Co-ordination of Legionella pneumophila virulence with entry into stationary phase by ppGpp


  • Brian K. Hammer,

    1. Department of Microbiology and Immunology, The University of Michigan Medical School, 6734 Medical Sciences Building II, Ann Arbor, MI 48109-0620, USA.
    Search for more papers by this author
  • Michele S. Swanson

    1. Department of Microbiology and Immunology, The University of Michigan Medical School, 6734 Medical Sciences Building II, Ann Arbor, MI 48109-0620, USA.
    Search for more papers by this author

Michele S. Swanson. E-mail; Tel. (+1) 734 647 7294; Fax (+1) 734 764 3562.


Legionella pneumophila survives in aquatic environments, but replicates within amoebae or the alveolar macrophages of immunocompromised individuals. Here, the signal transduction pathway that co-ordinates L. pneumophila virulence expression in response to amino acid depletion was investigated. To facilitate kinetic and genetic studies, a phenotypic reporter of virulence was engineered by fusing flaA promoter sequences to a gene encoding green fluorescent protein. When subjected to amino acid depletion, L. pneumophila accumulated ppGpp and converted from a replicative to a virulent state, as judged by motility and sodium sensitivity. ppGpp appeared to initiate this response, as L. pneumophila induced to express the Escherichia coli RelA ppGpp synthetase independently of nutrient depletion accumulated ppGpp, exited the exponential growth phase and expressed flaAgfp, motility, sodium sensitivity, cytotoxicity and infectivity, five traits correlated with virulence. Although coincident with the stationary phase, L. pneumophila virulence expression appeared to require an additional factor: mutant Lp120 accumulated ppGpp and acquired two stationary phase traits but none of six virulence phenotypes analysed. We propose that, when nutrients are limiting, ppGpp acts as an alarmone, triggering the expression of multiple traits that enable L. pneumophila to escape its spent host, to survive and disperse in the environment and to re-establish a protected intracellular replication niche.


Legionella pneumophila, a ubiquitous aquatic Gram-negative bacterium, can parasitize amoebae, the presumed natural reservoir for this pathogen (Fields, 1996). When immune-impaired individuals aspirate or inhale aerosolized water contaminated with L. pneumophila, the pathogen can invade and replicate within alveolar macrophages (Horwitz and Silverstein, 1980) and cause the severe pneumonia Legionnaires' disease (McDade et al., 1977). The aim of this study was to identify the regulatory pathway that enables this microbe to sense and adapt to these distinct extracellular and intracellular environments.

L. pneumophila uses amino acids as the sole source of both carbon and energy (George et al., 1980; Hoffman, 1984). Therefore, whether its local environment is favourable for growth could be gauged by monitoring amino acid levels. A previous study showed that, when exponential phase cells are shifted to broth depleted of amino acids, L. pneumophila converts to a virulent, post-exponential phase, characterized by motility, osmotic resistance, sodium sensitivity, cytotoxicity and competence to avoid phagosome–lysosome fusion (Byrne and Swanson, 1998). Accordingly, the following model of L. pneumophila virulence trait regulation was proposed. Plentiful nutrients and other factors within host cells support bacterial replication. When nutrients become limiting, intracellular bacteria express virulence factors that mediate escape from the spent host, survival and dispersal in the environment and the establishment of a replication niche in a new host cell.

The cellular response to starvation has been studied in detail for a variety of microorganisms. Amino acid depletion of Escherichia coli initiates the stringent response, a rapid inhibition of stable RNA, ribosome and protein synthesis, leading to growth arrest (for a review, see Cashel et al., 1996). This global response is a consequence of the binding of uncharged tRNAs to ribosomes, which activates the enzyme RelA, a guanosine 3′,5′-bispyrophosphate (ppGpp) synthetase. ppGpp accumulation triggers the adaptive response, including the synthesis and accumulation of the stationary phase sigma factor, σs (RpoS) (Gentry et al., 1993). Similarly, (p)ppGpp appears to initiate as a response to amino acid depletion fruiting body development by Myxococcus xanthus (Harris et al., 1998) and antibiotic and pigment production by Streptomyces coelicolor (Chakraburtty and Bibb, 1997). Thus, a variety of microbes use ppGpp as an alarmone to co-ordinate an adaptive response to amino acid starvation.

To investigate whether ppGpp controls entry into stationary phase and expression of virulence by L. pneumophila, the guanine nucleotide pools of replicating and starved cultures were analysed, and the effects of gratuitous relA expression on L. pneumophila growth and virulence were examined. To begin to define genetically the putative virulence regulon of L. pneumophila, the ability of a candidate regulatory mutant, Lp120, to express virulence traits in response to ppGpp was examined. A model is proposed in which ppGpp acts as a second messenger to co-ordinate L. pneumophila virulence expression as an adaptive response to amino acid starvation.


Accumulation of ppGpp in L. pneumophila cultures

The hypothesis that L. pneumophila, like E. coli, accumulates ppGpp in a manner analogous to the stringent response to amino acid depletion was tested. After radiolabelling the guanosine nucleotide pools of L. pneumophila cultures for 6 h (two doubling times), ppGpp levels were analysed by thin-layer chromatography (Cashel, 1969). As a standard for nucleotide localization, we analysed ppGpp accumulation by E. coli carrying a plasmid (pMMBrelA) from which transcription of E. coli ppGpp synthetase relA could be activated by IPTG (Schreiber et al., 1991; Fig. 1A). Similarly, L. pneumophila carrying pMMBrelA also accumulated ppGpp in response to the addition of IPTG (Fig. 1A).

Figure 1.

. ppGpp accumulation in L. pneumophila. After 32P labelling of the nucleotide pools, cell extracts were prepared and separated by PEI-TLC as described in Experimental procedures. A. Nucleotide pools of E. coli strain DH5α/pMMBrelA (lanes 1 and 2) and L. pneumophila strain Lp02/pMMBrelA (lanes 3 and 4) incubated during exponential phase without (−) or with (+) IPTG for 1 h and 4.5 h, respectively, and L. pneumophila strain Lp02 (lanes 5 and 6) grown to the exponential phase (E) and post-exponential phase (PE). Wild-type Lp02 (B) and mutant Lp120 (C) exponential phase cells were transferred to either exponential phase supernatant (− starvation) or post-exponential phase Lp02 supernatant (+ starvation) for the period shown. Then, nucleotide pools were analysed. Representative chromatograms from one of two independent experiments are shown for each L. pneumophila strain. D. Sodium sensitivity of non-radioactive wild-type Lp02 cultures prepared in parallel with those analysed in (B) (open circles, − starvation; closed circles, + starvation). The percentage sodium resistant was determined by plating duplicate samples on CYET and on CYET containing 100 mM NaCl. The mean values obtained from one representative experiment of two performed are shown.

Two approaches were taken to test whether L. pneumophila accumulated ppGpp in response to amino acid starvation. Consistent with studies of growth control in other bacteria (Cashel, 1969; Harris et al., 1998), L. pneumophila ppGpp was below the limit of detection during exponential phase, but accumulated in the stationary phase (Fig. 1A). Next, we analysed ppGpp levels in exponential phase L. pneumophila cells transferred to broth depleted of amino acids, a treatment that induces virulence expression (Byrne and Swanson, 1998). After 30 min in amino acid-depleted medium, the amount of L. pneumophila ppGpp had increased significantly, and it remained elevated for 3 h (Fig. 1B), a period that corresponds to approximately one generation time. Accumulation of pppGpp, the metabolic precursor of ppGpp (Cashel, 1994), was also detected under certain conditions (Fig. 1). Thus, like E. coli, L. pneumophila accumulated ppGpp in response to amino acid depletion.

To address whether ppGpp might initiate conversion of replicating L. pneumophila to the virulent form, the kinetics of ppGpp accumulation and virulence trait expression by L. pneumophila in response to amino acid depletion were compared. In particular, we quantified the efficiency of colony formation on medium containing sodium, which decreases ≈1000-fold when virulent L. pneumophila enter the stationary phase (Sadosky et al., 1993; Vogel et al., 1996; Byrne and Swanson, 1998). As expected, L. pneumophila cultures maintained in the exponential phase were relatively sodium resistant. However, after 2 h in amino acid-depleted medium, sodium sensitivity was apparent; after 3 h, it had increased ≈100-fold (Fig. 1D). Likewise, maximal cytotoxicity and motility, two other activities correlated with L. pneumophila virulence, occur more than 3 h after exponential phase cells are transferred to spent medium (B. Byrne and M. S. Swanson, unpublished; Byrne and Swanson, 1998). That ppGpp accumulates before expression of L. pneumophila virulence traits is consistent with its postulated role as an inducer in this regulatory pathway.

flaAgfp, a reporter of L. pneumophila virulence

To facilitate more detailed genetic and kinetic studies of L. pneumophila virulence regulation, we designed a visual and quantifiable marker of the L. pneumophila virulent phenotype. A number of observations suggested that the L. pneumophila flaA promoter could be used to construct such a genetic reporter. First, L. pneumophila becomes motile as it exits the exponential phase of growth in broth and in cells (Rowbotham, 1986; Byrne and Swanson, 1998). Secondly, flagellar synthesis appears to be regulated co-ordinately with virulence (Pruckler et al., 1995; Merriam et al., 1997). Thirdly, flagella production is regulated during growth in macrophages (Byrne and Swanson, 1998). Fourthly, transcription of the L. pneumophila flagellin gene flaA is regulated by temperature (Heuner et al., 1997). Reasoning that flaA expression was also likely to be regulated by nutrient levels, we constructed a gene fusion in which the L. pneumophila flaA promoter controlled the expression of green fluorescent protein (GFP).

Whether expression of the flaAgfp reporter was growth phase dependent in broth cultures of L. pneumophila was evaluated first. Wild-type strain Lp02 carrying plasmid pflaG expressed flaA specifically in the stationary phase as judged by fluorescence microscopy. Exponential phase cells were neither motile nor fluorescent, whereas at least 95% of post-exponential phase cells were motile and intensely fluorescent (Fig. 2A).

Figure 2.

. flaAgfp is expressed by post-exponential phase L. pneumophila. Wet mounts of exponential phase (E) and post-exponential phase (PE) L. pneumophila Lp02/pflaG (A) and Lp120/pflaG (B) broth cultures were examined by phase and fluorescence microscopy. For wild type, cultures of OD600 0.4 (E) and 2.2 (PE) were analysed; for mutant Lp120, cultures of OD600 0.8 (E) and 2.3 (PE) were analysed. An identical exposure time was used for each of the fluorescence micrographs shown. Similar patterns of expression were observed for several hundred cells examined during each of at least three independent experiments.

To determine the kinetics of flaA induction, GFP-dependent fluorescence of Lp02/pflaG broth cultures growing at 37°C was measured by spectrofluorometry. Bacterial fluorescence increased sharply and more than 30-fold as cultures reached the post-exponential phase (Fig. 3A). Similar results were obtained with cultures expressing flaAgfp on ColE1 replicon pTLPflaG (data not shown). Interestingly, growth phase regulation of GFP-dependent fluorescence was also observed when cells were cultured at 30°C (data not shown), indicating that the promoter sequences used in this study were not sufficient to confer the temperature-regulated flaA expression reported previously (Heuner et al., 1997). Thus, flaAgfp expression coincided with the conversion of L. pneumophila from the replicative to the virulent form.

Figure 3.

. flaAgfp expression is regulated by growth phase in broth cultures of L. pneumophila. Lp02/pflaG (A) and Lp120/pflaG (B) cultures were sampled at ≈3 h intervals for spectrophotometric determination of optical density (circles) and spectrofluorometric analysis of fluorescence intensity (squares), as described in Experimental procedures. Similar results were obtained in two additional independent experiments.

Next, we tested whether flaAgfp was also regulated by L. pneumophila growing intracellularly. Macrophages were infected with post-exponential phase Lp02/pflaG at a low multiplicity to ensure that only one bacterium entered each macrophage. Then, expression of flaAgfp was assessed by fluorescence microscopy. Macrophages infected for 1 h typically harboured a single fluorescent bacterium (Fig. 4). After 5 h, bacterial replication was evident, but GFP-mediated fluorescence of the bacteria was markedly diminished (Fig. 4). By 19 h after infection, macrophages contained large numbers of non-fluorescent bacteria (Fig. 4). After 21 h, when macrophage lysis was first apparent by phase microscopy, flaAgfp fluorescence was visible within the crowded phagosome (Fig. 4). Likewise, intracellular L. pneumophila become sodium sensitive and flagellated during this late stage of infection (Byrne and Swanson, 1998). That flaAgfp expression was coincident with the virulent phase of L. pneumophila cultured in broth or in macrophages indicates that it is a valid phenotypic marker of the virulent state.

Figure 4.

. flaAgfp expression by L. pneumophila growing intracellularly. Mouse macrophages infected with stationary phase Lp02/pflaG for the times shown were fixed and stained with DAPI to localize macrophage nuclei and intracellular bacteria. GFP-mediated fluorescence was detected immediately after infection (0 h), but diminished during the intracellular replication period (5 and 19 h). flaAgfp expression was detected again at the end of the primary infection period when cell lysis began to occur (21 h). Representative results from one of two independent experiments are shown.

Effects of E. coli relA induction on exponential phase L. pneumophila cultures

In the virulent, post-exponential phase, L. pneumophila contained elevated levels of ppGpp, whereas replicating bacteria did not (Fig. 1). Whether ppGpp can trigger conversion of replicating L. pneumophila to the virulent state was examined using a molecular genetic approach previously applied by others (Schreiber et al., 1991; Harris et al., 1998). To study the effect of ppGpp accumulation independent of nutrient levels or growth state, we used plasmid pMMBrelA from which transcription of E. coli relA by L. pneumophila could be activated by IPTG (Fig. 1A). After inducing exponential cultures to express ppGpp synthetase, we measured their growth and the expression of four traits previously correlated with virulence: flaAgfp-dependent fluorescence, sodium sensitivity, cytotoxicity and infectivity.

To learn whether gratuitous ppGpp synthetase expression triggered a switch to the virulent state, flaAgfp-dependent fluorescence of cells induced to express relA was analysed. For this purpose, a L. pneumophila strain carrying both pMMBrelA and a compatible replicon encoding flaAgfp (pTLPflaG) was constructed. As expected, the fluorescence of wild-type cells carrying both plasmids increased upon entry into stationary phase (Fig. 5A). However, exponential phase cells induced to express relA also became fluorescent, concomitant with an apparent decrease in the growth rate (Fig. 5A). When wet mounts were observed by phase-contrast and fluorescence microscopy, at least 90% of the cells were fluorescent and 10% were motile (data not shown). Thus, an accumulation of ppGpp appeared to trigger the transition into stationary phase and initiate flagellar biosynthesis.

Figure 5.

. Growth and GFP-mediated fluorescence of exponential phase L. pneumophila induced to express relA. Cultures of L. pneumophila wild-type Lp02 (A) or mutant Lp120 (B) strains containing both pTLPflaG and pMMBrelA were divided at the time indicated (arrows); one was treated with IPTG to induce relA (closed symbols) and the other was not (open symbols). At the times shown, samples were collected for spectrophotometric determination of optical density (circles) and spectrofluorometric determination of relative fluorescence (squares). Similar results were obtained in two other independent experiments.

In response to amino acid depletion, L. pneumophila accumulated ppGpp and converted to a sodium-sensitive form (Fig. 1). Therefore, we used a culture of L. pneumophila carrying pMMBrelA to investigate whether ppGpp was sufficient to trigger sodium sensitivity independent of culture conditions. Whereas uninduced exponential phase cultures were relatively sodium resistant, cultures induced to express relA were ≈1000-fold more sensitive to sodium (Fig. 6A), similar to the effect measured for amino acid-starved cultures (Fig. 1D) and post-exponential phase cultures (Fig. 6A and Table 1[link]).

Figure 6.

. Sodium sensitivity of exponential phase L. pneumophila induced to express relA. Sodium sensitivity of cultures of wild-type Lp02 (A) and mutant Lp120 (B) strains containing pMMBrelA grown to the exponential (E) or the post-exponential (PE) phase that either were induced for 4.5 h with IPTG (+) or were not (−) was determined by plating duplicate samples on CYET and CYET supplemented with 100 mM NaCl. The values plotted represent the mean ± standard error determined in three independent experiments.

Table 1.  . Comparison of L. pneumophila avirulent mutant Lp120 with wild-type Lp02 during exponential and post-exponential growth. a. The percentage sodium resistant was measured as described in Fig. 6. Values represent the mean ± standard error for three experiments performed in duplicate.b. The percentage osmotic resistant was measured by incubating broth cultures for 1 h in AYET that contained 0.3 M KCl (AYET + KCl) or did not, diluting in water or AYET + KCl, then plating duplicate samples on CYET to quantify cfu. The values represent the mean ± standard error for three experiments performed in duplicate.c. Percentage heat resistant was determined by diluting broth cultures in H2O to ≈1 × 107 cfu, incubation at room temperature or 57°C for 20 min, then plating on CYET to quantify cfu. The values represent the mean ± standard error for three experiments performed in duplicate.Thumbnail image of

We next examined whether gratuitous RelA activity induced contact-dependent cytotoxicity, a stationary phase activity thought to be important for L. pneumophila to establish and/or escape from its replication vacuole within amoebae and macrophages (Husmann and Johnson, 1994; Byrne and Swanson, 1998; Kirby and Isberg, 1998; Kirby et al., 1998). To determine whether relA induction was sufficient to convert exponential phase L. pneumophila to the cytotoxic form, exponential phase cultures of Lp02/pMMBrelA were treated with IPTG for 4.5 h; then, their cytotoxicity was compared with that of two control cultures. As expected, stationary phase cultures killed macrophages efficiently, while exponential phase cultures of Lp02/pMMBrelA did not (Fig. 7A). However, exponential phase cultures induced to express relA killed macrophages with an efficiency similar to that of stationary phase cultures (Fig. 7A). Therefore, ppGpp accumulation triggered L. pneumophila to express cytotoxicity, a trait correlated with virulence.

Figure 7.

. Cytotoxicity to macrophages of exponential phase L. pneumophila induced to express relA. Broth cultures of Lp02/pMMBrelA (A) and Lp120/pMMBrelA (B) grown to the exponential phase and treated for 4.5 h with IPTG (triangles) or not (circles) or to the stationary phase (squares) were added to triplicate wells of macrophage cultures at the multiplicities of infection (MOIs) shown. After a 1 h incubation, cytotoxicity was measured as the ability of viable macrophages to reduce the colorimetric dye Alamar blue as described in Experimental procedures. The values plotted represent the mean ± standard error for triplicate samples determined in one of three similar experiments.

Finally, we analysed whether relA induction enhanced L. pneumophila entry and survival in macrophages. Consistent with results published previously, within 2 h of infection, ≈5% of stationary phase L. pneumophila cells carrying pMMBrelA associated with macrophages (Fig. 8), whereas less than 1.0% of exponential phase Lp02/pMMBrelA cells did so (Fig. 8). However, exponential phase Lp02/pMMBrelA cells expressing relA entered and survived in macrophages at least as efficiently as stationary phase cells (11%; Fig. 8). Together, the results of this series of experiments indicate that, when L. pneumophila is starved of amino acids, ppGpp accumulates and initiates both entry into stationary phase and expression of multiple virulence traits, including motility, sodium sensitivity, cytotoxicity and enhanced infectivity.

Figure 8.

. Infection efficiency of exponential phase L. pneumophila induced to express relA. Wild-type strain Lp02 containing pMMBrelA was grown to the post-exponential phase (PE) or to the exponential phase (E) and treated with IPTG for 4.5 h (+) or not (−) before infection of macrophages. The number of cfu added to macrophage monolayers at 0 h (white bars) and the number associated with monolayers after a 2 h infection period (black bars) was determined by plating onto CYE triplicate samples of infection medium and macrophage lysates respectively. The mean cfu ± standard error determined in one of two similar experiments are shown.

Avirulent mutant Lp120 may identify a virulence regulator

Although co-ordinate expression of several L. pneumophila virulence traits has been observed in the stationary phase (Byrne and Swanson, 1998), the regulatory proteins that presumably control the timing of virulence expression have not yet been identified. To begin to identify such regulators genetically, we considered the phenotypes of a collection of intracellular growth mutants characterized previously (Swanson and Isberg, 1996a). When mutant Lp120 is cultured within macrophages, the majority of bacteria in the inoculum are degraded in the lysosomes, while a minority replicates as efficiently as wild type (Swanson and Isberg, 1996a). Based on the supposition that the partial loss of function of a virulence activator could cause the observed heterogeneity, mutant Lp120 was chosen for further study.

Reasoning that a mutation in a global regulator would probably have a pleiotropic effect on virulence, we compared mutant Lp120 and wild-type cultures in four assays of virulence. First, although virulent L. pneumophila converts from a sodium-resistant to a sodium-sensitive form upon exiting exponential phase, mutant Lp120 remained sodium resistant in all stages of growth (Table 1). Secondly, unlike stationary phase cultures of wild-type L. pneumophila, broth cultures of Lp120 did not exhibit contact-dependent cytotoxicity at any growth phase or bacterial density measured (Fig. 7B, data not shown). Finally, post-exponential phase cultures of Lp120/pflaG did not induce flaAgfp expression, as judged by microscopic examination (Fig. 2B) and spectrofluorometry (Fig. 3B), nor did cells become motile (data not shown). Because it was completely defective for the expression of all four virulence-associated phenotypes assessed, including cytotoxicity, sodium sensitivity, motility and expression of flaAgfp, and is partially defective for growth in macrophages, evasion of phagosome–lysosome fusion and association with the endoplasmic reticulum (Swanson and Isberg, 1996a,b), mutant Lp120 is likely to harbour a mutation affecting a global activator of L. pneumophila virulence.

Because the expression of L. pneumophila virulence is co-ordinately regulated with entry into stationary phase, a common signal transduction pathway may control both. To investigate whether the putative Lp120 activator was also required for the expression of stationary phase phenotypes, we compared the expression by wild-type and mutant Lp120 cells of two hallmark stationary phase traits described for E. coli : resistance to osmotic shock (McCann et al., 1991) and heat shock (Jenkins et al., 1990).

Unlike wild-type E. coli and L. pneumophila, which become resistant to osmotic shock upon entering post-exponential phase (McCann et al., 1991; Byrne and Swanson, 1998), mutant Lp120 was osmotically sensitive during both exponential and post-exponential phases (Table 1). Like E. coli, wild-type L. pneumophila acquired heat resistance in the stationary phase: after a 20 min incubation at 57°C, less than 0.01% of exponential phase cells were recovered, whereas 20% of post-exponential cells survived the heat shock (Table 1). Interestingly, in this assay, mutant Lp120 resembled wild-type Lp02: cells were heat sensitive in the exponential phase, but heat resistant in the post-exponential phase (Table 1). The phenotype of mutant Lp120 predicts that L. pneumophila encodes a factor that co-ordinately regulates a number of virulence genes, but is dispensable for ppGpp-dependent growth arrest and the expression of at least one stationary phase trait, heat resistance.

Gratuitous relA expression does not restore virulence gene expression by mutant Lp120

Its pleiotropic phenotype suggested that mutant Lp120 either fails to synthesize ppGpp or lacks a regulatory factor that responds to ppGpp accumulation. To investigate where in the signal transduction pathway the Lp120 gene product acts, we subjected exponential phase cultures of the mutant to amino acid depletion, then analysed ppGpp synthesis. Similar to wild type, mutant cells accumulated ppGpp within 0.5 h of transfer to supernatant obtained from wild-type stationary phase cultures, and the levels remained elevated for at least 3 h (Fig. 1C). Therefore, mutant Lp120 appeared to be competent to respond to amino acid depletion by synthesizing ppGpp.

As an independent approach to place the putative Lp120 activator in the virulence regulation pathway, we investigated whether gratuitous relA expression suppressed the virulence defects of the mutant. In particular, we examined the effect of gratuitous relA expression on mutant Lp120 growth and three virulence-associated traits: flaAgfp expression, sodium sensitivity and cytotoxicity. Lp120/pMMBrelA, like Lp02/pMMBrelA, responded to relA induction by reducing its growth rate (Fig. 5B). Therefore, the Lp120 mutation did not interfere with the accumulation of ppGpp or the subsequent growth response. In contrast to its effects on wild-type L. pneumophila, replicating mutant cells induced to express relA did not convert to the virulent form. Broth cultures of mutant Lp120 expressing the E. coli relA gene were amotile and did not express flaAgfp (Fig. 5B), sodium sensitivity (Fig. 6B) or cytotoxicity (Fig. 7B). Together, these data support the hypothesis that mutant Lp120 lacks a factor that co-ordinates L. pneumophila virulence expression in response to ppGpp, the intracellular signal of amino acid starvation.


In its natural habitat, the opportunistic pathogen L. pneumophila survives in fresh water but replicates within amoebae. Its ability to sense and respond to these alternative conditions is presumably critical for survival and must require co-ordinate regulation of particular traits required in each environment. In broth cultures, L. pneumophila responds to amino acid depletion by switching from a replicative to a virulent form, a state characterized by cytotoxicity, osmotic resistance, motility, sodium sensitivity and the capacity to evade lysosomal degradation (Byrne and Swanson, 1998). This report demonstrates that, in response to amino acid depletion, L. pneumophila accumulated ppGpp. Furthermore, gratuitous expression of the RelA ppGpp synthetase induced ppGpp accumulation and triggered L. pneumophila virulence expression, independent of nutrient levels or cell density. Finally, in response to amino acid depletion, mutant Lp120 produced ppGpp, exited the exponential phase and became heat resistant, yet failed to convert to the virulent phenotype. Accordingly, we propose that ppGpp acts as an alarmone, which requires the Lp120 factor to co-ordinate expression of L. pneumophila virulence traits as an adaptive response to amino acid starvation (Fig. 9). Virulent progeny can then efficiently escape from the spent host, survive and disperse in the aquatic environment and re-establish a protected intracellular replication niche.

Figure 9.

. Model for co-ordination of L. pneumophila virulence by ppGpp as described in the text. The ‘effectors of virulence’ refer to the class of genes presumably required for traits correlated with virulence, including motility, sodium sensitivity, cytotoxicity and evasion of phagosome–lysosome fusion.

ppGpp has been shown to initiate in a variety of environmental microbes developmental programs that are critical for surviving periods of starvation (Chakraburtty and Bibb, 1997; Wendrich and Marahiel, 1997; Harris et al., 1998). In response to amino acid depletion, RelA synthesis of ppGpp by Myxococcus xanthus initiates the formation of a multicellular fruiting body, which subsequently differentiates into environmentally resistant myxospores (Harris et al., 1998). ppGpp accumulation by Streptomyces coelicolor plays a role in the production of antibiotic and the pigmentation characteristic of mature spores (Chakraburtty and Bibb, 1997). When the soil bacterium Bacillus subtilis is starved of amino acids, ppGpp accumulates and activates the expression of stress response proteins that may contribute to sporulation (Wendrich and Marahiel, 1997). Similarly, the aquatic bacterium L. pneumophila appears to use ppGpp to trigger the expression of virulence traits, which may facilitate its escape from a nutrient-poor environment and subsequent transmission to a new host.

In E. coli, accumulation of ppGpp alters cell physiology dramatically. In addition to a rapid block in the synthesis of proteins and stable RNAs, known as the stringent response, effects of ppGpp on metabolism, DNA replication, cell cycle control and accumulation of stationary phase sigma factor σs (RpoS) have been described (for a review, see Cashel et al., 1996). However, the molecular mechanism by which ppGpp arrests bacterial growth and induces stationary phase processes is not completely understood. Indirect evidence indicates that inhibition of RNA synthesis during the stringent response may result from direct binding of ppGpp to RNA polymerase (Reddy et al., 1995). Numerous studies have described the effects of ppGpp on transcription at either the level of initiation (Baracchini and Bremer, 1988), pausing (Hernandez and Bremer, 1993) or readthrough (Faxén and Isaksson, 1994), as well as on translation (Svitil et al., 1993), underscoring the complexity and lack of consensus regarding the mechanism of ppGpp-mediated growth arrest.

E. coli strains that lack RelA and SpoT, two ppGpp synthetases, have a pleiotropic phenotype similar to that of rpoS mutants, suggesting that much of the stringent response is a consequence of RpoS induction by ppGpp (Cashel et al., 1996). By analogy with E. coli, RpoS may activate expression of L. pneumophila virulence genes in the stationary phase. Indeed, RpoS is required for maximal virulence of several pathogens, including Salmonella species (Fang et al., 1992), Shigella flexneri (Small et al., 1994), toxigenic E. coli (Waterman and Small, 1996) and the plant pathogen Erwinia carotovora (Mukherjee et al., 1998). However, it has been reported that an L. pneumophila rpoS mutant strain resembles wild type in its capacity to a kill a macrophage cell line (L. M. Hales and H. A. Shuman, unpublished). Whether RpoS is required for expression of the panel of virulence traits examined here remains to be determined.

Based on its pleiotropic phenotype and its ability to synthesize ppGpp, we predict that the Lp120 mutation will identify an activator of the L. pneumophila virulence regulatory pathway (Fig. 9). Whether Lp120 lacks an RpoS activity has not been tested directly. However, like E. coli rpoS mutants, stationary phase Lp120 cells remain elongated (Fig. 2B) and sensitive to osmotic shock (Table 1), and they lose viability when subjected to prolonged incubation in nutrient-limiting conditions (B. K. Hammer and M. S. Swanson, unpublished; Lange and Hengge-Aronis, 1991a,b; McCannet al., 1991). On the other hand, mutant Lp120 cells acquire heat resistance in the post-exponential phase (Table 1), a trait that is RpoS dependent in E. coli (Lange and Hengge-Aronis, 1991b). Interpretation of the phenotype of mutant Lp120 is complicated: because this strain was generated by chemical mutagenesis, it may carry a partial loss-of-function allele or multiple mutations. Therefore, isolation and characterization of the respective null mutant strain is required to define the role of the Lp120 gene product in the co-ordinate expression of virulence and stationary traits by L. pneumophila.

We have presented evidence that an extracellular signal, amino acid starvation (Byrne and Swanson, 1998), stimulates the production of an intracellular second messenger, ppGpp, which co-ordinates the expression of traits critical for L. pneumophila virulence. It remains possible that additional environmental signals act through this or another signal transduction pathway to regulate L. pneumophila virulence expression. For example, environmental regulation of S. typhimurium virulence requires numerous transcriptional regulators (Guiney et al., 1995; Slauch et al., 1997). Nonetheless, the stringent response paradigm will probably facilitate identification of the presumptive transcriptional regulators and their target effector genes, which together form the L. pneumophila virulence regulon.

Experimental procedures

Bacterial strains and culture conditions

L. pneumophila strain Lp02, a virulent thymine auxotroph (Berger and Isberg, 1993), and Lp120, an intracellular growth mutant isolated by chemical mutagenesis of Lp02 (Swanson and Isberg, 1996a), were grown in N-(2-acetamino)-2-aminoethanesulphonic acid (ACES; Sigma)-buffered yeast extract broth supplemented when necessary with 100 μg ml−1 thymidine (AYET) at 37°C with agitation. Bacteria were plated on ACES-buffered charcoal yeast extract agar supplemented with 100 μg ml−1 thymidine (CYET) at 37°C. Where indicated, chloramphenicol was added to a final concentration of 5 μg ml−1 and gentamicin to 10 μg ml−1.

Macrophage cultures

Bone marrow-derived macrophages were prepared from female A/J mice (Jackson Laboratory) as described previously (Swanson and Isberg, 1995). After a 7 day culture period in L-cell conditioned medium, macrophages were collected by centrifugation, suspended in RPMI-1640 containing 10% fetal bovine serum (RPMI-FBS; Gibco BRL), and plated as described below for the analysis of intracellular flaAgfp expression and infectivity.


To exploit the observed correlation between motility and expression of virulence traits, a genetic reporter was constructed in which transcription of a gene encoding green fluorescent protein (GFP) was controlled by the L. pneumophila flaA promoter. Plasmid pflaG was generated by subcloning the 0.8 kb XbaI/Pst I fragment encoding GFP and the phage T7 gene 10 ribosome binding site obtained from plasmid pGFPmut3 (Cormack et al., 1996) into the multiple cloning site (MCS) of the IncQ plasmid pJB98, which encodes thymidylate synthetase as a selectable marker. Plasmid pJB98 was generated from pKB5 (Berger and Isberg, 1993) by deletion of an ≈1.5 kb HpaI/EcoRI fragment containing the Ptac promoter and the lacIq gene (J. P. Vogel and R. R. Isberg, unpublished). The flaA promoter sequences, including its −35, −10 and transcriptional start sites, were amplified from wild-type Lp02 chromosomal DNA by the polymerase chain reaction (PCR). The forward primer was 5′-AAA AAG GAT CCT TTG CGA CTT TTC AAA AAA GC-3′, which introduced a BamHI restriction site, and the reverse primer was 5′-AAA AAT CTA GAC CTG AAT CCC TGG TAT GGG AC-3′, which introduced an XbaI restriction site. The BamHI/XbaI flagellar promoter product (PflaA) generated from Lp02 chromosomal DNA was ligated to the BamHI and XbaI sites located directly upstream of the gfp gene (Heuner et al., 1997). The plasmid pTLPflaG was constructed by subcloning an ≈1 kb BamHI/SphI flaAgfp fragment from pflaG into the ≈4.6 kb BamHI/SphI fragment of the ColE1 replicon pTLP6 (McClain et al., 1996), which encodes chloramphenicol resistance as a selectable marker.

To analyse the effects of expression of the E. coli relA gene in E. coli and L. pneumophila, plasmid pMMBrelA was constructed by subcloning the EcoRI/PstI relA fragment of pALS13 (Svitil et al., 1993), which encodes a truncated, metabolically active form of RelA (Schreiber et al., 1991), into the MCS of the broad-host-range IncQ vector pMMB67EH (Frey et al., 1983). The location of the MCS directly downstream of the Ptac promoter and the presence of lacIq on pMMB67EH enabled IPTG induction of the E. coli relA gene. Preliminary experiments determined that 200 μM IPTG was sufficient for maximum growth arrest (data not shown). To permit selection in L. pneumophila, an EcoRI fragment encoding gentamicin resistance was treated with Klenow enzyme, then ligated to the unique DraI site of pMMB67EH.

Detection of 32P-labelled ppGpp in L. pneumophila

Analysis of ppGpp accumulation by thin-layer chromatography was performed using an established protocol (Cashel, 1969). To provide a standard for ppGpp localization, we analysed the nucleotide pools of E. coli strain DH5α carrying pMMBrelA, which were or were not induced with 200 μM IPTG for 1 h, as described previously (Schreiber et al., 1991). To ensure uniform labelling of nucleotide pools, L. pneumophila wild-type Lp02 and mutant Lp120 cultures grown in AYET were labelled with carrier-free [32P]-phosphoric acid at 100 μCi ml−1 (ICN Pharmaceuticals) for ≈6 h (two generation times) before sampling.

To analyse the effect of relA induction on ppGpp accumulation in L. pneumophila, aliquots were removed after incubation of exponential phase Lp02/pMMBrelA (OD600 ≈0.2) cultures without or with 200 μM IPTG for 4.5 h. For experiments determining the growth phase-dependent accumulation of ppGpp, analysis was performed with aliquots collected from wild-type L. pneumophila strain Lp02 grown to the exponential phase (OD600 ≈0.4) and stationary phase (OD600 ≈2.1). To subject L. pneumophila to amino acid depletion, aliquots of L. pneumophila wild-type Lp02 and mutant Lp120 cultures grown to exponential phase (OD600 ≈0.4) were collected by centrifugation and resuspended with supernatants obtained by filter sterilization of either corresponding exponential phase non-radioactive cultures (OD600 ≈0.4) or post-exponential phase Lp02 supernatant (OD600 = 2.1–2.3). Aliquots were removed for analysis before dividing the culture and at designated times after starvation. The starvation protocol was essentially as described previously (Byrne and Swanson, 1998), except the supernatants were supplemented with thymidine (100 μg ml−1), cysteine (400 μg ml−1) and ferric nitrate (135 μg ml−1), and 32P (100 μCi ml−1) was added to labelled culture supernatants.

Formic acid extracts of each 50 μl culture aliquot were prepared, applied to a PEI-cellulose thin-layer chromatography (TLC) sheet (20 × 20 cm) and developed with 1.5 M KH2PO4 (pH 3.4) as described previously (Cashel, 1969; Cashel, 1994). Equal volumes (40 μl) of extracts were chromatographed in each experiment, except for the volume of post-exponential phase extract in the growth phase experiment, which was reduced to compensate for the calculated increase in cfu ml−1 with growth phase. Autoradiographs of each chromatogram were converted to digital images with a Hewlett Packer ScanJet IIcx scanner, and the image quality was optimized using Adobe Photoshop software (Adobe Systems). Optical density and cfu ml−1 were measured for non-radioactive cultures grown under identical conditions.

Osmotic sensitivity

Osmotic sensitivity was measured by diluting aliquots of exponential and stationary phase cultures to ≈1 × 107 cfu ml−1 AYET that did not or did contain 0.3 M KCl (AYET + KCl). After incubation for 1 h at 37°C with agitation, the cultures were diluted with either H2O or AYET + KCl, then plated onto CYET. The percentage of bacteria that were osmotically resistant was determined for samples incubated in AYET + KCl according to the following formula: (cfu ml−1‘H2O-diluted’ samples)/(cfu ml−1‘AYET + KCl-diluted’ samples) × 100. As expected, the plating efficiencies of control broth samples treated only with H2O or only with AYET + KCl were identical.

Heat resistance

Heat resistance was quantified by diluting cultures to ≈1 × 107 cfu ml−1 H2O, transferring a 300 μl aliquot of each sample to a 1.5 ml culture tube, then incubating for 20 min at room temperature or in a 57°C water bath. Next, cultures were diluted in H2O and plated on CYET to enumerate surviving cfu. The percentage of bacteria that were heat resistant was determined by the following calculation: (cfu ml−1 heated sample)/(cfu ml−1 unheated sample) × 100.

Sodium sensitivity

The sodium sensitivity of wild-type Lp02 and mutant Lp120 broth cultures was quantified by diluting aliquots in H2O and plating on CYET containing or lacking 100 mM NaCl. The percentage of bacteria that were sodium resistant was determined by the following calculation: (cfu ml−1 on CYET + 100 mM NaCl)/(cfu ml−1 on CYET) × 100 as described previously (Byrne and Swanson, 1998).


Cytotoxicity of wild-type Lp02 and mutant Lp120 carrying plasmid pMMBrelA was determined essentially as described previously (Byrne and Swanson, 1998). Exponential phase cultures at OD600 < 1.0 were divided into two samples: one sample was induced with 200 μM IPTG and the other was not. After an additional 4.5 h incubation, both exponential phase cultures and an untreated stationary phase culture were each resuspended in RPMI-FBS and added at the indicated ratios to macrophage monolayers in 96-well tissue culture plates. The multiplicity of infection (MOI) was calculated by plating the broth cultures on CYET in duplicate. After incubation at 37°C for 1 h, extracellular bacteria were washed away, and the macrophages were incubated with RPMI-FBS containing 10% Alamar blue (Acumed) for ≈4 h to allow viable cells to reduce the colorimetric dye. To generate a standard curve, twofold dilutions of uninfected macrophages cultures in the range of ≈5 × 104 (100%) to 1.6 × 103 (3.1%) per well were included in each assay. The absorbance of samples was determined at 570 nm and 600 nm using a Microtek plate reader. The percentage of macrophages that were viable was calculated from the average A570/600 of duplicate or triplicate samples, and the slope of the plot of A570/600 versus uninfected macrophages was obtained from the standard curve.


The infectivity assay was performed essentially as described previously (Byrne and Swanson, 1998). Because strains carrying plasmid pMMBrelA were gentamicin resistant, extracellular bacteria were removed at 2 h by washing monolayers three times with RPMI rather than by gentamicin treatment. Colony-forming units added at time 0 h and intracellular cfu at time 2 h were calculated from plating triplicate aliquots on CYET.

flaAgfp expression in broth cultures and in infected macrophages

Expression of flaAgfp by L. pneumophila broth cultures was analysed by phase-contrast and fluorescence microscopy. Wet mounts of exponential and post-exponential phase cultures were photographed with Kodak T-Max ASA 400 film. Negatives were converted to digital images with an Archiboldi Leafscan-45 film scanner, and the image quality was optimized using Adobe Photoshop software (Adobe Systems).

Expression of flaAgfp by intracellular L. pneumophila was analysed by fluorescence microscopy. Macrophages cultured on 12 mm glass coverslips were infected at an MOI of ≈1 for 1 h, washed to remove the majority of extracellular bacteria and then incubated in fresh medium. At that time (0 h), and after specific time periods (5, 19 and 21 h), a set of coverslips was fixed with periodate–lysine–paraformaldehyde (McLean and Nakane, 1974) prewarmed to 37°C, washed three times with PBS, methanol extracted and again washed three times with PBS before staining the bacterial DNA fluorescently with 0.1 μg of 4′,6-diamidino-2-phenylindole (DAPI) per ml of PBS. Photographs were processed as above.

Quantification of fluorescence

Expression of flaAgfp was quantified by fluorometry for L. pneumophila wild-type Lp02 and mutant Lp120 cultured in AYET containing appropriate antibiotics. For experiments measuring the effect of relA expression on GFP-mediated fluorescence, 200 μM IPTG was added to one-half of the culture at the indicated time. At the times indicated, one aliquot of each culture was collected to determine optical density. To ensure that similar bacterial concentrations were analysed by fluorometry, one aliquot of cells was then collected by centrifugation and resuspended in water to an OD600 of ≈0.10. The maximum signal was established using a stationary phase culture of wild-type Lp02 expressing flaAgfp, and all fluorescence values for a given experiment were determined relative to the positive control. Relative fluorescence intensity was measured on a SPF-500C spectrophotometer (SLM Instruments) using an excitation of 488 nm with a bandpass width of 2 nm, and emission of 510 nm with bandpass width of 1 nm.


We are grateful to Melody Neely for critical advice and insight during the course of this work, Eric Olson for suggestions regarding this manuscript, and Amrita Joshi for providing the pTLPflaG plasmid. We also thank David Friedman for assistance with ppGpp measurements, and Joel Swanson for assistance with spectrofluorometry. This work was supported by the National Institutes of Health (R29AI40694-01 BM) and the Rackham Faculty Research Fund. B.H. was supported in part by the National Institutes of Health Cellular Biotechnology Training Grant (T32GM08353).