During its life cycle, Legionella pneumophila alternates between a replicative and a transmissive state. To determine their contributions to L. pneumophila differentiation, the two ppGpp synthetases, RelA and SpoT, were disrupted. Synthesis of ppGpp was required for transmission, as relA spoT mutants were killed during entry to and exit from macrophages. RelA, which senses amino acid starvation induced by serine hydroxamate, is dispensable in macrophages, as relA mutants spread efficiently. SpoT monitors fatty acid biosynthesis (FAB), since following cerulenin treatment, wild-type and relA strains expressed the flaA transmissive gene, but relA spoT mutants did not. As in Escherichia coli, the SpoT response to FAB perturbation likely required an interaction with acyl-carrier protein (ACP), as judged by the failure of the spoT-A413E allele to rescue transmissive trait expression of relA spoT bacteria. Furthermore, SpoT was essential for transmission between macrophages, since secondary infections by relA spoT mutants were restored by induction of spoT, but not relA. To resume replication, ppGpp must be degraded, as mutants lacking spoT hydrolase activity failed to convert from the transmissive to the replicative phase in either bacteriological medium or macrophages. Thus, L. pneumophila requires SpoT to monitor FAB and to alternate between replication and transmission in macrophages.
To cope with environmental fluctuations such as changes in temperature, osmolarity and nutrient availability, bacteria modify their physiology. To increase resilience and promote survival in response to deteriorating conditions, some bacteria also undergo morphological adaptations. Examples of microbial differentiation include fruiting body formation by soil dwelling Myxococcus xanthus (Zusman et al., 2007), spore formation by species of Bacillus and Clostridium (Paredes et al., 2005), and alternation between replication and transmission by the intracellular pathogens Coxiella burnetti (Voth and Heinzen, 2007), Chlamydia trachomatis (Abdelrahman and Belland, 2005) and Legionella pneumophila (Swanson and Hammer, 2000; Molofsky and Swanson, 2004). In each case, differentiation is integral to the microbe's resilience and versatility.
Legionella pneumophila is ubiquitous in aquatic environments where it resides either in biofilms or within freshwater protozoa. When humans inhale aerosols of contaminated water, this opportunistic pathogen can infect alveolar macrophages. Upon phagocytosis, transmissive L. pneumophila resist phagosome acidification and maturation and instead become sequestered within a protective, ER-derived vacuole (Shin and Roy, 2008). Favourable metabolic conditions within the vacuole prompt repression of transmissive genes and induction of genes required for protein synthesis and replication (Molofsky and Swanson, 2003; Sauer et al., 2005; Bruggemann et al., 2006). Following multiple rounds of cell division that presumably exhaust host cell nutrients, the bacteria differentiate back to the transmissive form, which can subsequently develop into a ‘mature intracellular form’ suited for prolonged viability in the environment (Garduño et al., 2008). In broth cultures, lysosome evasion, type IV secretion, flagella expression, sodium sensitivity and cytotoxicity are co-ordinately induced in stationary phase (Byrne and Swanson, 1998; Alli et al., 2000; Molofsky and Swanson, 2004); since each trait is predicted to enhance spread from one cell to another, we refer to this as the transmissive phase. In both broth and amoebae, L. pneumophila exhibits a biphasic gene expression pattern that corresponds to the phenotypic switch observed (Bruggemann et al., 2006). Thus, the transition between the exponential (E) phase and the post-exponential (PE) phase in broth cultures reproduces in large part the differentiation from the replicative to the transmissive phase observed in host cells.
In Gram-negative bacteria such as E. coli and Salmonella spp., the levels of ppGpp in the cell are regulated by the ppGpp synthetase RelA and the bifunctional synthetase/hydrolase SpoT (Potrykus and Cashel, 2008). During amino acid starvation, uncharged tRNAs trigger rapid synthesis of ppGpp by ribosome-bound RelA. During exponential growth, SpoT hydrolyses ppGpp. Since SpoT hydrolase activity is required to prevent RelA-dependent ppGpp from accumulating unabatedly, spoT is essential to bacteria that encode relA. When the RelA synthetase is not active, spoT is dispensable for replication and its function can be deduced by comparing relA mutants to relA spoT strains (Potrykus and Cashel, 2008).
SpoT can also generate ppGpp in response to a variety of stresses, including carbon source deprivation, phosphate starvation, iron starvation and FAB inhibition (Potrykus and Cashel, 2008). For E. coli to respond to perturbations in FAB, SpoT interacts with acyl-carrier protein (ACP), a protein critical for FAB (Battesti and Bouveret, 2006). Point mutations within the C-terminal TGS domain of SpoT not only disrupt its interaction with ACP, but also alter its activity such that ppGpp synthesis is favoured over hydrolysis (Battesti and Bouveret, 2006). Phosphate starvation results in SpoT-dependent ppGpp accumulation that requires SpoT hydrolase activity, not synthetase activity (Spira and Yagil, 1998; Bougdour and Gottesman, 2007). It remains to be determined if other starvation conditions that elicit ppGpp accumulation alter the SpoT–ACP interaction or result from either an inhibition of ppGpp hydrolysis or an increase in synthesis.
Legionella pneumophila requires RelA both for ppGpp accumulation upon entry into stationary phase and for maximal expression of the primary flagellar subunit flaA (Zusman et al., 2002). However, RelA is dispensable for L. pneumophila growth in Acanthamobae castellanii or HL-60 human-derived macrophages (Zusman et al., 2002; Abu-Zant et al., 2006). SpoT was proposed to be essential to L. pneumophila viability, since attempts to disrupt the locus in both wild-type (WT) and relA mutant backgrounds were unsuccessful (Zusman et al., 2002).
Here we undertook a comprehensive analysis of the contributions of RelA and SpoT to the L. pneumophila life cycle. We report that distinct metabolic cues trigger SpoT and RelA activity and that, during the L. pneumophila life cycle in host macrophages, SpoT is critical not only for replicative cells to differentiate to the transmissive state, but also for transmissive cells to re-enter the replicative phase.
In the absence of relA, spoT is not essential in L. pneumophila
To rigorously test the hypothesis that the stringent response controls L. pneumophila virulence expression, we constructed and analysed relA and relA spoT mutant strains (Table 1). In the stationary phase, L. pneumophila requires RelA to produce pyomelanin, a secreted pigmented molecule important for iron metabolism (Zusman et al., 2002; Chatfield and Cianciotto, 2007). Therefore, to verify that relA was disrupted, we assayed the extracellular pigment of stationary-phase cultures. As expected, our relA insertion mutant was completely defective for pyomelanin production (data not shown).
pMMB206-Δmob with 2.7 kb SalI fragment from pGEM-relAi ligated at the SalI site in the MCS, colinear with the PtaclacUV5 promoter, lacIq, inducible RelA expression, CamR
pMMB206-Δmob with 2.5 kb SalI/HindIII fragment from pGEM-spoTi ligated between the SalI/HindIII sites in the MCS, colinear with the PtaclacUV5 promoter, lacIq, inducible SpoT expression, CamR
pMMB206-Δmob with 2.5 kb SalI/HindIII fragment from pGEM-spoT-A413E ligated between the SalI/HindIII sites in the MCS, colinear with the PtaclacUV5 promoter, lacIq, inducible spoT-A413E expression, CamR
150 bp flaA promoter fragment fused to GFPmut3 in pKB5 with Ptac and lacIq removed, td(Δ)I
As observed for other Gram-negative bacteria, we and others were unable to generate spoT null mutant strains of L. pneumophila, and attempts to construct a conditional spoT mutant were also unsuccessful (Zusman et al., 2002). Therefore, to analyse the contribution of SpoT to the L. pneumophila life cycle, we constructed a relA spoT double mutant in which 136 amino acids near the N-terminus, including the consensus His–Asp (HD) motif required for ppGpp hydrolysis (Potrykus and Cashel, 2008) and a significant portion of the putative ppGpp synthetase domain (Gentry and Cashel, 1996), were replaced by a kanamycin resistance cassette. As the cultures entered late PE phase, the relA spoT double mutant cells adopted a hyperfilamentous cellular morphology (data not shown), a phenotype reminiscent of that described for relA spoT mutants of E. coli (Xiao et al., 1991; Traxler et al., 2008). Other than hyperfilamentation in late PE phase, in rich medium the growth kinetics of E-phase relA spoT mutant cells were identical to WT bacteria, as judged by optical density and colony-forming units (cfu) (Fig. 1C). Therefore, when relA is absent, spoT is not essential for L. pneumophila to replicate in rich bacteriological medium. Why previous attempts to isolate a relA spoT double mutant were unsuccessful is difficult to assess since the allele construction was not described (Zusman et al., 2002). However, spoT mutations that eliminate hydrolase but not synthetase activity would be lethal.
RelA and SpoT contribute to ppGpp accumulation, motility and survival in stationary phase
To assess in greater detail the contributions of RelA and SpoT to L. pneumophila biology, we tested the ability of relA and relA spoT mutants to accumulate ppGpp in stationary phase when cultured in rich broth containing yeast extract. After radiolabelling the L. pneumophila guanosine nucleotide pools for 6 h (two doubling times), ppGpp levels were analysed by thin-layer chromatography (TLC). WT L. pneumophila cultured into PE phase accumulated a pool of ppGpp, much of which was dependent on relA (Fig. 1A). The relA mutants also accumulated a small but appreciable pool of ppGpp that was not observed in the relA spoT double mutant (Fig. 1A). Thus, when L. pneumophila are cultured in rich broth to the PE phase, RelA synthetase activity accounts for the majority of ppGpp, whereas SpoT synthetase activity contributes modestly. Since L. pneumophila obtains carbon and energy from amino acids, not sugars (Tesh and Miller, 1981; Tesh et al., 1983), the predominant role for RelA in these yeast extract broth cultures is not surprising.
To test the biological consequences of the small amount of spoT-dependent ppGpp that accumulates in PE phase, we assayed flaA expression and motility, two hallmarks of stationary-phase L. pneumophila. The relA and relA spoT mutants were transformed with a plasmid containing a flaA:gfp transcriptional fusion that enables promoter activity of flaA, the primary flagellar subunit of L. pneumophila, to be analysed by fluorometry (Hammer and Swanson, 1999). As expected, the flaA promoter was inactive during E phase in WT, relA and relA spoT cultures (Hammer and Swanson, 1999). Upon entry into PE phase, both WT and relA mutant L. pneumophila activated the flaA promoter; however, in relA mutant cultures the flaA promoter activity was decreased relative to WT, consistent with a previous report (Zusman et al., 2002). In contrast, relA spoT mutant cultures failed to activate flaA expression upon entry into PE phase, indicating that residual flaA promoter activity observed in relA mutants is dependent on SpoT (Fig. 1B). Consistent with their flaA expression, the relA mutants also exhibited a 30–50% decrease in the fraction of motile bacteria per field relative to WT cultures, while relA spoT mutants failed to express motility at all PE-phase time points analysed (data not shown). Therefore, both RelA and SpoT contribute to flaA expression by PE L. pneumophila, and the small spoT-dependent ppGpp pool that accumulates in PE phase (Fig. 1A) is nevertheless sufficient to activate partial flaA expression and motility by L. pneumophila.
To investigate the importance of the stringent response to survival in the PE phase, we assessed the culture density and viability of WT, relA and relA spoT cultures over time. Both parameters were similar for all strains between 2 and 30 h. However, by 44 h, or ∼20 h after entry into PE phase, relA spoT cells appeared hyperfilamentous relative to WT and relA, which had each adopted the more coccoid-like morphology that is characteristic of PE-phase L. pneumophila (Molofsky and Swanson, 2004). Concomitant with filamentation, by 44 h relA spoT cultures began to exhibit a loss in cfu relative to WT and relA (Fig. 1C). This decrease in viability of relA spoT cells was even more pronounced after 66 h. Therefore, as reported for other bacteria, including C. jejuni and Helicobacter pylori (Gaynor et al., 2005; Mouery et al., 2006), L. pneumophila requires the stringent response for differentiation and survival in stationary phase.
RelA-dependent ppGpp synthesis is required for flaA expression following amino acid starvation
To verify that, like E. coli, L. pneumophila relies on RelA to initiate the stringent response following amino acid starvation, we subjected WT and mutant cultures to treatment with serine hydroxamate. An analogue of the amino acid serine, serine hydroxamate, inhibits the attachment of serine to transfer RNA, and it has been used extensively to characterize in E. coli the mechanism by which RelA senses accumulation of uncharged tRNAs at the ribosome (Tosa and Pizer, 1971; Magnusson et al., 2005). Unlike mock-treated cultures, WT cells subjected to serine hydroxamate treatment accumulated an appreciable pool of ppGpp (Fig. 2A). In contrast, neither relA nor relA spoT mutant L. pneumophila appeared to respond to serine hydroxamate treatment.
To determine the physiological relevance of the ppGpp pool detected, we asked if the relA-dependent ppGpp that accumulated after serine hydroxamate treatment was sufficient to activate flaA expression. Indeed, the pattern of flaA expression in serine hydroxamate-treated cultures (Fig. 2B, top) mimicked that observed for ppGpp production (Fig. 2A). Whereas WT cultures activated flaA as early as 3 h post treatment, neither relA nor relA spoT treated with serine hydroxamate activated flaA throughout the time-course analysed (Fig. 2B, top). Mock-treated cultures activated flaA with kinetics similar to that of control PE-phase cultures (Fig. 1B). Serine hydroxamate treatment did partially inhibit growth of all strains relative to mock-treated control samples (Fig. 2B, bottom), an effect that occurred independently of appreciable ppGpp accumulation (Fig. 2A). Together these observations verify that, in response to amino acid starvation, RelA synthesizes adequate ppGpp to activate flaA in L. pneumophila.
SpoT-dependent ppGpp accumulation activates flaA when FAB is inhibited
In E. coli, RelA responds to amino acid starvation, whereas SpoT responds to carbon, phosphate and iron starvation and FAB inhibition (Magnusson et al., 2005). Therefore, we tested whether L. pneumophila RelA and SpoT also respond to distinct metabolic cues. For this purpose we used the antibiotic cerulenin, a specific inhibitor of FAB in bacteria that acts on the 3-ketoacyl-[acyl carrier-protein (ACP)] synthases I and II, FabB and FabF (Seyfzadeh et al., 1993). Appreciable amounts of ppGpp were detected in WT and relA cultures treated with cerulenin, whereas relA spoT mutant L. pneumophila failed to accumulate detectable ppGpp under these conditions (Fig. 2C). The response was specific to the antibiotic, since cultures receiving the DMSO vehicle did not exhibit ppGpp.
To investigate the phenotypic consequences of the modest SpoT-dependent ppGpp pool, we assayed the ability of WT and mutant cultures to activate flaA following cerulenin treatment. As observed after the serine hydroxamate treatment, the pattern of flaA expression reflected the size of the ppGpp pool in each strain (Fig. 2D, top). Cerulenin-treated WT and relA mutant L. pneumophila activated flaA, but relA spoT cells did not (Fig. 2D, top). As observed for cultures treated with serine hydroxamate, cerulenin treatment also inhibited growth independently of ppGpp accumulation in the strain (Fig. 2D, bottom). Control cultures treated with DMSO were not inhibited for growth and activated fluorescence only upon entry into PE phase. Thus, L. pneumophila is equipped to differentiate in response to amino acid starvation via RelA, and inhibition of FAB using SpoT.
Amino acid substitution in SpoT confers enzyme activity that is blind to fatty acid signals
In addition to their response to FAB inhibition (Fig. 2), L. pneumophila that are exposed to excess short-chain fatty acids (SCFA) also activate transmissive phenotypes in a SpoT-dependent manner (R.L. Edwards, unpublished). Like cerulenin treatment, addition of either 10 mM acetic or propionic acid to E-phase L. pneumophila triggers expression of several transmission traits, including lysosome evasion, flaA expression and motility, as well as cytotoxicity (R.L. Edwards, unpublished). In E. coli, SpoT monitors and responds to perturbations in fatty acid metabolism through a specific interaction with ACP (Battesti and Bouveret, 2006). Accordingly, we tested the hypothesis that a SpoT–ACP interaction is essential for L. pneumophila to activate transmissive phenotypes when FAB is perturbed by excess SCFA. To do so, we designed plasmids for the inducible expression of either WT spoT or spoT-A413E, an allele encoding an enzyme predicted to be defective for ACP interaction (Battesti and Bouveret, 2006). Within the ∼60-amino-acid TGS domain in which the critical alanine residue resides, L. pneumophila SpoT exhibits 67% identity and 86% similarity to the E. coli protein, and the seven amino acids C-terminal to the alanine residue are identical. In particular, we asked whether expression of plasmid-borne spoT or spoT-A413E was sufficient to restore flagellin expression by relA spoT bacteria exposed to 10 mM SCFA. Since similar origins of replication and antibiotic resistance markers for the relevant plasmids complicated use of the flaA reporter, we instead measured cytotoxicity, a phenotype dependent on flagellin as well as type IV secretion (Molofsky and Swanson, 2004).
Wild-type or mutant spoT expression was first induced for 1 h with 25 μM isopropyl-beta-d-thiogalactopyranoside (IPTG) before the cells were cultured for 3 h with 10 mM SCFA and 25 mM IPTG. When exposed to propionic acid, WT spoT restored relA spoT mutant cytotoxicity towards macrophages, but spoT-A413E did not (Fig. 3A). Likewise, only WT spoT rescued relA spoT cytotoxicity after acetic acid or cerulenin addition (data not shown). The plasmid-encoded spoT-A413E allele was functional: when the IPTG concentration was doubled to 50 μM, relA spoT mutants carrying the spoT-A413E plasmid were even more cytotoxic than those harbouring either WT spoT or the empty vector (Fig. 3B). Furthermore, the enhanced cytoxicity occurred independently of SCFA treatment. Thus, the spoT-A413E mutation appears to render the enzyme insensitive to perturbations in fatty acid metabolism and also increases the ppGpp pools, presumably by disrupting the balance between ppGpp synthesis and hydrolysis (Battesti and Bouveret, 2006).
The stringent response activates transmissive traits
Legionella pneumophila co-ordinates entry into stationary phase with expression of not only motility but also virulence-associated phenotypes like sodium sensitivity, cytotoxicity and lysosome avoidance (Byrne and Swanson, 1998). Accordingly, to assess the contributions of RelA and SpoT to virulence trait expression in PE L. pneumophila, we assayed WT and mutant bacteria for sodium sensitivity and lysosome evasion. As predicted, unlike WT, relA spoT mutants remained sodium resistant even when cultured into PE phase (Fig. 4A). PE-phase relA mutant L. pneumophila, which accumulate low amounts of ppGpp, activate flaA and become motile (Fig. 1), also become more sensitive to sodium. Likewise, whereas PE WT and relA resisted phagosome–lysosome fusion and remained intact following macrophage phagocytosis, E-phase WT and PE-phase relA spoT were degraded (Fig. 4B). Therefore, even the modest SpoT-dependent pool of ppGpp is sufficient to induce significant sodium sensitivity and to protect L. pneumophila from lysosomal degradation.
Induction of either ppGpp synthetase is sufficient to restore the transmissive phenotype of relA spoT mutants in broth
Since L. pneumophila requires ppGpp for flaA expression, motility, sodium sensitivity and lysosome avoidance (Figs 1 and 4) we predicted that, during PE-phase starvation, induction of either synthetase would equip replicative relA spoT mutants to enter the transmissive phase. Indeed, IPTG induction of plasmid-borne relA, spoT or spoT-A413E suppressed both the flagellar and the cell-remodelling defect of relA spoT mutant bacteria, as judged by the motility of coccoid bacteria. In contrast, relA spoT bacteria carrying only the vector remained amotile and elongated (data not shown). Furthermore, when either relA, spoT or spoT-A413E was induced, relA spoT L. pneumophila killed more macrophages than either relA spoT mutant bacteria carrying vector alone or E-phase WT (Fig. 5A). Induction of relA, spoT and spoT-A413E also increased the infectivity of relA spoT bacteria: when harbouring plasmids encoding either ppGpp synthetase and cultured with IPTG to the PE phase, significantly more relA spoT mutants were infectious than those carrying the empty vector (Fig. 5B). Thus, expression of either relA, spoT or spoT-A413E was sufficient to trigger replicative-phase relA spoT mutants to enter the transmissive phase. These results not only verify that the phenotypic defects of relA spoT mutants were not due to secondary site mutations or polar effects, but also indicate that, regardless of its source, ppGpp co-ordinates the replicative to transmissive phase transition in broth.
The stringent response is essential during macrophage infection
As observed previously, relA was dispensable for L. pneumophila growth and survival in macrophages (Fig. 6A; Zusman et al., 2002; Abu-Zant et al., 2006). In contrast, relA spoT mutants infected with a decreased efficiency relative to PE-phase WT, a phenotype similar to WT in E phase (Fig. 6A; 2 h time point). Those relA spoT and E-phase WT bacteria that survived the initial attack moved on to replicate efficiently between 2 and 24 h. However, after 24 h, E-phase WT cfu increased to values equivalent to PE-phase WT, but relA spoT mutants exhibited no further increase in cfu over the remainder of the period analysed (Fig. 6A).
Since 24 h corresponds to the period of flagellar expression and initiation of secondary infection (Byrne and Swanson, 1998), we postulated that relA spoT mutants were defective for either escape from an exhausted host or entry into a naïve one. To distinguish between these two hypotheses, we performed immunofluorescence microscopy. At 16 h, both WT and relA spoT bacteria occupied replication vacuoles (data not shown); however, between 18 and 24 h, a unique class of infected macrophages emerged in relA spoT cultures. Intact bacteria were no longer observed; instead, degraded L. pneumophila particles were dispersed throughout the macrophage (Fig. 6B). In addition, there was scant evidence of secondary infection by relA spoT mutants at time points beyond 18 h (data not shown). Thus, the majority of mutant bacteria were apparently destroyed following the replicative growth period, prior to escape. In the rare macrophages that contained one or two bacteria at time points beyond 18 h, the relA spoT mutants were also degraded, likely due to their inability to avoid phagosome–lysosome fusion (Fig. 6B). At 18 h, the percentage of infected macrophages that harboured either single or multiple degraded relA spoT mutant bacteria was nearly 80%, approximately three times greater than the ∼25% of infected macrophages containing degraded WT bacteria (Fig. 6C). Therefore, the stringent response is required for L. pneumophila transmission between macrophages, and SpoT is a critical component of its developmental switch.
SpoT is essential for transmission between macrophages
To investigate further the contributions of RelA and SpoT to differentiation of replicative bacteria to the transmissive state in macrophages, we tested whether induction of either relA or spoT was sufficient to rescue the intracellular relA spoT mutants. The relA spoT mutant L. pneumophila were cultured in the absence of IPTG to early PE phase and then added to macrophages. After ∼16 h, IPTG was added to the cultures and maintained for the remainder of the infection. Like relA spoT bacteria harbouring empty vector, double mutants in which relA had been induced exhibited an ∼20-fold increase in cfu between 2 and 24 h, but no appreciable increase beyond 24 h (Fig. 7). Likewise, constitutive expression of relA by relA spoT mutants in rich broth was also incompatible with replication in broth, as judged by measuring both the optical density and cfu of the cultures (Fig. S1C). In contrast, when spoT was induced, relA spoT mutants replicated similarly to WT in macrophages (Fig. 7) and in broth (Fig. S1). In mouse macrophages, the cfu increase observed after 24 h reflects secondary and tertiary infections; therefore, SpoT activity is critical for L. pneumophila transmission between host cells.
To re-enter the replicative phase in macrophages, L. pneumophila requires SpoT
To investigate whether SpoT synthetase and/or hydrolase activity is critical for the life cycle of L. pneumophila in macrophage cultures, we tested whether induction of relA, spoT or spoT-A413E equipped transmissive bacteria to differentiate into the replicative phase in both macrophages and on solid medium. Broth-grown cultures of relA spoT bacteria carrying inducible relA, spoT or spoT-A413E were cultured with IPTG into PE phase to activate transmission traits. Next, transmissive bacteria were either added to macrophages in the presence of IPTG (Fig. 8A), or plated onto rich medium with or without IPTG (Fig. 8B). During macrophage infection with the relA spoT strains, induction of WT spoT resulted in a pattern of intracellular growth identical to that of PE-phase WT bacteria (Fig. 8A). In contrast, induction of neither relA nor spoT-A413E equipped relA spoT L. pneumophila to replicate intracellularly (Fig. 8A). Similar patterns were observed when transmissive bacteria were plated onto bacteriological medium. All the strains formed colonies when transferred to medium without IPTG (Fig. 8B, left). On medium containing IPTG, mutant relA spoT L. pneumophila that either expressed WT spoT or carried the empty vector readily formed colonies (Fig. 8B, right). In contrast, consistent with their intracellular growth defect, relA spoT mutants harbouring either inducible relA or spoT-A413E failed to replicate when IPTG was maintained in the medium (Fig. 8B, right). These genetic data suggest that the enzyme encoded by spoT-A413E favours ppGpp accumulation, more closely resembling RelA. Together, these results indicate that SpoT hydrolase activity is required for transmissive cells to re-enter the replicative phase either during macrophage infection or on bacteriological medium.
To endure starvation and other stresses, many bacteria rely on the stringent response pathway to alter their transcriptional profiles and enhance their fitness. When nutrients become scarce in broth, L. pneumophila generates ppGpp to co-ordinate its transition from a replicative to a transmissive form (Hammer and Swanson, 1999; Zusman et al., 2002). Although L. pneumophila utilizes RelA to produce ppGpp upon entry into the PE phase in yeast extract broth, this enzyme is dispensable for replication in either amoeba or human macrophages (Zusman et al., 2002; Abu-Zant et al., 2006). Like several other microbes, L. pneumophila also encodes the bifunctional and essential enzyme SpoT, which can either synthesize or hydrolyse ppGpp depending on local cues. Analysis of broth cultures indicates that the two ppGpp synthetases confer versatility to L. pneumophila, enabling conversion to the transmissive state when either protein or fatty acid synthesis is compromised (Fig. 2). However, in macrophages only SpoT is required, and its role is bipartite (Figs 6A and 8A). After establishing its protective vacuole, intracellular L. pneumophila relies on SpoT to hydrolyse ppGpp to reduce the alarmone pool and switch from the transmissive to the replicative form. When conditions deteriorate, the intracellular progeny elicit SpoT synthetase activity to co-ordinate their conversion to the transmissive state and escape from the exhausted host (Fig. 9).
In broth, L. pneumophila requires ppGpp from either RelA or SpoT to initiate the replicative to transmissive transition (Fig. 9). When L. pneumophila is cultured in broth, a large proportion of its ppGpp pool is generated by RelA (Fig. 1A). The more minor contribution by SpoT can be observed as the residual ppGpp detected in relA mutants, since only two ppGpp synthetases are apparent in the L. pneumophila genome, and no alarmone is detectable in relA spoT double mutant cells (Fig. 1A). Like E. coli SpoT, the L. pneumophila enzyme may be a weak synthetase (Seyfzadeh et al., 1993). Alternatively, the low levels of SpoT-dependent ppGpp observed may be a consequence of experimental conditions: its fastidious character precludes use of a defined medium where competing inorganic phosphate levels can be minimized. Here L. pneumophila was cultured in a rich medium that contains amino acids as the sole energy source, which not only reduces the efficiency of the radiolabelling of the phosphate pool, but also favours RelA-dependent ppGpp production. Nevertheless, the weak ppGpp signal generated by SpoT in this medium is sufficient to induce with timing similar to WT L. pneumophila not only the flaA promoter, but also motility, sodium sensitivity and evasion of lysosomes (relA mutant; Figs 1B and 4, and data not shown).
In addition to triggering its panel of transmission traits, ppGpp from either synthetase is also critical for L. pneumophila to survive the PE phase in broth cultures (Fig. 1C). In E. coli, ppGpp-dependent control of fatty acid and phospholipid metabolism allows the bacterium to selectively modify its membranes during stress. For example, ppGpp inhibits fatty acid and phospholipid biosynthesis through repression of plsB (sn-glycerol-3-phosphate acyltransferase), resulting in accumulation of long-chain acyl-ACPs and cell shortening (Heath et al., 1994). Indeed, the E. coli stringent response both activates and represses several genes in the fatty acid and phospholipid biosynthetic pathways, as judged by transcriptional profile analysis (Durfee et al., 2008). Accordingly, we predict that the stringent response pathway induces synthesis of specific membrane constituents that are critical for L. pneumophila cells to survive periods of stress, including that encountered during entry into and exit from host phagocytes. Taken together, our phenotypic and biochemical analysis of WT, relA and relA spoT bacteria establishes that synthesis of ppGpp from either enzyme is sufficient to co-ordinate L. pneumophila differentiation into the transmissive form and to promote survival in detrimental environments.
To alternate between transmissive and replicative forms in macrophages, L. pneumophila uses the SpoT enzyme to tightly regulate ppGpp synthesis and degradation. As reported previously (Zusman et al., 2002; Abu-Zant et al., 2006), the RelA enzyme itself is dispensable for L. pneumophila to survive phagocytosis by macrophages and to replicate intracellularly (Fig. 6A). On the other hand, the alarmone ppGpp is critical to induce the virulence factors that equip L. pneumophila to establish its protected intracellular niche, since mutants that lack both relA and spoT survive poorly within macrophages (Fig. 6). For the replicative to transmissive transition in broth, the source of ppGpp is inconsequential, since expression of either RelA or SpoT largely bypasses the transmissive trait defects of relA spoT double mutant bacteria (Fig. 5A and B). Once replication ensues in the macrophage, ppGpp is no longer required, since the few relA spoT mutant bacteria that do survive the initial infection replicate efficiently during the subsequent 24 h primary infection period (Fig. 6A). When vacuolar conditions deteriorate and L. pneumophilatransit to a new host, SpoT is required, since induction of SpoT but not RelA bypasses the growth defect of relA spoT mutant bacteria beyond the 24 h time point (Fig. 7). During the transmission period, the synthetase activity of SpoT is sufficient to promote escape from and entry into a new host (relA mutant bacteria; Fig. 6A), whereas its hydrolase is absolutely required for newly engulfed bacteria to initiate replication (Fig. 8A). Thus, SpoT governs differentiation of L. pneumophila at two critical nodes of its life cycle in macrophage cultures.
Legionella pneumophila has evolved two ppGpp synthetases capable of responding to distinct metabolic inputs during growth in vitro (Fig. 2); however, during intercellular transmission only SpoT is essential (Figs 6A and 7). Therefore, the stress signal that promotes replicative bacteria to convert to the transmissive form in macrophages must be sensed by SpoT. In E. coli, RelA responds to amino acid limitation, whereas SpoT activates the stringent response pathway during all other types of stresses and starvation (Potrykus and Cashel, 2008). Amino acid depletion also stimulates L. pneumophila to produce ppGpp and induce differentiation by a RelA-dependent pathway (Fig. 2A and B). On the other hand, to respond to perturbations in FAB, L. pneumophila relies on SpoT (Fig. 2C and D). Since SpoT alone is sufficient to induce transmission in macrophages (Figs 6A and 7), we postulate that the ability of SpoT to sense variations in fatty acid metabolism equips L. pneumophila to respond to fluctuations in the lipid supply during its intracellular life cycle. Whether SpoT also equips L. pneumophila to respond to other metabolic stimuli that are critical within host cells remains to be explored experimentally.
Our genetic data suggest that, as in E. coli, the L. pneumophila response to perturbations in FAB requires SpoT interaction with ACP; by some mechanism, this interaction regulates SpoT enzyme activity (Battesti and Bouveret, 2006). In particular, Battesti and Bouveret identified a single amino acid substitution in SpoT that abrogates a physical interaction with ACP and renders the mutant bacteria blind to fatty acid starvation (Battesti and Bouveret, 2006). Likewise, the analogous spoT-A413E mutant L. pneumophila fail to trigger cytotoxicity in response to excess SCFA (Fig. 3A). In addition, interaction between SpoT and ACP affects the ability of SpoT to reduce the ppGpp pool, as judged by three phenotypic assays. L. pneumophila treated to express strongly the spoT-A413E allele apparently contain excessive ppGpp, as the cells are more cytotoxic than bacteria expressing WT spoT (Fig. 3B), and they survive yet fail to replicate when cultured either in macrophages or on rich medium (Fig. 8A and B). Based on the capacity of the analogous E. coli spoT-A404E mutant to adapt to histidine starvation, Battesti and Bouveret (2006) also deduced that excess ppGpp accumulates when the SpoT–ACP interaction is disrupted. We note two caveats to our experimental approaches. The contribution of a physical interaction between SpoT and ACP to the L. pneumophila fatty acid response is based on a precedent established in E. coli and our analogous genetic analysis, which remains to be verified by direct biochemical assays. Second, the simple interpretation that increased concentrations of IPTG used to induce either the spoT or the spoT-A413E allele increases the amount of the respective proteins (Fig. 3) has not been verified. Our attempts to monitor SpoT protein levels by Western analysis using a previously described polyclonal antibody specific to the E. coli enzyme (Gentry and Cashel, 1996) were not successful.
Given our genetic data and the published genetics and biochemistry (Battesti and Bouveret, 2006), several working models remain in play. When FAB is perturbed, an unidentified signalling ligand either accumulates or is depleted, thereby altering ACP in a manner that is sensed by SpoT. Consequently, ACP could either repress SpoT-dependent ppGpp hydrolysis, or activate SpoT-dependent ppGpp synthesis. Biochemical studies can distinguish between these possibilities. In any case, when FAB is perturbed, SpoT-dependent ppGpp accumulates and replicative L. pneumophila transition to the transmissive form.
In addition to monitoring FAB, SpoT activity is also critical for transmissive bacteria to re-enter the replicative state when nutrients are abundant, likely due to the need to hydrolyse the alarmone (Fig. 8A and B). The intracellular phenotype of L. pneumophila that generate excess ppGpp (i.e. relA spoT mutant bacteria that constitutively express either relA or the mutant allele spoT-A413E) is similar to that of phtA mutant L. pneumophila, which lack the high-affinity transporter for threonine (Sauer et al., 2005). In each case, the mutant bacteria persist as single, intact, rods within their vacuoles, while WT microbes replicate robustly (Sauer et al., 2005). Without the capacity to acquire threonine, as essential amino acid, L. pneumophila may become locked in a state of ppGpp synthesis. Thus, the local nutrient supply dictates when transmissive bacteria activate replicative functions. To resume multiplying, L. pneumophila must activate the SpoT hydrolase to reduce its ppGpp pool (Fig. 9B).
The intracellular pathogen M. tuberculosis encodes a single-bifunctional Rel–Spo homologue (RSH), Rel, required for long-term survival in vitro (Primm et al., 2000) and long-term persistence in mice (Dahl et al., 2003; Karakousis et al., 2004). The ability of this pathogen to differentiate to a persister state may contribute to the failure of front-line antimicrobial agents (Warner and Mizrahi, 2006). Thus, understanding the biochemistry of the stringent response enzymes provides an avenue to analyse the life cycle of a variety of intracellular pathogens and, more importantly, to develop new strategies and therapeutics to eradicate them.
Encoding both RelA and SpoT confers versatility to L. pneumophila and likely enables the bacterium to survive a variety of assaults encountered in the extracellular environment. Although the precise stimuli that cue L. pneumophila differentiation in vivo remain to be determined, our evidence points to a SpoT-dependent signal. Furthermore, the bifunctional SpoT enzyme plays two critical and distinct roles during the intracellular life cycle. By acting as a receiver of metabolic cues, SpoT co-ordinates a rapid response when either favourable or unfavourable environmental circumstances are encountered. Thus the stringent response pathway is fundamental to the virulence of L. pneumophila, an opportunistic pathogen that not only replicates within but also transits between macrophages.
Bacterial strains, culture conditions and reagents
Legionella pneumophila strain Lp02 (thyA hsdR rpsL; MB110), a virulent thymine auxotroph derived from Philadelphia 1 (Berger and Isberg, 1993), was the parental strain for all the strains analysed. L. pneumophila was cultured at 37°C in 5 ml aliquots of N-(2-acetamido)-2-aminoethanesulphonic acid (ACES; Sigma)-buffered yeast extract (AYE) broth with agitation or on ACES-buffered charcoal yeast extract (CYE), both supplemented with 100 μg ml−1 thymidine (AYET, CYET) when necessary. Bacteria obtained from colonies < 5 days old were cultured in broth overnight, then subcultured into fresh AYET prior to experimentation. For all experiments, E cultures were defined as having an optical density at 600 nm (OD600) of 0.3–2.0; PE samples were obtained from cultures having an OD600 of 3.0–4.5 during the period 20–30 h after subculture, when viability and cell shape of the strains were similar (Fig. 1C). Where indicated, ampicillin (amp; Fisher) was added to a final concentration of 100 μg ml−1, kanamycin (kan; Roche) to 10 μg ml−1, gentamicin (gent; Fisher) to 10 μg ml−1, chloramphenicol (cam; Roche) to 5 μg ml−1, cerulenin (cer; Sigma) to 0.5 μg ml−1, serine hydroxamate (Sigma) to 1 mM, propionic acid to 10 mM, acetic acid to 10 mM and IPTG to the concentrations specified. To determine cfu, serial dilutions of L. pneumophila were plated on CYET and incubated at 37°C for 4–5 days.
relA and relA spoT mutant strain construction
To construct a relA insertion mutant, we generated pGEM-relA by amplifying the relA locus (lpg1457) plus flanking sequence from Lp02 genomic DNA using the primers relA1 and relA2. The kanamycin resistance gene cassette from pUC4K and the gentamicin cassette from PUCGent were obtained from their respective plasmids by EcoRI digestion, and pGEM-relA was digested with SnaBI. The fragments containing the resistance cassettes were blunted with Klenow, gel extracted and ligated into the blunted pGEM-relA plasmid, creating pGEM-relA::kan and pGEM-relA::gent. After verification by PCR and restriction enzyme digest, the insertion alleles were used to transform Lp02 by natural competence, and the resulting transformants were selected using the appropriate antibiotic (Stone and Abu Kwaik, 1999). Recombination of the desired relA insertion alleles onto the Lp02 chromosome was confirmed by PCR, and the resulting relA::kan and relA::gent mutant L. pneumophila were designated MB696 and MB695, respectively.
To construct the relA spoT double mutant, we generated pGEM-spoT by amplifying the spoT locus (lpg2009) plus flanking sequence from Lp02 genomic DNA using the primers spoT1 and spoT2. pGEM-spoT was digested with HindIII, which cuts at sites ∼98 and ∼588 bp 3′ to the translational start, releasing an ∼490 bp fragment encoding the consensus HD motif required for hydrolase activity, as well as a significant portion of the predicted synthetase coding regions at the N-terminus of SpoT. The deletion leaves intact 32 amino acids at the extreme N-terminus, as well as amino acids 195–715 at the C-terminus. By analogy to deletion analysis conducted with the E. coli spoT gene (Gentry and Cashel, 1996), our deletion should have eliminated both enzymatic activities of SpoT. The linear fragment containing the spoT deletion was blunted, and a kanamycin cassette was ligated to the blunted fragment, generating pGEM-spoT::kan. After verification by PCR and restriction enzyme digest, the deletion/insertion alleles were used to transform relA::gent mutant L. pneumophila MB695 by natural competence, and transformants were selected using the appropriate antibiotic (Stone and Abu Kwaik, 1999). Recombination of the desired spoT deletion/insertion allele into the relA::gent mutant chromosome was confirmed by PCR. The resulting relA::gent spoT::kan double mutant L. pneumophila was designated MB697. MB355 contains the pflaAgfp plasmid that encodes thymidylate synthetase as a selectable marker and a transcriptional fusion of the flaA promoter to gfp (Hammer and Swanson, 1999; Hammer et al., 2002). MB696 and MB697 were transformed with the pflaAgfp reporter plasmid to create MB684 and MB685 respectively.
Inducible relA and spoT expression
To generate strains in which expression of either relA or spoT could be induced, we cloned promoterless fragments of relA and spoT into pMMB206-Δmob, a plasmid containing a PtaclacUV5 IPTG-inducible promoter and a chloramphenicol resistance cassette. To construct pGEM-relAi, the relA locus was amplified from Lp02 genomic DNA using the primers relAi1 and relAi2, each containing a SalI restriction site. The fragment was excised from pGEM-relAi using SalI and ligated into pMMB206-Δmob at the SalI site within the MCS immediately 3′ of the PtaclacUV5 promoter, generating prelAi. Insertion and orientation were confirmed by both PCR and restriction enzyme digest. prelAi was used to transform MB697, and transformants were selected on chloramphenicol, creating MB687 for inducible relA expression. To construct pGEM-spoTi, the spoT locus was amplified from Lp02 genomic DNA using the primers spoTi1, which contains a SalI restriction site, and spoTi2, which contains a HindIII site. pGEM-spoTi and pMMB206-Δmob were each digested with SalI and HindIII and then ligated, generating pspoTi. Insertion and orientation of spoT in the MCS of pMMB206-Δmob were confirmed by PCR and restriction enzyme digest. pspoTi was used to transform MB697, generating MB688 for inducible spoT expression.
Inducible spoT-A413E expression
pGEM-spoT was used as a template for spoT-A413E mutant allele construction. To generate site-directed mutations in spoT, Stratagene's QuickChange® II XL Site-Directed Mutagenesis Kit was used. To synthesize spoT-A413E, pGEM-spoT plasmid DNA was amplified with PfuUltra HF DNA polymerase (Stratagene) and the complementary primers spoTA413E1 and spoTA413E2. Parental DNA was digested with DpnI, and DH5α transformants were selected on ampicillin. To verify the GCC to GAA codon change, candidate plasmids were sequenced and then designated pGEM-spoT-A413E. pGEM-spoT-A413E was digested with SalI and HindIII, and the locus cloned into pMMB206-Δmob, as described for the WT spoT allele, generating pspoT-A413E. MB697 was transformed with pspoT-A413E, and transformants were selected on chloramphenicol, generating MB689 for inducible spoT-A413E expression.
Detection of ppGpp
Accumulation of the ppGpp in the PE phase and in response to amino acid starvation and FAB inhibition was detected by TLC as described (Cashel, 1969; 1994; Hammer and Swanson, 1999). Briefly, to detect PE-phase ppGpp production in WT and mutant L. pneumophila, 100 μCi ml−1 carrier-free 32P phosphoric acid (ICN Pharmaceuticals) was added in late E phase, and bacteria were cultured at 37°C on a roller drum to the PE phase, or approximately 6 h. To detect ppGpp accumulation following amino acid starvation and FAB inhibition, E-phase cultures were diluted to an OD600 = 0.25 and labelled with approximately 100 μCi ml−1 carrier-free 32P phosphoric acid for 6 h, or two generation times, at 37°C on a roller drum. After incorporation of the radioactive label, cultures were supplemented with a carrier control, 1 mM serine hydroxamate or 0.5 μg ml−1 cerulenin. Cultures were incubated for an additional 1.5 h at 37°C. To extract the nucleotides, 50 μl aliquots were removed from each culture, added to 13 M formic acid, and then incubated on ice for 15 min. Samples were subjected to two freeze–thaw cycles and stored at −80°C until chromatographed. Formic acid extracts (35 μl in the case of the treated cultures and 25 μl in the case of the PE samples) were applied to a PEI-cellulose TLC plates (20 × 20; Sorbent) and developed with 1.5 M KH2PO4, pH 3.4 as described (Cashel, 1969; 1994; Hammer and Swanson, 1999). TLC plates were exposed to Kodak BioMax MR Film (18 × 24 cm) for 72 h before development. To monitor growth in these experiments, OD600 was determined for non-radioactive cultures grown under identical conditions.
To monitor expression of the flagellin promoter in WT and mutant cultures, L. pneumophila containing the reporter plasmid pflaAgfp were cultured in AYE media. Overnight cultures in mid-E-phase OD600 = 1.0–1.75 were back diluted to early E-phase OD600 = 0.50–0.85 (T = 0). At the times indicated, the cell density of each culture was measured as OD600. To analyse similar bacterial concentrations, aliquots were collected by centrifugation, and the cells were normalized to OD600 = 0.01 in 1× PBS. An aliquot of each sample (200 μl) was transferred to black 96-well plates (Costar), and the relative fluorescence intensity was measured using a Synergy™ HT microplate reader and 485 nm excitation, 530 nm emission, sensitivity of 50. A similar protocol was used to monitor flaA promoter activity in cultures supplemented with 1 mM serine hydroxamate and 0.5 μg ml−1 cerulenin. Overnight cultures in mid-E-phase OD600 of 1.0–1.75 were back diluted to early E-phase OD600 of 0.50–0.85 (T = 0), at which time WT and mutant bacteria were treated with either serine hydroxamate or cerulenin. L. pneumophila cultures supplemented with water or DMSO served as negative and vehicle controls respectively. Optical density and relative fluorescence were measured as described above.
To calculate the percentage of L. pneumophila that were sensitive to sodium, PE bacteria or E cultures were plated onto CYET and CYET containing 100 mM NaCl. After a 6-day incubation at 37°C, cfu were enumerated, and the percentage of sodium-sensitive microbes calculated as described (Byrne and Swanson, 1998).
Macrophages were isolated from femurs of female A/J mice (Jackson Laboratory) and cultured in RPMI-1640 containing 10% heat-inactivated fetal bovine serum (FBS) (RPMI/FBS; Gibco BRL) as described previously (Swanson and Isberg, 1995). Following a 7-day incubation in L-cell supernatant-conditioned media, macrophages were plated at either 5 × 104 per well for cytotoxicity or 2.5 × 105 per well for degradation assays, infectivity assays and intracellular growth curves.
The percentage of intracellular L. pneumophila that remain intact after a 2 h macrophage infection was quantified by fluorescence microscopy (Bachman and Swanson, 2001). Briefly, macrophages were plated at 2.5 × 105 onto coverslips in 24-well plates. Then, E- or PE-phase microbes were added to macrophage monolayers at a multiplicity of infection (moi) ∼1. The cells were centrifuged at 400 g for 10 min at 4°C and then incubated for 2 h at 37°C. Extracellular bacteria were removed by washing the monolayers three times with RPMI/FBS, and the macrophages were fixed, permeabilized and stained for L. pneumophila as described (Molofsky et al., 2005). Except for longer incubation times (18 and 24 h), identical procedures were used to image and score infected macrophages at later time points of the primary infection period (Fig. 6B and C).
Infectivity and intracellular growth
Infectivity is a gauge of the ability of L. pneumophila to bind, enter and survive inside macrophages during a 2 h incubation, as previously described (Byrne and Swanson, 1998; Molofsky and Swanson, 2003). In brief, macrophages were infected with strains at an moi of ∼1, with centrifugation at 400 g for 10 min. Cells were incubated for 2 h at 37°C, washed three times with fresh RPMI, and lysed in a 1:1 solution of RPMI and saponin, then intracellular cfu were enumerated. Results are expressed as the percentage of the inoculated cfu recovered from lysed macrophages at 2 h post infection. To gauge intracellular growth, the pooled macrophage supernatant and lysate were plated for cfu at various times post infection, as described elsewhere (Molofsky and Swanson, 2003). Results for intracellular growth are expressed as either total cfu recovered or relative cfu. In the latter, cfu recovered at 24, 48 and 72 h are divided by the number of cfu at 2 h. Culturing conditions for WT and mutant L. pneumophila varied depending on the experimental question addressed. For experiments assaying infectivity and intracellular survival of WT, relA and relA spoT bacteria lacking expression vectors, cultures were grown to E or PE phase prior to infection. To test the contribution of relA and spoT to the replicative to transmissive phase transition in broth (Fig. 5), overnight cultures in mid-E-phase OD600 of 1.0–1.75 were back diluted to early E-phase OD600 of 0.50–0.85. Cultures were incubated for approximately 3 h at 37°C to an OD600 of 1.0–1.75; then, 200 μM IPTG was added and maintained until PE phase. To assess the role of relA and spoT in the transmissive to replicative phase transition within host cells (Fig. 8A), bacteria were cultured in broth with 200 μM IPTG to PE phase, and 200 μM IPTG was maintained throughout macrophage infection. A similar protocol was used to test the transmissive to replicative transition in bacteriological medium by plating bacteria that had been induced with 200 μM IPTG and cultured into PE phase onto CYET with 200 μM IPTG (Fig. 8B).
To determine contact-dependent cytotoxicity of L. pneumophila following macrophage infection, bacteria were added to macrophage monolayers at the indicated moi. After centrifugation at 400 g for 10 min at 4°C (Molofsky et al., 2005), the cells were incubated for 1 h at 37°C. For quantification of macrophage viability, RPMI/FBS containing 10% alamarBlue™ (Trek Diagnostic Systems) was added to the monolayers for 6–12 h, and the reduction of the colorimetric dye was measured spectrophoretically as described (Molofsky et al., 2005). Protocols for culturing WT and mutant microbes varied depending on the experimental question. To determine the effect relA and spoT induction on cytoxicity of relA spoT mutant bacteria (Fig. 5A), microbes were cultured with 200 μM IPTG to PE phase in broth prior to the assay, as described in Infectivity and intracellular growth. To determine if expression of WT and mutant spoT was sufficient to rescue the cytoxicity defect of relA spoT mutant bacteria in which FAB had been perturbed (Fig. 3), overnight cultures in mid-E-phase OD600 of 1.0–1.75 were back diluted to early E-phase OD600 of 0.45–0.60 and treated with either 25 or 50 μM IPTG for 1 h in the case of the 25 μM induction and 2 h for the 50 μM induction. After incubation on a roller at 37°C, mutant cultures were split and either mock-treated or treated with propionic acid, acetic acid or cerulenin. Cultures were incubated at 37°C for an additional 3 h prior to assaying cytotoxicity.
We thank Mike Cashel for sharing reagents, technical advice and insightful discussion, and Fred Neidhardt for many discussions and support. This work was supported by the Molecular Mechanisms of Microbial Pathogenesis Training Program, the Cellular and Molecular Biology Training Program, the University of Michigan Rackham Predoctoral Fellowship and NIH Grant 2 R01 AI44212.