost uniformly red. Furthermore, this time course was significantly more rapid than would be expected for apoptotic death in general and the apoptotic cell death observed 24 h after L. pneumophila infection by Muller et al. (1996). Survey of many microscopic fields revealed only very rare cells with apoptotic morphology (< 0.1%) with no obvious difference between infected macrophages and uninfected controls.

Unfixed Lp02-infected (dotA+) cells formed large membrane blebs that broke off leaving behind the nucleus and phase-dense granular material, consistent with osmotic lysis. In electron micrographs, the nuclear contour of Lp02-infected cells remained generally intact without sharply demarcated peripheral chromatin condensation or nuclear fragmentation characteristic of apoptosis (Fig. 4A; Kerr et al., 1995). Gross swelling and degeneration of cytoplasmic organelles and loss of cell membrane integrity could also be observed. In contrast, Lp03-infected (dotA) cells showed no degenerative changes, despite similar numbers of internalized bacteria (Fig. 4B). Therefore, cytotoxicity showed morphological features consistent with necrotic cell death accompanied by rapid osmotic lysis.


Figure 4. . Cytotoxicity caused by L. pneumophila shows features of necrotic cell death. Transmission electron microscopy images of macrophages infected with either Lp02 (A, dotA+) or Lp03 (B, dotA) at an MOI of 500:1 for 1 h at 37°C.

Download figure to PowerPoint

After a 28 h infection with Lp02 at low multiplicity (MOI of 1:1 or 1:10), we also observed macrophages with condensed fragmented nuclei consistent with apoptotic forms admixed with necrotic cells as assessed by EtdBr-AO staining (data not shown). Lp03 (dotA) caused minimal cytotoxicity at later time points. Therefore, cell death may follow a biphasic pattern with a rapid, necrotic cell death observed with high-multiplicity infection and, as reported previously (Muller et al., 1996), a later phase of apoptotic cell death following low-multiplicity infection.

Cytotoxicity is cell associated

Supernatants from monolayers of macrophages infected with Lp02 or Lp03 were either filtered through a low-protein binding filter or centrifuged to pellet bacteria and applied to fresh macrophages. At the end of a 1 h incubation at 37°C, cells were stained with ethidium bromide and acridine orange, and cell death was quantified. There was only a background level of cell death and no significant difference between values obtained using supernatants from Lp02- or Lp03-infected macrophages prepared in either manner (data not shown; four coverslips analysed per condition).

During a high-multiplicity infection, non-cytotoxic Lp03 associated as well as, or better than, the cytotoxic Lp02 strain (Fig. 5A; Experimental procedures). In the presence of opsonizing antibody, more than 100 bacteria were found associated with most macrophages for both Lp02 and Lp03 (data not shown; Experimental procedures). As the number of non-cytotoxic, opsonized Lp03 associated with macrophages is more than 10-fold greater than the number of cytotoxic, non-opsonized Lp02 (Fig. 5B), cytotoxic activity cannot be the result of an enhanced ability of replication-competent strains to form a stable association with macrophages.


Figure 5. . Effect of cytochalasin D and opsonizing antibody on cytotoxicity and bacterial uptake. A. Average number of bacteria associated with macrophages infected at an MOI of 500:1. Pre-Imm, bacteria incubated with preimmune serum; cytoD, infection in the presence of cytochalasin D. Infections were for 1 h at 37°C in the presence of 30 mM dextran 6000 to prevent lysis of macrophages (see below). Values and error bars are the means and standard deviations of the average number of cell-associated bacteria per macrophage counted from three coverslips; 100 macrophages examined per coverslip. B. LDH released from macrophages infected at an MOI of 500:1. Anti-Lp. opsonizing antibody present; Pre-Imm, bacteria incubated with premmune serum; cytoD, infection in the presence of cytochalasin D. Infections were for 1 h at 37°C. Similar data were observed in separate experiments using ethidium bromide and acridine orange to quantify cytotoxicity. C. Bacterial uptake by macrophages measured by gentamicin protection (Experimental procedures). Anti-Lp, opsonizing antibody present; Pre-Imm, bacteria incubated with preimmune serum; cytoD, infection in the presence of cytochalasin D. Infections were for 2 h at 37°C at an MOI of 1:1. For B and C, values and error bars represent means and standard deviations for experiments performed in quadruplicate.

Download figure to PowerPoint

To establish whether phagocytosis of bacteria was necessary for cytotoxicity, infections were carried out in the presence or in the absence of cytochalasin D in a manner similar to the experiments of Husmann and Johnson (1994). This drug had been shown previously by electron microscopy to inhibit the internalization of Legionella pneumophila (Elliott and Winn, 1986). In the presence of preimmune serum, cytochalasin D efficiently prevented Lp02 from becoming internalized within the eukaryotic cell (Fig. 5C; Isberg et al., 1987), while reducing cytotoxicity by only 50% (Fig. 5B). Therefore, for non-opsonized Lp02, substantial cytotoxicity still occurred in the absence of uptake, although phagocytosis appeared to be necessary for the highest level of cytotoxicity. In addition, similar levels of Lp02 and Lp03 appeared to be ingested by macrophages under all the conditions tested (Fig. 5C). Therefore, the ability of Lp02, but not Lp03, to cause cytotoxicity does not result from a more invasive phenotype.

Husmann and Johnson (1994) described a cell-associated cytotoxic activity from L. pneumophila that was enhanced by opsonizing antibody. Although we found opsonization allowed full cytotoxicity in the presence of cytochalasin D (Fig. 5B), opsonized bacteria became internalized as efficiently in the presence of this drug as non-opsonized bacteria in its absence (Fig. 5C). We, therefore, cannot rule out the necessity of phagocytosis for full cytotoxicity, even in the presence of tight binding between bacteria and macrophages promoted by IgG-coated bacteria, because all conditions that allowed maximum cytotoxicity also permitted a detectable level of phagocytosis.

In contrast to cytochalasin D, a variety of reagents that neutralize intracellular acid compartments had no effect on cytotoxicity (data not shown). Cytotoxicity was unaffected if infections were carried out in the presence of inhibitors of the H+ vacuolar ATPase (250 nM bafilomycin A1 or 100 nM concanamycin A) or an H+ ion-specific ionophore (20 μM monensin) added in addition to 100 nM concanamycin A (Papini et al., 1993; Menard et al., 1996). This is despite the fact that there was a complete absence of acidification of intracellular compartments based on acridine orange staining.

Contact-mediated haemolysis

To test whether the cytotoxic activity was specific for macrophages, the ability to lyse sheep red blood cells (RBCs) was assessed. Haemolysis was observed only when L. pneumophila competent for intracellular replication (Lp02 or Lp01) was pelleted with red blood cells, but was not seen with any of the replication-defective isogenic mutants [dotA, dotB, dotE, dotG, dotH and dotI (Fig. 6; dotO data not shown]. In addition, haemolysis did not occur in the absence of centrifugation (Fig. 6A; Lp02 NS, Lp01 NS) or after an incubation at 0°C (data not shown). Contact-dependent haemolysis also occurred efficiently with rabbit and horse blood (data not shown). Introduction of a plasmid containing a dotA+ gene but not vector alone into a dotA mutant strain (Fig. 6B; Lp03 pdotA+, Lp03 vector) resulted in restoration of the haemolytic activity to roughly 33% of wild-type levels. The more complete complementation observed in the macrophage cytotoxicity assay (see Fig. 1) may reflect the higher MOI used in the macrophage infections.


Figure 6. . Haemolysis by L. pneumophila requires factors necessary for intracellular targeting and growth. A. Sheep red blood cells (0.375%) were pelleted or mixed but not pelleted (NS) with replication-competent L. pneumophila (Lp01 or Lp02) or strains defective for targeting and intracellular replication at an MOI of 25:1 and incubated for 2 h at 37°C. A415 indicates spectrophotometric determination of haemoglobin released into supernatant. B. As above, sheep red blood cells were pelleted with replication-competent strains, Lp02 or Lp03 pdotA+ [JV382; Lp03 dotA/pKB12 (dotA+)] or strains defective for targeting and intracellular replication, Lp03 or Lp03 vector (JV383; Lp03/pKB5), and incubated for 1 h at 37°C. Values and error bars represent means and standard deviations for assays performed in quadruplicate.

Download figure to PowerPoint

The kinetics of haemolysis by Lp02 appeared slightly slower than macrophage lysis, with a t1/2 of approximately 30 min (Fig. 7A). A significantly increased level of haemolysis in comparison with Lp03 (dotA) occurred with an MOI as low as 0.39:1 (= 0.02; Fig. 7B), with 50% maximal haemolyis occurring at an MOI = 5:1. After a 10 min incubation, approximately 50% of red cells pelleted with Lp02, but not with Lp03, were observed to be lysed after fixation and examination by electron microscopy (data not shown). There was also no observable phagocytosis of bacteria by red cells (data not shown). Therefore, haemolysis appeared to be similar to macrophage cytotoxicity in terms of kinetics, contact dependence and dependence on factors involved in intracellular growth, suggesting that the same activity was promoting both.


Figure 7. . Quantitative evaluation of haemolysis by L. pneumophila.A. Sheep red blood cells (0.25%) were pelleted with either Lp02 or Lp03 at an MOI of 25:1 (Experimental procedures) and incubated at 37°C for the times indicated. B. Sheep red blood cells (0.25%) were pelleted with either Lp02 or Lp03 at the indicated MOI and incubated for 1 h at 37°C. Values and error bars represent means and standard deviations for experiments performed in quadruplicate.

Download figure to PowerPoint

Evidence for a pore-forming toxin

Osmotic lysis of macrophages and red blood cells could result from a number of different insults. For instance, the Na+/K+ ATPase pump of eukaryotic cells creates an electrochemical gradient that counterbalances the excess osmotic pressure inside the cell (Guyton, 1986). Direct or indirect inhibition of the pump will lead to osmotic lysis, but only after several hours, inconsistent with the rapid kinetics of L. pneumophila-mediated cell lysis (Tosteson and Hoffman, 1960). In contrast, damage to the plasma membrane through the insertion of a pore-forming toxin or the action of a detergent should lead to the rapid lysis observed (t1/2 < 20 min for macrophages).

The ability of different-sized compounds added to the external medium to protect cells from osmotic lysis allows differentiation of pore formation from other forms of membrane damage (Lobo and Welch, 1994). High concentrations of an osmotically active compound added to the external medium should keep the external medium hypertonic compared with the interior of the cell. In the presence of a pore-forming toxin, however, compounds will equilibrate across the cell membrane. Compounds that approach or are larger than the size of the pore will serve as osmoprotectants, because they diffuse slowly through the pore. Therefore, the differential ability of high-molecular-weight sugars, dextrans and glycols to suppress osmotic lysis indicates the presence of a pore-forming toxin and allows the approximation of pore size (Bhakdi et al., 1986; Clinkenbeard et al., 1989; Lalonde et al., 1989; Iwase et al., 1990; Noronha et al., 1996; reviewed in Lobo and Welch, 1994).

Consistent with L. pneumophila cytotoxicity resulting from pore insertion, 30 mM PEG3350 (Experimental procedures; average molecular weight = 3350) was able to prevent obvious blebbing of macrophages infected with Lp02(dotA+), whereas PEG1000 (average molecular weight = 1000) was not (data not shown). It should be noted that PEG3350 does not prevent cell death. Lp02-infected cells treated with PEG3350 became permeable to ethidium bromide, indicating cytotoxicity in the absence of osmotic lysis (data not shown; see Fig. 9). Without gross membrane damage in PEG-protected cells, ethidium bromide might enter the cell through inserted pores in a manner similar to its ability to pass through some types of gap junctions (Elfgang et al., 1995) or toxin channels (Finck-Barbancon et al., 1993).


Figure 9. . Low-molecular-weight dyes stain osmoprotected macrophages infected by L. pneumophila. A. Macrophages infected with Lp02 or Lp03 in the presence of opsonizing antibody and 30 mM dextran 6000 added as an osmoprotectant were stained with lucifer yellow CH (LY), TOTO-1 and Texas red-X phalloidin (PHALL). Molecular weights (MW) and charge of dye are indicated. Representative fluorescent and corresponding Hoffman modulation contrast images are shown. B. Macrophages treated with Lp02 or with both Lp02 and streptolysin O (+ STREP O) in the presence of opsonizing antibody and 30 mM dextran 6000 were stained with Texas red dextran 3000.

Download figure to PowerPoint

To estimate the size of a potential pore inserted by L. pneumophila, the ability of PEGs of various sizes and osmolarities to protect RBCs from contact-mediated haemolysis was determined, assessing the osmolarity necessary for 50% protection by each. 9Figure 9A shows the osmolarity necessary for 50% protection for each size of PEG plotted as a function of its Einstein–Stokes molecular diffusion radius (Scherrer and Gerhardt, 1971). As a control, a similar analysis was performed for α-haemolysin from Staphylococcus aureus, a pore-forming toxin with a solved crystal structure (Song et al., 1996). An asymptotic line, defining a threshold below which protection does not occur at infinite osmolarity, approximates a pore radius of roughly 1.5 nm for both α-haemolysin and L. pneumophila (Fig. 8A).


Figure 8. . Assessment of osmoprotection of red blood cells from lysis by Lp02, Staphylococcusα-haemolysin and streptolysin O. A. Rabbit red blood cells (0.375%) were either pelleted with Lp02 at an MOI of 25:1 or mixed with 15 units of α-haemolysin from Staphylococcus aureus in the presence of polyethylene glycols of different molecular weights (PEG600, 1000, 2000, 3350, 8000) in twofold concentration increments ranging from 1.56% to 25% by weight. For each PEG, the osmolarity necessary for 50% suppression of haemolysis was determined graphically. Osmolarity for 50% protection is plotted as a function of the Einstein–Stokes molecular diffusion radius for each PEG (Scherrer and Gerhardt, 1971). Data were derived from the means of triplicate assays. The size of each PEG derivative is denoted by the symbols in the legend to 8Fig. 8B. B. Rabbit red blood cells (0.375%) were mixed with 30 units of streptolysin O and incubated for 1 h at 37°C. The osmolarity of each PEG solution is plotted against released haemoglobin as assessed by A415. Data were derived from the means of triplicate assays. C. Sheep red blood cells (0.375%) were pelleted with Lp02 or Lp03 at an MOI of 25:1 in the presence of either 30 mM dextran 6000 or 30 mM sucrose, incubated for 1 h at 37°C and processed for the determination of haemolysis. The supernatants from parallel quadruplicate assays incubated for 1 h with 30 mM dextran 6000 were withdrawn and replaced with either 30 mM sucrose (new sucrose) or 30 mM dextran 6000 (new dextran). Samples were incubated for an additional 5 min at 37°C, and haemolysis was determined. Data show means and standard deviations of assays performed in quadruplicate.

Download figure to PowerPoint

In contrast to the above results indicating a pore of defined size, haemolysis caused by streptolysin O was poorly suppressed by any of the PEG solutions (Fig. 8B), consistent with its known ability to form a pore larger than the haemoglobin molecule (Buckingham and Duncan, 1983). Except for the one or two highest osmolarities tested for each PEG solution, haemolysis appeared to increase with increasing PEG osmolarities, presumably by forcing haemoglobin out of the cell (Fig. 8B). Similarly, haemolysis caused by digitonin, a membrane detergent known to form larger membrane lesions without a discrete size threshold, was not suppressed by any of the PEG solutions (data not shown; Thelestam and Florin, 1994).

Control experiments were performed to rule out the possibility that high-osmolarity solutions prevented the activity of the L. pneumophila cytotoxin. Red blood cells were infected with Lp02 (dotA+) or Lp03 (dotA) for 1 h in the presence or absence of 30 mM dextran 6000 or 30 mM sucrose (Fig. 8C). Lp02 caused significant haemolysis in the presence of sucrose (molecular weight 342; Fig. 8C). In contrast, essentially no haemolysis occurred in the presence of dextran 6000, indicating that protection is not specific to polyethylene glycols. After a 1 h incubation, the supernatant from Lp02- or Lp03-infected RBCs incubated in the presence of 30 mM dextran was replaced by 30 mM sucrose or 30 mM dextran 6000. Lp02-infected samples lysed almost immediately after the addition of sucrose, indicating that membrane damage had occurred during incubation with dextran 6000. Similarly, membrane damage occurred in the presence of osmoprotection by PEG3350, detected as haemolyis after resuspension in sucrose (data not shown).

To characterize the pore-forming activity further, the ability of various-sized membrane-impermeant dyes to gain access to the cytoplasm of osmotically protected macrophages infected with Legionella was assessed (Experimental procedures). As was true for EtdBr, TOTO-1 (molecular weight 795), a polycationic DNA-binding dye, stained all Lp02-infected macrophages, but only rare Lp03-infected macrophages (Fig. 9A). Similar results were seen with lucifer yellow CH (molecular weight 443), an anionic dye, and Texas red-X phalloidin (approximate molecular weight 1490), a dye that binds F-actin. In contrast, Texas red-labelled dextran 3000 did not stain Lp02- or Lp03-infected macrophages efficiently. However, macrophages permeabilized with streptolysin O, a pore-forming toxin that should allow free passage of dextran 3000, stained well. Therefore, L. pneumophila appears to allow passage of dyes of molecular weight less than 1500 into the mammalian cytoplasm. A higher molecular weight dye is excluded, consistent with the pore size suggested by osmoprotection experiments.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

We have described an activity from L. pneumophila that causes rapid lysis of macrophages and red cells in a process that requires the function of a number of factors known to promote intracellular growth. The inability of supernatants from infected macrophages to cause lysis, the dependence on pelleting bacteria and RBCs together for haemolysis and the role of phagocytosis in contributing to full cytotoxic activity indicate that the cytotoxin is cell associated. Osmoprotection and dye exclusion experiments suggested that lysis was caused by the insertion of a pore of a defined size into the plasma membrane.

The cytotoxic activity appears to be distinct from the diffusable Zn2+-dependent metalloprotease and legiolysin, which are not required for intracellular growth (Szeto and Shuman, 1990; Moffat et al., 1994; Wintermeyer et al., 1994). The observed necrotic cell death is also distinct from the apoptotic death observed by Muller et al. (1996) after a low-multiplicity infection 24 h or more after infection. However, it is probably the same as the activity described by Husmann and Johnson (1994), based on a similar association with virulence and dependence on cell contact. Husmann and Johnson (1994) did not observe the protective effect of cytochalasin D as we did (Fig. 5), perhaps because of the lower cytotoxicity in their assays. Previously, Caparon and Johnson (1988) described a cytochalasin D-inhibited cytotoxic activity that was contact mediated and expressed only by virulent bacteria. The inconsistent inhibition by cytochalasin D probably reflects experimental variables and not a difference in the identity of otherwise similar activities.

For osmotic lysis to occur with a t1/2 of less than 20 min after infection, the pore was probably inserted just after association of bacteria with the macrophage. Presumably this occurred before phagocytosis or before closure of the phagocytic vacuole, when the damaged membrane was still in contact with the relatively hypotonic external medium. Alternatively, the pores formed in mature vacuoles could be recycled to the plasma membrane and cause lysis at a later time.

In support of a model of insertion of a pore upon initial interaction of bacteria and macrophage, Legionella is presumably able to insert a pore into eukaryotic membranes in the absence of phagocytosis. We have shown that L. pneumophila can effect rapid osmotic lysis of red blood cells, which are not phagocytic for L. pneumophila. In addition, experiments with cytochalasin D and non-opsonized bacteria show that significant lysis of macrophages occurred with minimal amounts of bacterial uptake. The process of enclosing L. pneumophila in a nascent phagosome may help to establish the close contact necessary for pore insertion.

The DotA-dependent mechanism altering endocytic trafficking events must act during or soon after phagocytosis, as wild-type L. pneumophila is found in a compartment phenotypically distinct from a dotA mutant 5 min after uptake (C. Roy, personal communication). The probable insertion of a pore on initial interaction of bacteria with macrophages is kinetically compatible with a role in causing early changes in trafficking that must take place to allow a replicative phagosome to form. In this model, the inability of the dotA, dotB, dotE, dotG, dotH, dotI and dotO mutants to allow pore insertion, most probably by eliminating the components of a transport machinery, is a cause of their defect in targeting. A local ion flux from the pore itself may influence targeting of the phagosome. Alternatively, the pore may have an additional activity that becomes activated or gains access to the cytoplasm after insertion. The pore may also act as a conduit for other bacterial proteins to enter the mammalian cytoplasm and alter the endocytic machinery. In this respect, it would be functionally analogous to YopB/YopD from Yersinia pseudotuberculosis, IpaB/IpaC from Shigella flexnerii and SipB/SipC from Salmonella typhimurium (High et al., 1992; Chen et al., 1996; Hakansson et al., 1996). These proteins have been implicated by homology and/or function as forming a pore in the mammalian cell membrane. Analogous to the L. pneumophila cytotoxin, the YopB/YopD and IpaB/IpaC proteins have been shown to be necessary for contact-dependent haemolysis by Yersinia and Shigella respectively. Based on osmoprotection experiments, the YopB protein was also predicted to form a pore of roughly the same size as the L. pneumophila cytotoxin (Hakansson et al., 1996). It is also possible that L. pneumophila opens a pre-existing channel in the eukaryotic membrane.

Assuming that the pore plays a necessary role in allowing intracellular growth, the insertion of only a small number of pores may be needed to establish a replication vacuole. In a high-multiplicity infection, however, the insertion of a large number of pores results in rapid host cell death, clearly not an efficient strategy for establishing a productive infection. During infection of amoebae, a natural reservoir for Legionella, or initial infection of humans by inhalation of aerosols or aspiration, a high-multiplicity infection is unlikely to occur, as, in the former case, bacteria should be readily dispersed in the environment and, in the latter, they should be distributed evenly within segments of lung. In Legionnaires' pneumonia, however, high local concentrations of organisms must occur after successive rounds of replication and lysis of infected macrophages. The ability of the toxin to disrupt membranes from red cells of three different species and macrophages from mice argues that the activity uses a shared receptor or physical characteristics. Therefore, we predict that high local concentrations of bacteria would be toxic to alveolar and bronchiolar epithelial cells in human disease in addition to macrophages. Rapid necrotic lysis of macrophages in vivo might also increase the inflammation and release of destructive enzymes contributing to local tissue destruction. Perhaps this type of cytotoxicity also leads to the leukocytoclastic process characteristic of this pneumonia (Winn and Myerowitz, 1981).

The estimated pore diameter of 3 nm probably overestimates the actual pore size because of the polydispersed nature of polyethylene glycols (e.g. PEG2000 contains a range of PEGs of average molecular weight 2000) and their lipid solubility (Scherrer and Gerhardt, 1971). The Staphylococcus aureusα-haemolysin pore size is 1.4 nm in narrowest diameter by crystal structure and 1.1 nm in diameter by electrophysiological measurements (Menestrina, 1986; Song et al., 1996), showing that the assay used here overestimated the pore size for this toxin by about a factor of two. By analogy, we predict that L. pneumophila produces a pore size closer to the actual size of α-haemolysin, large enough for the non-specific passage of ions, small sugars and peptides. This property may allow access to nutrients in the cytoplasm as has been proposed for the vacuolar pore from Toxoplasma gondii (Schwab et al., 1994).

Intriguingly, both Mycobacterium tuberculosis and Mycobacterium haemophilum, pathogens that replicate within phagocytic vacuoles, demonstrate contact-dependent haemolysis (King et al., 1993; Fischer et al., 1996), indicating a membrane-disrupting or pore-forming capability. It is, therefore, possible that pore formation is a widely used strategy, employed by pathogens that replicate within phagosomes to alter the nature of their replicative vacuole. Determining the precise role of the Legionella pneumophila cytotoxin in modifying the physiology of the host cell will be the subject of future investigations.

Experimental procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Strains, media and chemicals

All L. pneumophila strains are derived from Philadelphia serogroup 1 (CDC). Lp01 (rpsL hsdR ), Lp02 (rpsL hsdR thyA ) and Lp03 (Lp02 dotA ) have been described previously (Berger and Isberg, 1993). The derivation and characterization of JV303 (dotB ), JV328 (dotE ), JV573 (dotG ; J. P. Vogel and R. R. Isberg, unpublished), HL1700 (dotH ), HL056 (dotI ) and HL1400 (dotO ; H. L. Andrews and R. R. Isberg, unpublished) are to be described elsewhere. JV382 and JV383 are Lp03 (dotA ) strains containing pKB12 (a plasmid expressing a dotA+ gene, identical to pKB6 described in Berger et al., 1994) or pKB5 (parent vector; Berger and Isberg, 1993). Bacteria were passaged on buffered charcoal yeast extract supplemented with 100 μg ml−1 thymidine (BCYET) as described previously (Berger and Isberg, 1993). Unless otherwise indicated, all chemical reagents were from Sigma Chemical.

Preparation and infection of macrophages

Bone marrow-derived macrophages were isolated from female A/J mice that were 6–10 weeks old (Jackson Labs) and cultured in conditioned medium as described previously (Swanson and Isberg, 1995). Macrophages were replated in Falcon 24-well tissue culture plates (Becton Dickinson) at a density of 1.5 × 105 macrophages per well in RPMI medium (Irvine Scientific) supplemented with 10% fetal bovine serum (FBS; Gibco BRL). For microscopy experiments, macrophages were replated on coverslips (12 × 1 mm; Fischer Scientific). Before infection of macrophages, bacteria were patched onto BCYET plates and incubated for 2 days at 37°C. Just before infection, bacteria were resuspended in RPMI with or without FBS and phenol red and diluted to an appropriate MOI based on optical density (1 × 109 bacteria ml−1 for A600 = 1 assumed). For MOI plots, MOI values were calculated by titring colony-forming units from bacterial suspensions used to prepare the infectious inoculum. For infections performed in the absence of FBS and phenol red, macrophages were washed three times with fresh RPMI-1640 medium (Gibco BRL) without phenol red and FBS before the addition of the infectious inoculum prepared in the same medium. In the MOI experiment displayed in Fig. 3, medium contained 4% FBS. Otherwise, the bacterial inoculum was added directly to the macrophages without washing the macrophages. The presence or absence of FBS in the medium during the infection caused no apparent difference in cytotoxicity; however, FBS and phenol red resulted in a high background in the lactate dehydrogenase (LDH) assays described below and were deleted when appropriate. After the addition of bacteria, microtitre dishes were centrifuged at 100 g for 5 min at room temperature and placed at 37°C, 5% CO2 for 1 h or floated in a 37°C water bath for rapid equilibration necessary for time course experiments.

For cytotoxicity experiments in the presence of cytochalasin D, macrophages were preincubated for 2 h in fresh RPMI containing 3.9 μM cytochalasin D [from 1 mg ml−1 stock in dimethyl sulphoxide (DMSO)] or 0.2% DMSO. Uniform cell rounding, indicative of cytochalasin D treatment, was confirmed under phase-contrast microscopy. Bacterial infections were performed in fresh media containing the above concentrations of cytochalasin D or DMSO. Polyclonal rabbit anti-L. pneumophila antibody or preimmune serum (K. H. Berger and R. R. Isberg, unpublished data) was added to wells immediately before centrifugation for experimental convenience at a 1:2000 dilution. This dilution of antibody caused mild clumping of bacteria by the end of the incubation period. Preopsonization with antibody for 30 min before infection resulted in nearly identical (no statistical difference) cytotoxicity data, indicating that sufficient opsonization had occurred when antibody was added at the beginning of infection (data not shown).

Fluorescence microscopy

Ethidium bromide is excluded from cells with an intact cell membrane, whereas acridine orange stains all cells. Loss of membrane integrity is therefore indicated by orange–red nuclear staining, visible with either a fluorescein isothiocyanate (FITC) or rhodamine bandpass filter. Acridine orange stains the cytoplasm and nucleus of both living and dead cells green and acidic compartments of live cells orange. At the end of the incubation period, coverslips containing infected macrophages were inverted onto a 5 μl drop of phosphate-buffered saline (PBS) containing 25 μg ml−1 ethidium bromide and 5 μg ml−1 acridine orange placed on the surface of a glass slide. Coverslips were observed immediately using FITC and rhodamine filters and phase-contrast optics with a Zeiss Axioskop. To quantify EtdBr permeability, a Hamamatsu model C2400 video camera was used to record images with a 10 × objective using both the FITC and the rhodamine filter sets. The number of cells visible on saved images was quantified using IP Lab Spectrum Software (Signal Analytics). Two fields were counted from each coverslip (> 800 cells in total) after optimization of segmentation parameters.

Lactate dehydrogenase (LDH) release

Aliquots of media (50 μl) from wells containing macrophages incubated with bacteria were removed for the determination of LDH release. LDH determinations were made using the CytoTox 96 cytotoxicity assay kit (Promega), according to the manufacturer's instructions. After allowing 20–30 min of incubation with substrate, the A490 was determined using a Bio-Rad model 550 microplate reader. As a control for total cell-associated LDH, macrophages in selected wells were lysed with 0.9% Triton X-100. Fractional LDH release was calculated by dividing the A490 released from samples by total cell-associated LDH release. For other experiments, the raw A490 reading for experiments has been indicated.

Cytotoxicity of supernatants from infected macrophages

To test the ability of supernatants from macrophages to cause cytotoxicity in the absence of bacteria, 12 wells from 24-well microtitre dishes containing macrophages were incubated with either Lp02 or Lp03 for 1 h at 37°C and monitored for cytotoxicity by microscopy and LDH assay. Significant cell death of macrophages (Lp02 vs. Lp03) was confirmed by LDH assay (< 0.0001, n = 12). The supernatants were then collected after either filtration through a Millex GV 0.22 μm filter (Millipore) or centrifugation at 5500 g for 20 min at 4°C. Processed supernatants were then applied to fresh macrophages. At the end of a 1 h incubation at 37°C, cells were stained with ethidium bromide and acridine orange and cell death quantified as above.

Electron microscopy

Bone marrow-derived macrophages were infected for 1 h at an MOI of 500:1 and analysed as described previously (Swanson and Isberg, 1995). For contact-dependent haemolysis assays, red blood cell pellets were fixed in 2.5% glutaraldehyde in Sorensen's phosphate buffer (Dykstra, 1993), post-fixed with osmium tetroxide and then processed as above.

Gentamicin protection assay

Macrophages were incubated with bacteria as described above, except that macrophages were incubated with 3.9 μM cytochalasin D (1 mg ml−1 stock in DMSO) or DMSO in appropriate wells for 1 h in RPMI media with 10% FBS before infection with bacteria. Infections were performed at an MOI of 1:1 to avoid cytotoxicity from higher concentrations of bacteria. Throughout all subsequent steps, including gentamicin incubation, the identical concentrations of cytochalasin D or DMSO were present. As indicated above, anti-L. pneumophila antibody or preimmune serum was added to appropriate wells before centrifugation. After a 2 h incubation at 37°C, macrophages were washed once with media and incubated with 50 μg ml−1 gentamicin for 45 min. Macrophages were then washed three times with fresh media and lysed with distilled water. Colony-forming units were determined by serial dilutions on BCYET plates. Cytochalasin D had no obvious effect on bacterial viability. Percentage protection was determined by dividing the number of bacteria surviving the assay by the number of bacteria in the infectious inoculum.

Cell-associated bacteria assay

Macrophages were infected at an MOI of 500:1 with Lp02 or Lp03 in RMPI-1640 medium containing 30 mM dextran 6000 as an osmoprotectant in the presence or absence of cytochalasin D in the presence of preimmune serum or opsonizing antibody, as described above. At the end of a 1 h incubation at 37°C, macrophages were washed three times with PBS 30 mM dextran 6000 with or without cytochalasin D, fixed for 40 min in PBS formalin (Fischer Scientific) containing 30 mM dextran 6000, washed once with PBS with 50 mM ammonium chloride for 5 min, rinsed twice with PBS, permeabilized with methanol, blocked with PBS containing 2% goat serum (Gibco BRL) and then incubated with rabbit anti-L. pneumophila antibody at 1:1000, followed by FITC–goat anti-rabbit secondary antibody (Zymed Laboratories) at 1:1000. Cell-associated bacteria were counted using a Zeiss Axioskop microscope. The number of cell-associated bacteria in the presence of opsonizing antibody were too numerous to count accurately.

Haemolysis assays

Sheep, rabbit or horse red blood cells (Remel) were diluted in PBS, washed two or three times by centrifugation for 5 min at 17 000 g until the supernatant was essentially colourless, and the cells were counted with a haemocytometer. Reactions were set up in a final volume of 800 μl with the final concentration of RBCs as indicated in the figure legends and an MOI based on the red blood cell count. MOI ratios were generally approximated from the A600 of bacteria resuspended in PBS before incubation with RBCs as indicated above. For the data presented in 7Fig. 7B, MOI values were calculated by titring colony-forming units from bacterial suspensions used to prepare the infectious inoculum. Bacteria and RBCs were mixed, pelleted for 2 min at 17 000 g in a microfuge and incubated for 1 or 2 h at 37°C. At the end of the incubation period, the pellets were resuspended by vortexing and repelleted by centrifugation for 2 min at 17 000 g. Samples (100 μl) of supernatants were transferred to microtitre plates, and the A415 was recorded in a Bio-Rad microplate reader to assess haemoglobin release.

Osmoprotection assays

Macrophages incubated in RPMI medium supplemented with 30 mM PEG1000 or PEG3350 were infected at an MOI of 500:1 for 1 h at 37°C and stained with ethidium bromide and acridine orange as described above. Haemolysis assays were performed, as described above, except that bacteria and red cells were suspended in PBS containing the indicated osmolarities of PEG600, PEG1000, PEG3350, PEG8000 (Sigma) and PEG2000 (Aldrich Chemical). Alternatively, assays were performed in the presence of 30 mM dextran 6000 (Fluka) or 30 mM sucrose. Where indicated, supernatants from dextran 6000 assays were aspirated after a 1 h incubation and replaced with dextran 6000 or sucrose. Pellets were resuspended and haemolysis was determined after a further 5 min incubation at 37°C. For assays using purified Staphylococcus aureusα-haemolysin (Sigma) and streptolysin O (Sigma), red cells were first mixed with twofold serial dilutions of toxin and incubated for 1 h. The toxin concentration used in subsequent osmoprotection assays was the lowest amount necessary for complete or nearly complete haemolysis. Units of activity indicated for these toxins are from the manufacturer's determinations for its products. Streptolysin O was incubated with 10 mM dithiothreitol (Gibco BRL) for 5 min at room temperature before use in assays (Bhakdi et al., 1984).

Dye exclusion experiments

Dextran 6000 (30 mM) was present in all media and buffers used. Macrophages were infected with either Lp02 or Lp03 at an MOI of 500:1 in RPMI-1640 medium supplemented with anti-L. pneumophila antibody at 1:2000 as described above and incubated for 2 h at 37°C. For the last half hour of the incubation period, 333 units of dithiothreitol-activated streptolysin O was added to some Lp02-infected wells in PBS dextran 6000 buffer. At the end of the incubation period, macrophages were washed once with ice-cold PBS dextran 6000, incubated at 0°C for 10 min and then incubated with fluorescent dyes in RPMI-1640–dextran 6000 for 1 h on ice. At the end of the incubation period, macrophages were washed seven times with ice-cold PBS dextran 6000, and unfixed cells were viewed with a Nikon Eclipse TE300 inverted microscope. Photomicrographs were taken using Hoffman modulation contrast optics or fluorescent illumination with either Kodak Tmax 400 or Elite II film and scanned using a Polaroid SprintScan35 Plus slide scanner. Contrast adjustments for equal-length exposures were performed when necessary in parallel in Adobe Photoshop 3.0 (Adobe Systems) for each dye. Dyes were purchased from Molecular Probes and used at the following concentrations: lucifer yellow CH, lithium salt (625 μg ml−1); TOTO-1 iodide (500 nM); and Texas red-X phalloidin (1 U per coverslip). Texas red dextran 3000 was fractionated on a Sephadex G-15 (Pharmacia Biotech) column, and the fastest migrating fraction was used for staining following adsorption with SM-2 Bio-Beads (Bio-Rad) to remove any residual unincorporated dye (Schwab et al., 1994; Spack et al., 1986).


P-values were determined with STATVIEW 4.51 (Abacus Concepts) using the non-parametric Mann–Whitney U-test.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

We thank Jeannie T. Lee, Craig Roy and Jonathan Solomon for critical reading of the manuscript and Amiee Brown-Cormier for instruction and technical assistance with electron microscopy. This work was supported by the Howard Hughes Medical Institute (R.I.) and by K08 AI01402-01 (J.E.K.).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  • 1
    Berger, K.H. & Isberg, R.R. (1993) Two distinct defects in intracellular growth complemented by a single genetic locus in Legionella pneumophila. Mol Microbiol 7: 719.
  • 2
    Berger, K.H., Merriam, J.J., Isberg, R.R. (1994) Altered intracellular targeting properties associated with mutations in the Legionella pneumophila dotA gene. Mol Microbiol 14: 809822.
  • 3
    Bhakdi, S., Roth, M., Sziegoleit, A., Tranum-Jenson, J. (1984) Isolation and identification of two hemolytic forms of streptolysin-O. Infect Immun 46: 394400.
  • 4
    Bhakdi, S., Mackman, N., Nicaud, J., Holland, I.B. (1986) Escherichia coli hemolysin may damage target cell membranes by generating transmembrane pores. Infect Immun 52: 6369.
  • 5
    Brand, B.C., Sadosky, A.B., Shuman, H.A. (1994) The Legionella pneumophila icm locus: a set of genes required for intracellular multiplication in human macrophages. Mol Microbiol 14: 797808.
  • 6
    Buckingham, L. & Duncan, J.L. (1983) Approximate dimensions of membrane lesions produced by Streptolysin S and Streptolysin O. Biochim Biophys Acta 729: 115122.
  • 7
    Caparon, M. & Johnson, W. (1988) Macrophage toxicity and complement sensitivity of virulent and avirulent strains of Legionella pneumophila. Rev Infect Dis 10 (2): S377S381.
  • 8
    Chen, L.M., Kaniga, K., Galan, J.E. (1996) Salmonella spp. are cytotoxic for cultured macrophages. Mol Microbiol 21: 110115.
  • 9
    Clemens, D.L. & Horwitz, M.A. (1992) Membrane sorting during phagocytosis: selective exclusion of major histocompatibility complex molecules but not complement receptor CR3 during conventional and coiling phagocytosis. J Exp Med 175: 13171326.
  • 10
    Clemens, D.L. & Horwitz, M.A. (1995) Characterization of the Mycobaterium tuberculosis phagosome and evidence that phagosomal maturation is inhibited. J Exp Med 181: 257270.
  • 11
    Clinkenbeard, K.D., Mosier, D.A., Confer, A.W. (1989) Transmembrane pore size and role of cell swelling in cytotoxicity caused by Pasteurella haemolytica leukotoxin. Infect Immun 57: 420425.
  • 12
    Dykstra, M.J. (1993) A Manual of Applied Techniques for Biological Electron Microscopy. New York: Plenum Press.
  • 13
    Elfgang, C., Eckert, R., Lichtenberg-Frate, H., Butterwork, A., Traub, O., Klei, R.A., et al. (1995) Specific permeability and selective formation of gap junction channels in connexin-transfected HeLa cells. J Cell Biol 129: 80517.
  • 14
    Elliott, J.A. & Winn, W.C. (1986) Treatment of alveolar macrophages with cytochalasin D inhibits uptake and subsequent growth of Legionella pneumophila. Infect Immun 51: 3136.
  • 15
    Finck-Barbancon, V., Duportail, G., Meunier, O., Colin, D.A. (1993) Pore formation by a two-component leukocidin from Staphylococcus aureus within the membrane of human polymorphonuclear leukocytes. Biochim Biophys Acta 1182: 275282.
  • 16
    Fischer, L.J., Quinn, F.D., White, E.H., King, C.H. (1996) Intracellular growth and cytotoxicity of Mycobacterium haemophilum in a human epithelial cell line (Hec-1-B). Infect Immun 64: 269276.
  • 17
    Fraser, D.W., Tsai, T.R., Orenstein, W., Parkin, W.E., Beecham, H.J., Sharrar, R.G., et al. (1977) Legionnaires' disease: description of an epidemic of pneumonia. N Engl J Med 297: 118997.
  • 18
    Guyton, A.C. (1986) Textbook of Medical Physiology, 7th edn. Philadelphia, PA: W.B. Saunders.
  • 19
    Hakansson, S., Schesser, K., Persson, C., Galyov, E.E., Rosqvist, R., Homble, F., Wolf-Watz, H. (1996) The YopB protein of Yersinia pseudotuberculosis is essential for the translocation of Yop effector proteins across the target cell plasma membrane and displays a contact-dependent membrane disrupting activity. EMBO J 15: 58125823.
  • 20
    High, N., Mounier, J., Prevost, M.C., Sansonetti, P.J. (1992) IpaB of Shigella flexneri causes entry into epithelial cells and escape from the phagocytic vacuole. EMBO J 11: 199199.
  • 21
    Horwitz, M.A. (1983a) Formation of a novel phagosome by the Legionnaires' disease bacterium (Legionella pneumophila) in human monocytes. J Exp Med 158: 131931.
  • 22
    Horwitz, M.A. (1983b) The Legionnaires' disease bacterium (Legionella pneumophila) inhibits phagosome–lysosome fusion in human monocytes. J Exp Med 158: 210826.
  • 23
    Horwitz, M.A. (1987) Characterization of avirulent mutant Legionella pneumophila that survive but do not multiply within human monocytes. J Exp Med 166: 131028.
  • 24
    Husmann, L.K. & Johnson, W. (1994) Cytotoxicity of extracellular Legionella pneumophila. Infect Immun 62: 21112114.
  • 25
    Isberg, R.R. (1994) Intracellular trafficking of Legionella pneumophila within phagocytic cells. In Molecular Genetics of Bacterial Pathogenesis. Miller, V.L., Kaper, J.B., Portnoy, D.A., and Isberg, R.R. (eds). Washington, DC: American Society of Microbiology, pp. 263278.
  • 26
    Isberg, R.R., Voorhis, D.L., Falkow, S. (1987) Identification of invasin: a protein that allows enteric bacteria to penetrate cultured mammalian cells. Cell 50: 769778.
  • 27
    Iwase, M., Lally, E.T., Berthold, P., Korchak, H.M., Taichman, N.S. (1990) Effects of cations and osmotic protectants on cytolytic activity of Actinobacillus actinomycetemcomitans. Infect Immun 58: 17821788.
  • 28
    Keen, M.G. & Hoffman, P.S. (1989) Characterization of a Legionella pneumophila extracellular protease exhibiting hemolytic and cytotoxic activities. Infect Immun 57: 732738.
  • 29
    Kerr, J.F.R., Gobe, G.C., Winterford, C.M., Harmon, B.V. (1995) Anatomical methods in cell death. Methods Cell Biol 46: 127.
  • 30
    King, C.H., Mundayoor, S., Crawford, J.T., Shinnick, T.M. (1993) Expression of contact-dependent cytolytic activity by Mycobacterium tuberculosis and isolation of the genomic locus that encodes the activity. Infect Immun 61: 27082712.
  • 31
    Lalonde, G., McDonald, T.V., Gardner, P., O'Hanley, P.D. (1989) Identification of a hemolysin from Actinobacillus pleuropneumoniae and characterization of its channel properties in planar phospholipid bilayers. J Biol Chem 264: 1355913564.
  • 32
    Lobo, A.L. & Welch, R.A. (1994) Identification and assay of RTX family of cytolysins. Methods Enyzmol 235: 667678.
  • 33
    McGahon, A.J., Martin, S.J., Bissonnette, R.P., Mahboubi, A., Shi, Y., Mogil, R.J., et al. (1995) The end of the (cell) line: methods for the study of apoptosis in vitro. Methods Cell Biol 46: 172173.
  • 34
    Marra, A., Blander, S.J., Horwitz, M.A., Shuman, H.A. (1992) Identification of a Legionella pneumophila locus required for intracellular multiplication in human macrophages. Proc Natl Acad Sci USA 89: 96079611.
  • 35
    Menard, A., Altendorf, K., Breves, D., Mock, M., Montecucco, C. (1996) The vacuolar ATPase proton pump is required for the cytotoxicity of Bacillus anthracis lethal toxin. FEBS Lett 386: 161164.
  • 36
    Menestrina, G. (1986) Ionic channels formed by Staphylococcus aureus alpha toxin: voltage-dependent inhibition by divalent and trivalent cations. J Membr Biol 90: 177190.
  • 37
    Moffat, J.F., Edelstein, P.H., Regula, Jr, D.P., Cirillo, J.D., Tompkins, L.S. (1994) Effects of a Zn-metalloprotease-deficient mutant of Legionella pneumophila in a guinea-pig pneumonia model. Mol Microbiol 12: 693705.
  • 38
    Muller, A., Hacker, J., Brand, B.C. (1996) Evidence for apoptosis of human macrophage-like HL-60 cells by Legionella pneumophila infection. Infect Immun 64: 49004906.
  • 39
    Noronha, F.S.M., Ramalho-Pinto, F.J., Horta, M.F. (1996) Cytolytic activity in the genus Leishmania: involvement of a putative pore-forming protein. Infect Immun 64: 39753982.
  • 40
    Papini, E., Rappuoli, R., Murgia, M., Montecucco, C. (1993) Cell penetration of diphtheria toxin. J Biol Chem 268: 14671574.
  • 41
    Roy, C.R. & Isberg, R.R. (1997) Topology of Legionella pneumphila DotA: an inner membrane protein required for replication in macrophages. Infect Immun 65: 571578.
  • 42
    Sadosky, A.B., Wiater, L.A., Shuman, H.A. (1993) Identification of Legionella pneumophila genes required for growth within and killing of human macrophages. Infect Immun 61: 53615373.
  • 43
    Scherrer, R. & Gerhardt, P. (1971) Molecular sieving by the Bacillus megaterium cell wall and protoplast. J Bacteriol 107: 718735.
  • 44
    Schwab, J., Beckers, C., Joiner, K. (1994) The parasitophorous vacuole membrane surrounding intracellular Toxoplasma gondii functions as a molecular sieve. Proc Natl Acad Sci USA 91: 50913.
  • 45
    Song, L., Hobaugh, M.R., Shustak, C., Cheley, S., Bayley, H., Gouaux, J.E. (1996) Structure of Staphylococcalα-hemolysin, a heptameric transmembrane pore. Science 274: 18591866.
  • 46
    Spack, Jr, E.G., Packard, B., Wier, M.L., Edidin, M. (1986) Hydrophobic adsorption chromatography to reduce non-specific staining by rhodamine-labeled antibodies. Anal Biochem 158: 233237.
  • 47
    Swanson, M.S. & Isberg, R.R. (1995) Association of Legionella pneumophila with the macrophage endoplasmic reticulum. Infect Immun 63: 36093620.
  • 48
    Szeto, L. & Shuman, H.A. (1990) The Legionella pneumophila major secreted protein, a protease, is not required for intracellular growth or cell killing. Infect Immun 58: 25852592.
  • 49
    Thelestam, M. & Florin, I. (1994) Assay of cytopathogenic toxins in cultured cells. Methods Enzymol 235: 679690.
  • 50
    Tosteson, D.C. & Hoffman, J.F. (1960) Regulation of cell volume by active cation transport in high and low potassium sheep red cells. J Gen Physiol 44: 169194.
  • 51
    Vogel, J.P., Roy, C., Isberg, R.R. (1996) Use of salt to isolate Legionella pneumophila mutants unable to replicate in macrophages. Ann NY Acad Sci 797: 271272.
  • 52
    Winn, W.C. & Myerowitz, R.L. (1981) The pathology of the Legionella pneumonias. Hum Pathol 12: 401422.
  • 53
    Wintermeyer, E., Rdest, U., Ludwig, B., Debes, A., Hacker, J. (1991) Characterization of legiolysin (lly ), responsible for haemolytic activity, colour production and fluorescence of Legionella pneumophila. Mol Microbiol 5: 11351143.
  • 54
    Wintermeyer, E., Flugel, M., Ott, M., Steinhert, M., Rdest, U., Mann, K.-H., Hacker, J. (1994) Sequence determination and mutational analysis of the lly locus of Legionella pneumophila. Infect Immun 62: 11091117.