The Legionella pneumophila rpoS Gene Is Required for Growth within Acanthamoeba castellanii

Abstract

To investigate regulatory networks in Legionella pneumophila, the gene encoding the homolog of the Escherichia coli stress and stationary-phase sigma factor RpoS was identified by complementation of an E. coli rpoS mutation. An open reading frame that is approximately 60% identical to the E. coli rpoS gene was identified. Western blot analysis showed that the level of L. pneumophila RpoS increased in stationary phase. An insertion mutation was constructed in the rpoS gene on the chromosome of L. pneumophila, and the ability of this mutant strain to survive various stress conditions was assayed and compared with results for the wild-type strain. Both the mutant and wild-type strains were more resistant to stress when in stationary phase than when in the logarithmic phase of growth. This finding indicates that L. pneumophila RpoS is not required for a stationary-phase-dependent resistance to stress. Although the mutant strain was able to kill HL-60- and THP-1-derived macrophages, it could not replicate within a protozoan host, Acanthamoeba castellanii. These data suggest that L. pneumophila possesses a growth phase-dependent resistance to stress that is independent of RpoS control and that RpoS likely regulates genes that enable it to survive in the environment within protozoa. Our data indicate that the role of rpoS in L. pneumophila is very different from what has previously been reported for E. coli rpoS.

Legionella pneumophila is a gram-negative bacterium which normally exists in water, in soil, and within free-living unicellular protozoa, yet it has developed unique strategies that permit multiplication within the phagosomes of human macrophages. It is this ability which enables it to persist and cause the pneumonia known as Legionnaires’ disease (71). L. pneumophila binds cell surface complement receptors (62) and enters human mononuclear cells via a coiling phagocytosis mechanism (34). Intracellular survival requires that acidification of the Legionella-containing phagosome and fusion with lysosomes be prevented (33, 35, 36, 82). The bacteria sequentially recruit host cell smooth vesicles, mitochondria, and ribosomes, replicate within a specialized vacuole (32, 37, 79), and eventually lyse the cell. In contrast to other intracellular pathogens such as Leishmania, Mycobacterium, and Toxoplasma which use many of the same strategies for survival (73), L. pneumophila can be genetically manipulated and grown with relative ease both on bacteriological media and intracellularly within cell culture lines and protozoan hosts.

Intracellular pathogens encounter a variety of different environmental stresses upon entry into a eukaryotic cell. Studies of pathogenic organisms indicate that gene expression is coordinately regulated in response to environmental signals such as the absence of certain amino acids, pH, temperature, oxygen availability, and the concentration of ions like iron, calcium, and magnesium (25, 54). Sigma factors are one way that bacteria regulate the expression of specific sets of genes in response to environmental signals (84). The sigma factor ς^S or RpoS of Escherichia coli is known to specifically activate genes when the bacteria are entering stationary phase or encounter adverse conditions such as low nutrient availability, high osmolarity, reactive oxygen intermediates, or low pH. Cells in stationary phase exhibit increased osmotolerance, resistance to oxidative stresses such as H₂O₂, and survive starvation as a result of RpoS-dependent gene expression (19, 29, 48).

Induction of genes responsible for survival under adverse conditions is likely important for pathogenesis. Indeed, RpoS homologs have been identified in several pathogens, yet their roles vary among organisms (61, 74, 80). A well-studied example is Salmonella typhimurium. The oral dose of bacteria required to kill 50% of infected mice is 1,000-fold higher for an S. typhimurium rpoS null strain than for the wild-type strain (20, 42). Additionally, the mutant strain is less able to survive stress (20). Studies also show that RpoS likely regulates chromosomal as well as plasmid-encoded virulence genes in S. typhimurium (20, 42), including those required for acid tolerance (46). Yersinia enterocolitica rpoS is required for the expression of heat-stable enterotoxin (Yst) (39), but mutant strains lacking RpoS are wild type in the mouse model of virulence (6, 39). Vibrio cholerae rpoS mutants show reduced expression of hemagglutinin/protease and are stress sensitive but are wild type in the ability to colonize mice (85).

We sought to identify L. pneumophila homologs of global regulatory proteins which are known to be involved in regulation of virulence genes in other organisms in order to elucidate regulatory networks in L. pneumophila and to draw parallels with other pathogens. In particular, we wanted to determine if L. pneumophila encodes an RpoS-like protein and to determine if this protein is required for growth within eukaryotic cells. Here we describe the isolation of the L. pneumophila rpoS gene. We examined potential roles for RpoS in vivo by testing the ability of a strain containing a mutation in rpoS to survive stress conditions and replicate within eukaryotic cells. We report that the role of rpoS in L. pneumophila is distinctly different from what has described for E. coli rpoS.

(A preliminary report of this work has been presented elsewhere [26].)

MATERIALS AND METHODS

Bacterial strains and plasmids.

Bacterial strains and plasmids used in this study are listed in Table 1. Strain LM5005 was constructed by growing P1vir on strain RH90 and transducing the Tet^r marker into strain RO151 as described by Silhavy et al. (72).

TABLE 1.

Bacterial strains and plasmids used

Strain or plasmid	Genotype or features	Reference or source(s)
L. pneumophila
25D	Philadelphia-1 avirulent mutant	35
AM511	Philadelphia-1 Smr r− m+	51
JR32	AM511 salt-sensitive isolate	67
LELA14	pig::Tn903dIIlacZ	83
LELA2955	icmX2955::Tn903dIIlacZ	67
LM1376	JR32 rpoS4::Tn903dGent	This study
LM1381	JR32 rpoS5::Tn903dGent	This study
LM1386	JR32 rpoS13::Tn903dGent	This study
LM1389	LELA2955 rpoS4::Tn903dGent	This study
LM1392	LELA2955 rpoS5::Tn903dGent	This study
LM1395	LELA14 rpoS4::Tn903dGent	This study
LM1397	LELA14 rpoS13::Tn903dGent	This study
LM1580	LM1376 rpoS+	This study
LM1584	LM1381 rpoS+	This study
LM1588	LM1386 rpoS+	This study
E. coli
DH5α	F−endA1 hsdR17 (r− m+) supE44 thi-1 λ− recA1 gyrA96 (Nalr) relA1 Δ(argF-lacZYA)U169 deoR φ80dlacZ ΔM15	Lab collection
LE392Δlac	e14− (r− m+) hsdR514 supE44 supF58 Δ(lacIZY)6 galK2 galT22 metB1 trpR55 λ−	Lab collection
LM5005	RO151 rpoS379::Tn10	This study
LM5316	LW252/pLM507	This study
LM5408	Donor for transposition: LM5316/pLM598 (pool 1)	This study
LM5409	Donor for transposition: LM5316/pLM598 (pool 2)	This study
LW211	LE392Δlac with an integrated RP4 (ΔAmpr Tetr::Mu) Mob+ Kanr	L. Wiater and H. Shuman
LW252	LW211 Kanr::Tn7 (Kanr Trir Spcr Smr)	L. Wiater and H. Shuman
MC4100	F−araD139 Δ(argF-lacZYA)U169 rpsL150 relA1 flbB5301 deoC1 ptsF25 rbsR	72
MW67	DH5α containing a L. pneumophila Philadelphia-1 chromosomal DNA library consisting of partial EcoRI fragments (10–20 kb) in pMMB207	63
RH90	MC4100 rpoS379::Tn10	44
RO151	MC4100 csi-5::lacZ (osmY::lacZ)	81
Plasmids
pBluescript II KS(+)	oriR (f1) MCSaoriR (ColE1), Ampr	Stratagene
pBC SK(+)	oriR (f1) MCS oriR (ColE1) Camr	Stratagene
pBR322	oriR (ColE1) Ampr Tetr	New England Biolabs
pCR2.1	oriR (f1) MCS oriR (ColE1) Ampr Kanr	Invitrogen
pKD368	tnpA Tn903dIIlacZ (′lacZ Kanr) oriR (ColE1) Ampr	18
pLAW344	sacB MCS oriT (RK2) CamrloxP oriR (ColE1) AmprloxP	83
pLB41	HindIII fragment containing Gmr in pBR322	L. Babiss and D. Figurski
pLM507	pMMB207 containing two EcoRI fragments isolated from library MW67 (rpoS+)	This study
pLM546	2,502-bp PstI fragment from pLM507 cloned into pBluescript II KS(+)	This study
pLM549	3,201-bp PstI fragment from pLM507 cloned into pBluescript II KS(+)	This study
pLM579	HindIII fragment containing Gmr from pLB41 cloned into pYSF31	This study
pLM598	XbaI fragment containing Gmr from pLM579 cloned into pKD368 in place of the Kanr ′lacZ sequences	This study
pLM654	pLM507 rpoS4::Tn903dGent	This study
pLM655	pLM507 rpoS5::Tn903dGent	This study
pLM658	pLM507 rpoS13::Tn903dGent	This study
pLM670	pLAW344 containing rpoS4::Tn903dGent from pLM654	This study
pLM671	pLAW344 containing rpoS5::Tn903dGent from pLM655	This study
pLM673	pLAW344 containing rpoS13::Tn903dGent from pLM658	This study
pLM799	PCR fragment of rpoS cloned into pZErO-2.1	This study
pLM806	EcoRI-XbaI fragment containing rpoS from pLM799 cloned into pBC SK(+)	This study
pLM845	pLAW344 containing rpoS from pLM507	This study
pMMB207	oriR (RSF1010) MCS lacIq Camr	58
pYSF31	pACYC184 oriT (RP4) polylinker pBluescript II KS(+) in EcoRV site	S.-F. Yan and H. Shuman
pZErO-2.1	oriR (f1) MCS ccdB oriR (ColE1) Kanr	Invitrogen

^aMCS, multiple cloning site. 

Media and reagents.

Growth of L. pneumophila and E. coli, and chemicals and antibiotics used, are as described elsewhere (63). Tetracycline was used at 20 μg/ml for E. coli; gentamicin was used at 10 μg/ml for L. pneumophila. Bovine albumin (35% solution), isopropyl-β-d-thiogalactopyranoside (IPTG), and 30% (wt/vol) solution of hydrogen peroxide (H₂O₂) were purchased from Sigma Chemical Co. 5-Bromo-4-chloro-3-indolyl-β-d-galactoside (X-Gal) was purchased from Diagnostic Chemicals Limited.

DNA sequence analysis.

All nucleotide sequences were generated by the DNA Synthesis and Sequencing Facility of the Comprehensive Cancer Center, College of Physicians and Surgeons, Columbia University. Synthetic oligonucleotides were purchased from Life Technologies Inc. Open reading frames (ORFs) were compared to sequences in the GenBank/EMBL database by using the BLAST program. The PSORT program (at GenomeNet, University of Tokyo [60]) was used to identify potential transmembrane domains and to predict the cellular location of the predicted protein product. The LALIGN program (at the GeneStream network server, Institut de Génétique Humaine, Montpellier, France) was used to align the amino acid sequences of two proteins and calculate the percentages of sequence homology and identity.

Identification of the L. pneumophila rpoS gene by functional complementation.

A library (MW67 [63]) of L. pneumophila chromosomal DNA was electroporated into strain LM5005, and the transformants were plated on Luria-Bertani (LB) plates containing chloramphenicol and X-Gal. Strain LM5005 (rpoS::Tn10 osmY::lacZ) will form dark blue colonies on LB containing X-Gal in the presence of rpoS (81). Approximately 20,000 transformants were visually inspected for the ability to form a dark blue colony, and 11 were analyzed further. The plasmid DNA from these colonies was reintroduced into strain LM5005 to confirm that the blue colony phenotype was plasmid dependent. Each of the 11 plasmids isolated from the library was able to complement the catalase-negative phenotype of strain LM5005 (59) by using the catalase test (44) in which 10 μl of 30% (wt/vol) hydrogen peroxide solution is dropped onto a colony (data not shown).

Plasmid pLM507 was studied further and found to contain two EcoRI fragments (Fig. 1). A Southern blot of L. pneumophila genomic DNA probed with the E. coli rpoS gene hybridized with a 3-kb L. pneumophila PstI DNA fragment (76). Plasmid pLM507 contained a 3,201-bp PstI fragment and a contiguous 2,502-bp PstI fragment (Fig. 1). These two PstI fragments were subcloned into pBluescript II KS(+) to create plasmids pLM546 and pLM549, and both strands of DNA of the inserts from each plasmid were sequenced. One partial ORF from the 3,201-bp PstI fragment along with another partial ORF from the 2,502-bp PstI fragment (Fig. 1) together encoded for the N terminus and the C terminus, respectively, of a protein that is homologous to the E. coli rpoS gene product. Because the coding region for L. pneumophila rpoS was contained on two separate subclones, we confirmed that the rpoS gene was indeed contiguous by sequencing the DNA across the PstI site from plasmid pLM507. Together, sequence from these plasmids generated a contiguous PstI-EcoRI fragment from the L. pneumophila genome that was 5,658 bp in length.

FIG. 1.

Schematic diagram of the ORFs (indicated by arrows) encoded on pLM507. At the top is a restriction map of pLM507. Restriction sites: R, EcoRI; S, SalI; H, HindIII; P, PstI; V, EcoRV. Plasmid pLM507 contains two EcoRI fragments 7.2 and 0.8 kb in size. It is not known if these fragments are contiguous on the L. pneumophila chromosome, and the 0.8-kb fragment was not studied. A PstI digest of pLM507 produced two fragments 3.2 and 2.5 kb in size. One of the PstI sites of the 2.5-kb fragment is from the vector pMMB207 and is 39 bp from the EcoRI site of L. pneumophila DNA. The 5,658-bp L. pneumophila contiguous PstI-EcoRI fragment is represented below pLM507. An enlarged view of the ORFs (represented as arrows) is shown. The gene names given the respective ORFs are based on homology from searches of the GenBank database (Table 2).

A subclone of rpoS was generated by PCR. Two synthetic oligonucleotides (5′-GCGCGTTAATGCAGGGCAGG, which anneals to nucleotides 2209 to 2228, and 5′-CCAAAGAACTACTGGCAAG, which anneals to nucleotides 3652 to 3635) were used in a PCR with the Easy Start PCR premixes (Molecular Bio-Products Inc.) and plasmid pLM507 as the template. The 1,443-bp PCR product that was generated contained 213 bp upstream of the ATG codon of rpoS and 205 bp downstream of the stop codon of rpoS. This PCR product was cloned into pZErO-2.1 that had been digested with EcoRV to generate plasmid pLM799. Plasmid pLM799 was digested with EcoRI and XbaI to remove the fragment of DNA encoding rpoS and was ligated with pBC SK(+) to form plasmid pLM806, which was used in the complementation studies.

Construction of Tn903dGent.

Our laboratory already has a collection of strains (icm) containing Tn903dIIlacZ mutations (67). This transposon (18) is stably maintained in chromosome of L. pneumophila in the absence of the transposase TnpA (83). An alternate transposon was constructed for mutagenesis in order to study the regulation of genes containing Tn903dIIlacZ insertions. This allows a strain to contain both a Tn903dIIlacZ and a mutation in an additional gene (for example, LM1395 [see below]). TnpA requires only the 18-bp inverted repeats at the ends of the transposon for efficient transposition. Therefore, the sequences between these inverted repeats were deleted from Tn903dIIlacZ and replaced with a gentamicin resistance cassette. The gentamicin resistance cassette was cloned from pLB41 by digestion with HindIII, and subsequent ligation of this 2,976-bp fragment with HindIII-digested pYSF31 formed pLM579. When cloned into pYSF31, the HindIII fragment is flanked by XbaI sites. Therefore, pLM579 was digested with XbaI, and the gentamicin resistance cassette was ligated with XbaI-digested pKD368 to form pLM598, thereby replacing the kanamycin resistance cassette and ′lacZ genes within the 18-bp repeats of Tn903 with the gentamicin resistance cassette. This transposon, named Tn903dGent, transposes efficiently in E. coli (data not shown).

Transposition of Tn903dGent into rpoS.

First, pLM507 (Cam^r Mob⁺) was transformed into strain LW252 (Tet^s) to form strain LM5316. Then pLM598 encoding Tn903dGent (Amp^r Gent^r Mob⁻) was transformed two independent times into strain LM5316, and the transformants from each electroporation were pooled. Each pool was frozen in 10% glycerol and stored at −80°C as LM5408 and LM5409. Aliquots of LM5408 and LM5409 were grown overnight in LB medium containing chloramphenicol, ampicillin, and gentamicin. The following day, 0.5 ml of overnight culture was subcultured into LB medium containing no antibiotics and grown for 1.5 h. A bacterial mating was then performed with each donor (LM5408 and LM5409), using LM5005 (Tet^r) as a recipient in order to mobilize pLM507::Tn903dGent away from pLM598 containing tnpA. Strain LM5005 was used as an indicator of RpoS function to facilitate screening of the pLM507::Tn903dGent insertions for one that contained an insertion in rpoS. Transconjugants were plated on LB plates containing X-Gal, tetracycline, chloramphenicol, and gentamicin. Several hundred gentamicin-resistant, ampicillin-sensitive colonies were screened, and the light blue colonies were purified and examined in the catalase test. The plasmid DNA was purified and retransformed into LM5005 to confirm the phenotype.

The precise position of the transposon insertions was mapped by PCR. A reaction mixture containing oligonucleotides that hybridize to the ends of the transposon (5′-GGGGCTGACTTCAGGTGCTA and 5′-CGAATTCCTGCAGGCATGCC) along with oligonucleotides that hybridize to the DNA flanking the L. pneumophila rpoS sequences were used. PCR products from three plasmids, named pLM654, pLM655, and pLM658, were cloned into the vector pCR2.1 and sequenced to identify the transposon-rpoS junction sequence. The transposon insertions for pLM654 (rpoS4), pLM655 (rpoS5), and pLM658 (rpoS13) mapped to nucleotides 3179, 3235, and 3282, respectively.

Construction of the rpoS mutant alleles.

The DNA fragment containing rpoS::Tn903dGent was cloned from pLM654, pLM655, and pLM658 by digestion with BglII and NheI and subsequent ligation with pLAW344 that had been digested with BamHI and XbaI to form pLM670, pLM671, and pLM673, respectively. Allelic exchange of the rpoS::Tn903dGent fusions onto the chromosome of L. pneumophila JR32 was performed as described previously (83) and resulted in strains LM1376, LM1381, and LM1386, respectively. To rule out the possibility that the rpoS null strains acquired compensatory mutations which might contribute to the phenotypes observed, three other isolates (LM1375, LM1380, and LM1385) were also used in most experiments. In all cases, the phenotypes of all six strains were virtually identical (data not shown). In addition, we reconstructed the null strains a second time and performed several of the experiments again, obtaining the same results (data not shown). Plasmids pLM670 and pLM673 were used in an allelic exchange with strain LELA14 as described previously (83) to construct strains LM1395 and LM1397, respectively. Strain LM1397 gave the same results as LM1395 (data not shown). Plasmids pLM670 and pLM671 were used in an allelic exchange with strain LELA2955 as described previously (83) to construct strains LM1389 and LM1392, respectively. Strain LM1392 gave the same results as LM1389 (data not shown). All strain constructions were confirmed by Southern blot analysis using a Boehringer Mannheim DIG DNA labeling and detection kit according to the manufacturer’s instructions (data not shown).

The DNA fragment containing rpoS was cloned from pLM507 by digestion with BglII and NheI and subsequent ligation with pLAW344 that had been digested with BamHI and XbaI to form pLM845. To construct merodiploid strains containing both mutant and wild-type copies of the rpoS gene, strains LM1376, LM1381, and LM1386 were transformed with plasmid pLM845, which contains wild-type rpoS and is unable to replicate in L. pneumophila. This plasmid integrates into the chromosome by homologous recombination as described previously (83) and resulted in strains LM1580, LM1584, and LM1588, respectively. Chloramphenicol- and gentamicin-resistant, sucrose-sensitive transformants were selected and were confirmed to be merodiploid by PCR analysis (data not shown). The phenotypes of LM1584 and LM1588 were the same as those described for LM1580 (data not shown).

Western blot analysis.

Crude extracts of E. coli and L. pneumophila for Western blot analysis were prepared as follows. Strains were grown to logarithmic or stationary phase, and a volume of cells which would result in equivalent numbers of cells for all cultures was pelleted. Cell pellets were resuspended in 300 μl of sodium dodecyl sulfate sample buffer, boiled for 3 min, and clarified by centrifugation. Gel electrophoresis of proteins and Western blot analysis were performed as described elsewhere (54a). RpoS was detected on the blots by using rabbit anti-RpoS antibody (a generous gift from R. Hengge-Aronis) and a chemiluminescence kit (Pierce).

β-Galactosidase assays.

β-Galactosidase assays were performed as described elsewhere (55). L. pneumophila strains were grown in N-(2-acetamido-2-aminoethanesulfonic acid (ACES)-buffered yeast extract (AYE) broth, except that in the assays done in conjunction with pigmentation, bovine serum albumin (BSA) was added to the AYE broth to a final concentration of 7%. The induction and pattern of β-galactosidase activity of the pig::lacZ fusion were the same in the presence or absence of BSA (data not shown), indicating that neither BSA nor any of its amino acid components act as inducers of the pig gene. For both strains, cells were washed one time with 1× M63 salts (72) prior to assays. The substrate for lacZ hydrolysis was o-nitrophenyl-β-d-galactopyranoside.

Growth curves and pigmentation measurements.

L. pneumophila strains were grown for 2 days on ACES-buffered charcoal yeast extract (ABCYE) plates at 37°C; 5 ml of AYE medium was then inoculated, and cells were grown overnight. The overnight culture was subcultured into 30 ml of AYE so that the absorbance at 600 nm was 0.1. Cultures were grown at 37°C for 64 h; a 1-ml sample was removed every 2 to 4 h for a measurement of absorbance at 600 nm, then serially diluted in 1× M63 salts, and plated for CFU on ABCYE plates. Growth of L. pneumophila for pigmentation measurements was as described above except that BSA was added to the AYE to a final concentration of 7% to enhance browning. Pigmentation was analyzed by measuring the absorbance at 550 nm of culture supernatants (83).

Assays for survival under stress conditions.

L. pneumophila strains were grown for 2 to 3 days on ABCYE plates at 37°C; 5 ml of AYE medium was then inoculated, and cells were grown for at least 18 h for the stationary-phase stress experiments. The absorbance at 600 nm for stationary-phase cultures at the time of the assay was typically between 3 and 4, and the initial CFU count was about 10⁹ per ml. For the log-phase stress experiments, an overnight culture of L. pneumophila was subcultured to an absorbance at 600 nm of 0.1 and grown for 6 to 8 h. The absorbance at 600 nm for log-phase cultures at the time of the assay was typically between 0.4 and 0.7, and the initial CFU count was about 10⁸ per ml. Cells were centrifuged and resuspended in an equal volume of 1× M63 salts to measure the untreated CFU. For the different stress conditions, the cell pellet was resuspended in an equal volume solution of 5 M sodium chloride for osmotic stress, 10 mM H₂O₂ for oxidative stress, or 0.1 M citric acid at pH 3 for acid stress. Cells were incubated in a 37°C water bath. At each time point, the cells were washed with 1× M63 salts and serially diluted to determine CFU on ABCYE agar plates.

HL-60 and THP-1 cell culture methods and procedures.

The human leukemia cell line HL-60 and the human mononuclear phagocytic cell line THP-1 were used in tissue culture studies. Cells were maintained and differentiated as described elsewhere (67, 77a) except that no antibiotics were added to the tissue culture medium. Growth of L. pneumophila in HL-60 cells and determination of cytotoxicity of L. pneumophila for HL-60 and THP-1 cells were carried out as described elsewhere (52, 63).

A. castellanii culture methods and procedures.

Growth and maintenance of A. castellanii ATCC 30324 in PYG medium in 75-cm² tissue culture flasks were as described elsewhere (9, 57). The assay for replication in amoebae was based on previously described methods (9, 57). L. pneumophila was added at a multiplicity of infection (MOI) of 10 to an adherent monolayer of 1.2 × 10⁵ amoebae. After incubation for 30 min at 37°C to allow for infection, the wells were washed three times with 0.5 ml Ac buffer. A sample of the infection supernatant was removed once every 24 h for 4 days. CFU of extracellular bacteria were quantitated on ABCYE plates. An alternative spot assay was performed whereby 5 × 10⁵ amoebae were spread on a charcoal-yeast extract (CYE) plate. Individual colonies of L. pneumophila were spotted with a toothpick into the CYE plate and a CYE plate spread with amoebae. The plate was incubated at 28°C for 4 to 5 days and then visually inspected for the growth of each spot of L. pneumophila. Wild-type JR32 grows equally well on both plates. Strain 25D and icm mutant strains which are unable to grow within A. castellanii do not form visible growth on the CYE plate spread with the amoebae (70a).

Nucleotide sequence accession number.

The 5,658-bp PstI-EcoRI fragment from the L. pneumophila genome has been deposited in the GenBank database under accession no. AF117715; all references to nucleotide numbers correspond to this entry.

RESULTS

Isolation of the L. pneumophila rpoS gene.

The approach that we took to isolate genes encoding regulatory proteins from L. pneumophila involves complementation of a mutant gene product from E. coli. To clone the L. pneumophila homolog of E. coli rpoS, we constructed strain LM5005, which contains a Tn10 insertion in the rpoS gene and an rpoS-dependent lacZ fusion to the osmY gene, which encodes an osmotically induced periplasmic protein (81, 86). Strain LM5005 was electroporated with a library of L. pneumophila chromosomal DNA, and plasmid pLM507 was isolated and studied further (see Materials and Methods). Five complete and two partial ORFs were contained on a PstI-EcoRI fragment from pLM507 that was 5,658 bp in size (Fig. 1; Table 2). The DNA and deduced amino acid sequences for each ORF were used to search the appropriate databases. One 341-amino-acid ORF encoded a protein that is homologous to E. coli RpoS. We concluded that this ORF encoded the L. pneumophila homolog of the E. coli rpoS gene.

TABLE 2.

Characteristics of proteins encoded on pLM507

Gene	No. of amino acids of ORF	Predicted molecular mass (kDa)	Distinguishing characteristics/predicted locationa	% Amino acid sequence identity to other proteinsc	Proposed function in L. pneumophila based on GenBank homology
sfcA	>117	>13.5	CYTO	54.7b	Malate oxidoreductase
orfA	127	14.6	IM (1)		Unknown
surE	252	27	CYTO	57	Stationary-phase survival
nlpD	247	27.5	OM/IM, LP	39.2	Cell wall formation
rpoS	341	39.4	CYTO	59.5d	Sigma factor
hmgA	342	39	CYTO	48e	Phenylalanine metabolism
yebC	>238	>25.8	CYTO	58.4b	Unknown

^aCalculated by using PSORT. CYTO, cytoplasm; IM, inner membrane (number of transmembrane domains in parentheses), OM, outer membrane; LP, lipoprotein. 

^bCalculated from the partial sequence information. 

^cHomology to E. coli homologs in GenBank database except for hmgA, calculated by using LALIGN except as noted otherwise. 

^dCalculated by using the Genetics Computer Group (University of Wisconsin) software. 

^eHomology to Homo sapiens homolog in GenBank database. 

The L. pneumophila RpoS sequence was aligned with the E. coli RpoS sequence. The sequences were 78.4% similar and 59.5% identical at the amino acid level over their entire lengths. The homology of the N-terminal portion of the RpoS sequences is low, a characteristic typical of sigma factors (49). An alignment of the C-terminal amino acids of L. pneumophila RpoS and E. coli RpoS increases the sequence identity to 68.5% (LALIGN). Interestingly, the flagellar sigma factor (rpoF or ς²⁸) of L. pneumophila was found to be only 43% identical to that of E. coli (30).

Five of the other ORFs in the region of rpoS had homology to other genes in the databases (Fig. 1; Table 2; references 8, 22, 47, and 50). In an arrangement identical to that for E. coli (38, 45), the L. pneumophila nlpD gene mapped immediately upstream of the L. pneumophila rpoS gene. On the E. coli chromosome, the nlpD gene and the surE gene are not contiguous as they are on the L. pneumophila chromosome, but they are separated by the pcm (protein carboxyl methyltransferase) gene. Adjacent to rpoS and transcribed in the opposite direction is a gene containing homology to the hmgA gene (encoding homogentisate dioxygenase) of humans and fungi. At the time of submission, the L. pneumophila hmgA gene was the first prokaryotic homolog of this gene recorded in the GenBank database. In summary, the analysis of this 6-kb region of the L. pneumophila chromosome suggests that evolutionarily, L. pneumophila acquired three distinct regions similar to the E. coli chromosome (sfcA from min 33, yebC from min 41, and surE, nlpD, and rpoS from min 61) in addition to a homolog of a eukaryotic gene (hmgA).

To quantitate and confirm the complementation of the E. coli rpoS mutation by the L. pneumophila rpoS gene, β-galactosidase assays were performed to measure expression of the osmY-lacZ fusion in the presence of the L. pneumophila rpoS. Plasmid pLM507 or the vector pMMB207 were transformed into strain LM5005. The β-galactosidase levels from strains RO151 and RH90 were also measured as controls. In the absence of an rpoS gene, the rpoS-dependent osmY-lacZ fusion is expressed at a low level in strain RH90, accumulating only 10 Miller units of β-galactosidase in stationary-phase cells (Table 3). However, in the presence of the L. pneumophila rpoS on pLM507, transcription of the osmY-lacZ fusion is activated 19-fold (Table 3). These results confirm the functional complementation of the E. coli rpoS mutation by the L. pneumophila rpoS gene.

TABLE 3.

RpoS from L. pneumophila functionally complements E. coli RpoS

Straina	Mean β-galactosidase activityb ± SD
MC4100	0
RH90	0
RO151	519 ± 38
LM5005/pMMB207	10 ± 0
LM5005/pLM507	186 ± 28
LM5005/pLM654	12 ± 0
LM5005/pLM655	12 ± 0
LM5005/pLM658	11 ± 0

^aStrains were grown to stationary phase in LB medium prior to assay. 

^bExpressed as Miller units (55). 

Construction of a strain containing a mutation in the L. pneumophila rpoS gene.

To study possible functions of RpoS in L. pneumophila, we constructed a mutation in the rpoS gene on the chromosome of strain JR32. The transposon Tn903dGent was used in a random mutagenesis of pLM507 in E. coli. Three transposon insertions mapped within the coding region of the L. pneumophila rpoS gene. Alleles rpoS4, rpoS5, and rpoS13 were recombined separately onto the chromosome of L. pneumophila JR32 by allelic exchange as described elsewhere (83) to generate strains LM1376, LM1381, and LM1386, respectively.

To confirm that the transposon insertions indeed conferred an rpoS null phenotype, we transformed plasmids pLM654, pLM655, and pLM658 containing the three different alleles of rpoS::Tn903dGent into strain LM5005 and performed β-galactosidase assays. The insertional inactivation of the L. pneumophila rpoS gene by Tn903dGent resulted in a 17-fold reduction of the β-galactosidase levels of the osmY-lacZ fusion in LM5005, as expected (Table 3). We believe that polarity of the transposons in rpoS on downstream genes does not play a role in the phenotype we report because these β-galactosidase assays are specific for an rpoS⁺ or rpoS mutant condition; more importantly, there is not a gene immediately downstream of rpoS that is transcribed in the same direction on the L. pneumophila chromosome (Fig. 1).

Examination of RpoS levels in L. pneumophila.

To examine RpoS levels in L. pneumophila, crude cell extracts were prepared from wild-type and mutant strains grown to log or stationary phase and compared with extracts from E. coli wild-type and rpoS strains. Gel electrophoresis of proteins was followed by Western blot analysis using anti-E. coli RpoS antibody. The antibodies that were raised against E. coli RpoS recognized L. pneumophila RpoS. The results show that like E. coli RpoS (Fig. 2A), L. pneumophila RpoS is induced in stationary phase (Fig. 2B). Taken together, these results are consistent with L. pneumophila RpoS being a stationary-phase sigma factor. We observe that RpoS is not present in the L. pneumophila rpoS null strain LM1376 (Fig. 2B), as predicted. Additionally, the complementing plasmid pLM806 overexpresses RpoS in both E. coli and L. pneumophila (Fig. 2).

FIG. 2.

Western blot analysis of crude cell extracts. Six times more L. pneumophila than E. coli crude extract was loaded. (A) Relative amounts of E. coli RpoS from log-phase (lanes 1, 3, and 5) and stationary-phase (lanes 2, 4, and 6) cultures of LM5000 (lanes 1 and 2), LM5005 (lanes 3 and 4), and LM5005 containing pLM806 (lanes 5 and 6). (B) Relative amounts of L. pneumophila RpoS from log-phase (lanes 1, 3, and 5) and stationary-phase (lanes 2, 4, and 6) cultures of JR32 (lanes 1 and 2), LM1376 (lanes 3 and 4) and LM1376 containing pLM806 (lanes 5 and 6).

Growth of wild-type and rpoS mutant strains in AYE medium.

An E. coli strain with a mutation in rpoS is less able to survive in stationary phase (44, 53, 59). Therefore, we examined the growth and survival of the L. pneumophila wild-type and rpoS mutant strains. Growth was monitored for 64 h by reading the absorbance at 600 nm (Fig. 3A) and measuring CFU (Fig. 3B). The results show that the rpoS mutation in strain LM1376 does not have a dramatic effect on the ability of L. pneumophila to grow in batch culture or to survive during stationary phase in AYE broth. In fact, we consistently observed that the CFU of the wild-type strain decreased to undetectable levels approximately 4 to 6 h before the mutant strain (Fig. 3B), in contrast to findings reported for E. coli (44, 53, 59).

FIG. 3.

Growth curves of wild-type JR32 (squares) and rpoS null strain LM1376 (diamonds) in AYE medium. (A) Absorbance at 600 nm, measured every 2 to 4 h during a 64-h time course, plotted as a function of time; (B) CFU, measured by plating dilutions of bacterial strains on ABCYE plates during the same time course, plotted as a function of time.

Investigating possible candidates for rpoS-regulated genes in L. pneumophila.

L. pneumophila forms a brown pigment during the stationary phase of growth. We were interested in exploring the possibility that the pigment production gene (pig [83]) is regulated by rpoS. Therefore, we measured the absorbance of the culture supernatants at 550 nm from strains JR32 and LM1376. Strain LELA14, which contains a Tn903dIIlacZ mutation in the pig gene and does not form the characteristic brown pigment, was used as a control (83). We observed identical induction patterns of pigmentation for the wild-type and strain LM1376 (data not shown). To confirm this result, we investigated the possibility that rpoS activates transcription the pig gene itself. Transposition of Tn903dIIlacZ into the pig gene of strain LELA14 formed a reporter gene (lacZ) fusion to pig (83). We constructed a derivative of this strain, LM1395, which contained a null mutation in the rpoS gene and measured β-galactosidase activity. The β-galactosidase activity of the pig-lacZ fusion in the rpoS background followed the same pattern of induction, peak, and reduction as the activity of the strain in the rpoS⁺ background (data not shown). These results indicate that RpoS is not involved in the stationary-phase induction of the brown pigment of L. pneumophila.

Our laboratory has a collection of Tn903dIIlacZ mutants (icm strains) that are unable to multiply within or kill macrophages (67) or replicate within A. castellanii (70). We were interested in the possibility that rpoS regulates the expression of one or more of these genes. One indication of an rpoS-regulated gene would be an increase in its expression upon entry of cells into stationary phase. Because transposition of the Tn903dIIlacZ into the icm genes results in transcriptional and translational fusions to the lacZ gene, expression of the icm genes during growth is correlated with the β-galactosidase activity of the gene fusion. Therefore, we measured the activity of a representative group of the icm-lacZ fusions to see if one or more of them were induced during the stationary phase of growth. The results show that the levels of icm gene expression are not different in logarithmic and stationary phases (Table 4). Because we have no data indicating that RpoS is required for the expression of genes in stationary phase, we directly examined the requirement of RpoS for icm gene expression by constructing strain LM1389, which is a derivative of LELA2955 containing a mutation in the rpoS gene. We examined the β-galactosidase activity of the icmX::lacZ fusion of strain LM1389 in the logarithmic and stationary phases of growth and compared it to the β-galactosidase activity of LELA2955. The two strains were identical with respect to β-galactosidase expression (data not shown). These results further indicate that RpoS is not a general regulator of icm gene expression.

TABLE 4.

Growth phase regulation of icm genes

Straina	Geneb	Mean β-Galactosidase activityc ± SD
		Log-phase (A600 = 0.2–0.7)	Stationary phase (A600 = 4–6)
LELA1012	icmB	139 ± 3	228 ± 25
LELA1205	icmD	20 ± 3	22 ± 0
LELA4432	icmE	33 ± 2	51 ± 3
LELA1275	icmF	14 ± 2	9 ± 0
LELA4086	icmT	17 ± 1	6 ± 0
LELA3278	icmR	1,001 ± 57	1,079 ± 18
LELA3463	icmQ	223 ± 22	205 ± 1
LELA3037	icmP	5 ± 1	2 ± 0
LELA4004	icmX	98 ± 8	108 ± 6
LELA1747	icmV	44 ± 5	22 ± 1
JR32		2 ± 1	0

^aStrains contain respective complementing clones (27a, 63, 68, 69). 

^bName of gene containing Tn903dIIlacZ insertion (27a, 63, 67–69, 71). 

^cExpressed as Miller units (55). 

RpoS is not required for a growth phase-dependent response to stress.

Because L. pneumophila may encounter harsh conditions upon entering a eukaryotic host, we wished to test whether RpoS plays a role in survival of L. pneumophila during acid, oxidative, or osmotic stress. In E. coli, logarithmic phase cells are much more sensitive to stress conditions than stationary-phase cells. This resistance of stationary-phase cells to stress conditions is proposed to be controlled by rpoS (44). To determine whether L. pneumophila has a stationary phase-induced stress resistance, cultures of the wild-type strain JR32 were grown to logarithmic or stationary phase and subjected to various stress conditions (Fig. 4A to C). Indeed, when strain JR32 was subjected to conditions of pH 3 (Fig. 4A), 10 mM H₂O₂ (Fig. 4B), or 5 M sodium chloride (Fig. 4C), we found that the cells grown to stationary phase were much more resistant to the stress conditions than cells grown to log phase. When the wild-type strain was subjected to pH 3 or 10 mM H₂O₂, a striking immediate decrease in the CFU was observed. Under osmotic stress, however, the decrease was less dramatic. These results show that wild-type L. pneumophila possesses a stationary-phase-dependent stress resistance.

FIG. 4.

Ability of L. pneumophila strains to survive under stress. The log of the percent survival of each culture is plotted as a function of time. Individual experiments were performed two to four times, and the results of one representative experiment are shown. (A) Survival of log-phase (filled squares) and stationary-phase (open squares) JR32 at pH 3. (B) Survival of log-phase (filled squares) and stationary-phase (open squares) JR32 in the presence of 10 mM hydrogen peroxide. (C) Survival of log-phase (filled squares) and stationary-phase (open squares) JR32 in the presence of 5 M sodium chloride. (D) Survival of log-phase (filled squares) and stationary-phase (open squares) LM1376 at pH 3. (E) Survival of log-phase (filled squares) and stationary-phase (open squares) LM1376 in the presence of 10 mM hydrogen peroxide. (F) Survival of log-phase (filled squares) and stationary-phase (open squares) LM1376 in the presence of 5 M sodium chloride.

We then investigated a possible role for RpoS in the growth phase-dependent resistance to stress. We measured the ability of the strain containing a mutation in rpoS to survive stress conditions during growth in log and stationary phases by repeating the experiment described above with strain LM1376 (Fig. 4D to F). We observed that under conditions of pH 3 (Fig. 4D), 10 mM H₂O₂ (Fig. 4E), or 5 M sodium chloride (Fig. 4F), the rpoS mutant strain grown to logarithmic phase was much more sensitive to the stress conditions than a strain grown to stationary phase. These results suggest that unlike what has been reported for E. coli, L. pneumophila rpoS is not required for a growth phase-dependent resistance to stress. We note that the rpoS mutant strain seems to have a decreased ability to survive in 5 M sodium chloride in the logarithmic phase of growth, suggesting that RpoS may regulate genes involved in osmotic stress survival in log phase.

Examining the growth of the rpoS mutant strain in eukaryotic hosts.

We were next interested in examining the importance of the L. pneumophila RpoS for intracellular growth and host cell killing. First we measured the ability of strain LM1376 to replicate within the macrophage-like cell line HL-60. L. pneumophila wild-type strain JR32 and mutant strain 25D, which is unable to replicate intracellularly (35), were used as controls. The results show that rpoS mutant strain LM1376 replicated to the same degree as JR32 (Fig. 5A). Therefore, we conclude that rpoS is not required for L. pneumophila to replicate in HL-60 cells. We then examined the ability of strain LM1376 to kill HL-60-derived macrophages by using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay (52). The results show that strain LM1376 is able to kill HL-60-derived macrophages to the same extent as the wild-type strain (Fig. 5B). We also examined the ability of strain LM1376 to kill cells of the human mononuclear phagocytic line THP-1 (Fig. 5C) and found that the L. pneumophila rpoS is not required for killing of THP-1 cells. Taken together, these results indicate that rpoS is not required for L. pneumophila to replicate within and kill human macrophages.

FIG. 5.

Growth of L. pneumophila strains in eukaryotic hosts. Assays were performed in triplicate; the standard error bars indicate standard errors of the means and may not be visible. (A) Replication of wild-type JR32 (open squares), mutant 25D (circles), and rpoS null strain LM1376 (filled squares) in HL-60-derived macrophages. A monolayer of HL-60 cells was differentiated and infected with L. pneumophila bacteria at an MOI of 0.01. Infection wells were sacrificed, and the net log number of CFU was plotted as a function of time. (B and C) Cytotoxicity of HL-60 (B)- and THP-1 (C)-derived macrophages by wild-type JR32 (open squares), mutant 25D (circles), and rpoS null strain LM1376 (filled squares). An MTT assay was performed, the number of viable macrophages 5 days after infection with L. pneumophila was measured, and the absorbance at 570 nm was plotted as a function of the log of the number of bacteria in the wells. (D) Replication of wild-type JR32 (open squares), mutant 25D (circles), rpoS null strain LM1376 (filled squares), rpoS null strain LM1376 containing pLM806 (triangles), and JR32 containing pLM806 (hatched squares) within A. castellanii. CFU were measured from infection supernatants, and net log growth was plotted as a function of time.

L. pneumophila normally replicates in the environment within protozoan hosts such as A. castellanii (65, 66). We therefore tested the ability of the wild-type and rpoS mutant strains to replicate in these amoebae. Strains JR32, 25D, and LM1376 were used to infect A. castellanii at an MOI of 10 (Fig. 5D). While the wild-type strain was able to replicate 10⁴-fold within amoebae, strain LM1376, containing a mutation in rpoS, was unable to replicate (Fig. 5D). Therefore, L. pneumophila rpoS was absolutely required for multiplication within A. castellanii. These data indicate that rpoS likely regulates one or more genes in L. pneumophila that are required for growth in A. castellanii.

To examine the ability of a wild-type copy of L. pneumophila rpoS in trans to complement the growth defect in amoebae, we electroporated plasmid pLM806 into strain LM1376 and measured the ability of this strain to replicate in A. castellanii. Plasmid pLM806 was unable to complement the ability of strain LM1376 to grow within amoebae (Fig. 5D). Western blot analysis confirmed that RpoS is expressed from pLM806 in strain LM1376 (Fig. 2B); therefore, this phenotype is not due to lack of RpoS expression from pLM806. Interestingly, when transformed into wild-type strain JR32, pLM806 inhibited the growth of JR32 in amoebae (Fig. 5D). We attempted to complement the growth defect in amoebae by using the original larger clone, plasmid pLM507. This plasmid was also unable to complement the ability of strain LM1376 to replicate in amoebae (data not shown). Last, we constructed a merodiploid strain by integrating a wild-type copy of rpoS into the mutant strain LM1376 and assayed the ability of this strain to replicate in A. castellanii by using the spot assay as described in Materials and Methods. In this assay, wild-type JR32 is able to form a colony on the CYE plate containing amoebae whereas 25D is killed by the amoebae (data not shown). The merodiploid strain in which a wild-type copy of the rpoS gene is integrated into the chromosome was able to grow in the presence of A. castellanii (data not shown). This finding indicates that in single copy, the wild-type rpoS gene is able to complement the rpoS::Tn903dGent mutation.

DISCUSSION

In the environment and during growth within eukaryotic hosts, L. pneumophila likely encounters adverse conditions such as nutrient deprivation and osmotic or oxidative stress. Similar to other well-studied bacterial pathogens, L. pneumophila likely regulates gene expression in response to changes in its environment. Nothing is known about how L. pneumophila coordinately regulates the expression of genes which are needed during the course of infection. RpoS (ς^S) is a sigma factor known to be involved in regulation of genes induced in stationary phase and during stress conditions such as nutrient deprivation or hyperosmolarity (19, 29, 48). To elucidate global regulatory networks, we identified the L. pneumophila homolog of RpoS by complementation in E. coli. We found that the L. pneumophila rpoS was able to functionally substitute for E. coli rpoS in the activation of osmY-lacZ and the catalase gene katE. We constructed a null mutation in L. pneumophila rpoS and found that complementation of these phenotypes was abolished in E. coli. Thus, functional complementation in E. coli is a valid method to clone L. pneumophila genes (30, 31, 77).

We tested the ability of the L. pneumophila rpoS mutant strain for its ability to survive extreme oxidative, osmotic, and acidic stress conditions. As has been reported for E. coli (28, 40, 41), the wild-type strain of L. pneumophila possesses a stationary phase-induced stress resistance phenotype. However, our results show that the L. pneumophila rpoS gene is not required for this growth phase-dependent stress resistance. Perhaps another, as yet unidentified sigma factor or other global regulator plays a role in the stationary phase resistance of L. pneumophila to stress. The requirement of other regulators for survival during stress has been demonstrated for other pathogens (7, 12). It is not known what adverse conditions are encountered by L. pneumophila during intracellular growth. Possibly, stress is not encountered within the distinct phagosome in which L. pneumophila resides. L. pneumophila strains containing Tn903dIIlacZ insertions which render them unable to kill macrophages possess no increased sensitivity to oxidative stress (67), suggesting that these phenotypes are not correlated. Although it was demonstrated that some L. pneumophila proteins that are induced upon entry into macrophages are also induced by stress conditions (1), one protein that was identified, GspA, plays a role in the stress resistance in log phase but is not required for growth in eukaryotic hosts (2).

We investigated the possibility that rpoS regulates one or more of the icm genes, which are required for L. pneumophila intracellular growth and host cell killing (67). One characteristic of some rpoS-regulated genes is their induction in stationary phase (for example, references 5, 17, and 43). We measured β-galactosidase activity of the icm-lacZ fusions during log and stationary phases and found no evidence of growth phase regulation of any of the icm genes. Although these assays were performed on cells grown in laboratory medium, no induction of icmX::lacZ or dotA::lacZ was observed in β-galactosidase assays performed with L. pneumophila icm strains that had been replicating in HL-60-derived macrophages at 3 and 20 h postinfection (17a). Byrne and Swanson (13) proposed that L. pneumophila undergoes a phenotypic switch during growth within macrophages such that it converts from a replicative phenotype to a virulent phenotype in the post-exponential phase of growth, which suggests that L. pneumophila induces expression of virulence factors in stationary phase. Our data indicate that icm genes are probably not involved in this phenotypic switch.

In the environment, L. pneumophila likely survives by replicating within protozoan hosts. We show here that L. pneumophila rpoS is required for growth in A. castellanii but not in human macrophages. The infection pathways for macrophages and amoebae share some characteristics, but a few differences are apparent (3, 23). L. pneumophila has a subset of genes that are required for growth in both hosts (4, 14, 67, 70) which might act at steps in the infection process that are common between the macrophage and protozoan host, such as replication (23). L. pneumophila also has genes that are required for growth in one host but not another (24, 27, 70). The products of rpoS and these genes might act at steps in the infection process, such as entry, where the pathways differ (23). Virulence for a protozoan host was required for maximal intrapulmonary growth of L. pneumophila in a mouse model (11). Additionally, growth in protozoan hosts enhances the virulence (10, 15) and stress resistance (2) of L. pneumophila. The protozoan host may be more relevant as a model for L. pneumophila survival in the environment and may prove to be a more restrictive host.

We attempted to complement the defect of the rpoS null strain for growth within A. castellanii by providing the wild-type rpoS gene on a plasmid in two different constructions and were unable to complement the defect. Other groups have reported an inability to complement an rpoS mutant phenotype and attributed it to either differences in the rpoS expression from chromosomal and episomally encoded genes (85) or absence of the correct rpoS promoters (39, 74). The expression of E. coli rpoS is intricately regulated at the level of transcription, translation, and protein stability (48). Analysis of the mRNA control elements for L. pneumophila rpoS indicates it possesses the same regulatory signals as E. coli (29a) and thus is likely regulated in the same manner as observed in E. coli. Therefore, overexpression of rpoS in L. pneumophila from a plasmid may interfere with its regulation and disrupt the ability of the gene to complement. To avert this problem, we constructed a merodiploid strain by integrating a wild-type copy of rpoS into the chromosome of the rpoS mutant strain. We found that this strain was complemented, indicating that the copy number of the rpoS gene is a critical aspect of complementation.

Interestingly, the presence of plasmid pLM806 in the wild-type strain JR32 inhibited its growth within amoebae. Plasmid pLM806 did not, however, inhibit the growth of JR32 within HL-60 cells (data not shown), indicating that this inhibitory effect is specific for growth in amoebae. Therefore, we favor the possibility that overexpression of RpoS from the plasmid results in the overexpression of a factor which leads to an inability of the bacterium to grow in amoebae. Alternatively, examples of genes regulated by both ς⁷⁰ and RpoS have been reported, and overexpression may disrupt cell function or specific gene expression if competition between these two sigma factors is unbalanced (21).

Data suggest that global control mechanisms also exist in L. pneumophila and that expression of genes is changed in response to the specific environment of the macrophage (1, 56, 78). Elucidating the regulatory cascades in L. pneumophila will lead to a better understanding of the mechanisms of L. pneumophila pathogenesis and provide information on the host cell environment and the signals that bacteria encounter during infection. Additionally, this knowledge will contribute to the information on gene regulation in pathogenic organisms in general. This may aid in the development of novel drugs as global gene regulators can serve as potential targets for antimicrobial therapy (16, 64, 75).