Functional Analysis of the Alternative Sigma-28 Factor FliA and Its Anti-Sigma Factor FlgM of the Nonflagellated Legionella Species L. oakridgensis

ABSTRACT

Legionella oakridgensis causes Legionnaires' disease but is known to be less virulent than Legionella pneumophila. L. oakridgensis is one of the Legionella species that is nonflagellated. The genes of the flagellar regulon are absent, except those encoding the alternative sigma-28 factor (FliA) and its anti-sigma-28 factor (FlgM). Similar to L. oakridgensis, Legionella adelaidensis and Legionella londiniensis, located in the same phylogenetic clade, have no flagellar regulon, although both are positive for fliA and flgM. Here, we investigated the role and function of both genes to better understand the role of FliA, the positive regulator of flagellin expression, in nonflagellated strains. We demonstrated that the FliA gene of L. oakridgensis encodes a functional sigma-28 factor that enables the transcription start from the sigma-28-dependent promoter site. The investigations have shown that FliA is necessary for full fitness of L. oakridgensis. Interestingly, expression of FliA-dependent genes depends on the growth phase and temperature, as already shown for L. pneumophila strains that are flagellated. In addition, we demonstrated that FlgM is a negative regulator of FliA-dependent gene expression. FlgM seems to be degraded in a growth-phase- and temperature-dependent manner, instead of being exported into the medium as reported for most bacteria. The degradation of FlgM leads to an increase of FliA activity.

IMPORTANCE A less virulent Legionella species, L. oakridgensis, causes Legionnaires' disease and is known to not have flagella, even though L. oakridgensis has the regulator of flagellin expression (FliA). This protein has been shown to be involved in the expression of virulence factors. Thus, the strain was chosen for use in this investigation to search for FliA target genes and to identify putative virulence factors of L. oakridgensis. One of the five major target genes of FliA identified here encodes the anti-FliA sigma factor FlgM. Interestingly, in contrast to most homologs in other bacteria, FlgM in L. oakridgensis seems not to be transported from the cell so that FliA gets activated. In L. oakridgensis, FlgM seems to be degraded by protease activities.

INTRODUCTION

Legionella oakridgensis is less virulent than L. pneumophila and is pathogenic for guinea pigs (1, 2). Recently, two human cases of Legionnaires' disease caused by L. oakridgensis were reported in France (3), showing that the bacteria are able to replicate inside human cells (1, 4–7). Moreover, it was demonstrated that L. oakridgensis is able to multiply inside amoebae (4) and that the type IV secretion system is a key virulence factor; new virulence factors were also described (4). In general, no additional cysteine is needed to grow L. oakridgensis on agar plates and extracellular glucose cannot be metabolized by the bacterium (4). For L. oakridgensis, it was demonstrated that the species is nonflagellated and that almost all flagellar regulon genes are absent, including the master regulator protein FleQ. Nevertheless, the putative alternative sigma-28 factor FliA and the putative anti-sigma-28 factor FlgM are still present (4, 8). In a closely related nonflagellated species, Legionella longbeachae, FliA, FleN, and FlgM, as well as FleQ, FleSR, and FlgD, are still present, whereas the other flagellar regulon genes are absent. Interestingly, flagellated L. pneumophila strains are motile but do not harbor chemotaxis genes and do not show chemotaxis (9). Moreover, nonflagellated, nonmotile L. longbeachae species lack almost all genes of the flagellar regulon but have been shown to have a chemotaxis operon, and L. oakridgensis is negative for the flagellar regulon and the chemotaxis-encoding genes (4, 10, 11). In general, chemotaxis enables bacteria to locate special environmental conditions and get closer to higher concentrations of attractants (12–14). It will be interesting to further investigate the presence and role of chemotaxis genes in legionellae.

The flagellar regulon is known to be linked to the expression of the virulent phenotype of L. pneumophila. Furthermore, some regulatory proteins (FleQ and FliA) have been shown to actually be fitness factors (15–17). FliA is involved in the establishment of infections, cytotoxicity to bone marrow-derived macrophages (BMMs), and the in vivo fitness of L. pneumophila (15, 17–19). The fliA regulon contains several genes that do not belong to the flagellar regulon (20). Therefore, the presence of FliA and FlgM in pathogenic nonflagellated Legionella strains makes them suitable for investigation of the putative FliA-dependent virulence genes in detail. In L. oakridgensis, in contrast to L. pneumophila, deletion of fliA did not lead to a malfunction of the entire flagellar regulon (9, 20) and caused fewer (unspecific) feedback effects. Thus, it can be assumed that L. oakridgensis is a suitable species for investigation of the virulence-associated role of FliA.

The activity of FliA is, in general (Salmonella enterica, Escherichia coli, Vibrio cholerae, Pseudomonas aeruginosa), regulated on the protein level through the interaction with FlgM (21–25). By binding FliA, the anti-sigma-28 factor FlgM prevents the DNA binding of FliA to promoters with a sigma-28 consensus sequence. When the basal body of the flagellum is assembled, FlgM is secreted through this structure itself and FliA-dependent genes (e.g., the flagellin gene) are expressed (24–28). L. oakridgensis does not contain a basal body; thus, the regulation of FlgM must consequently be different. For Helicobacter pylori, it has been found that the FlgM protein is inactivated instead of being secreted (29).

The aims of this study were to investigate in detail the role of FliA in nonflagellated Legionella strain L. oakridgensis, to identify target genes of FliA, and to investigate the putative (regulatory) function of FlgM in modulating FliA activity.

RESULTS

The fliA gene of L. oakridgensis.

We started by analysis of the FliA protein in silico. The L. oakridgensis fliA gene (loa2161) codes for a protein of 241 amino acids with a predicted molecular mass of 26.9 kDa. The amino acid sequence of FliA of L. oakridgensis contains a FliA domain and the Pfam_Sigma-70_r2–4 superfamily domains that are generally involved in the interaction of a sigma factor with the core polymerase (Fig. 1A). This indicates that FliA may be a putative sigma-28 factor. The gene contains also putative −10 and −35 promoter sequence binding sites which are similar to the respective sites of the FliA protein of L. pneumophila Corby (Fig. 1A, underlined). However, the overall identities are 74% and 80% to FliA of L. pneumophila Corby and L. hackelia (Lha_2252), respectively. The fliA gene exhibits a sigma-70 promoter sequence (Fig. 1B, underlined) containing a putative discriminator site (Fig. 1B, line above sequence) (DksA binding) and a typical ribosome binding site 5 bp in front of the start codon (GTG).

FIG 1.

Amino acid sequence comparison of the FliA proteins of L. oakridgensis (241 amino acids [aa]; 26.9 kDa) and L. pneumophila Corby (238 aa) and the nucleotide sequence of the promoter region of the fliA gene of L. oakridgensis. (A) Amino acids involved in the recognition of the promoter consensus DNA sequences (−10 and −35) are shown and underlined. (B) A putative sigma-70 promoter sequence, the ribosome binding site (AAGGAGA), and the start codon (GTG) of fliA of L. oakridgensis are highlighted (underlined). The line above the DNA sequence indicates a putative discriminator site.

Functional characterization of FliA.

To determine if the fliA gene of L. oakridgensis encodes a functional sigma-28 factor, we used the fliA gene of L. oakridgensis (pfliA_Loa) to complement the defect in swarming of an E. coli fliA mutant (YK4104). Strains YK4104 (pfliA_Loa) and YK4104 (pfliA_Lpc [pKH27], coding for FliA of L. pneumophila Corby) were motile when inoculated into motility medium, while the vector control in the same background was immobile (Fig. 2). Complementation with pfliA_Loa was not able to restore the motility to the level seen using fliA of L. pneumophila Corby (15, 30). Nevertheless, the results showed that fliA of L. oakridgensis encodes a functional sigma-28 factor. To further investigate FliA of L. oakridgensis, we generated a ΔfliA mutant strain of L. oakridgensis ATCC 33761 by replacing fliA with a kanamycin resistance cassette (for details, please see Materials and Methods).

FIG 2.

Complementation of a fliA mutant strain of E. coli performed by using the cloned fliA gene of L. oakridgensis. Strains were inoculated on motility agar and incubated overnight at 30°C. Shown are motile E. coli wild-type strain YK410 harboring the pBCSK vector (YK410 pBCSK), the isogenic fliA mutant (YK4104 pBCSK), and the isogenic ΔfliA mutant complemented with the cloned fliA gene of L. pneumophila Corby (YK4104 pfliA_Lpc) and of L. oakridgensis (YK4104 pfliA_Loa).

To confirm that fliA of L. oakridgensis encodes a functional sigma-28 factor, we additionally investigated the ability of FliA to initiate the transcription of the FlaA promoter of L. pneumophila Corby. This promoter exhibits a FliA-binding consensus sequence that is known to mediate FliA-dependent expression in L. pneumophila Corby (8, 30, 31). The FliA-dependent transcription was determined using a reporter gene construct (luciferase) by measuring the luciferase activity. The FliA-dependent transcription was determined in the L. oakridgensis wild-type (WT) strain and the corresponding isogenic fliA mutant strain that has got either plasmid pKH23 (pFlaA-luxAB fusion) or plasmid pKH24 (promoterless luxAB construct) (31). Only the wild-type strain that harbored the pFlaA-luxAB fusion showed luciferase activity (Fig. 3A). In the FliA mutant strain, no luciferase activity was detected. The experiment demonstrated that FliA of L. oakridgensis encodes a functional sigma-28 factor which was able to initiate direct transcription from a promoter that contained a FliA-binding consensus sequence.

FIG 3.

Effects of FliA, temperature, and growth phase on the expression of a pflaA-reporter gene fusion in L. oakridgensis. (A) Expression of luciferase in the L. oakridgensis wild type (Loa WT) or the isogenic ΔfliA mutant (Loa ΔfliA) harboring the flaA promoter-luxAB fusion (flaA gene of L. pneumophila Corby) (Loa WT pKH23) or a promoterless flaA-luxAB fusion (Loa WT pKH24), respectively (left), and the respective agar plate (right). (B) Growth curve of the L. oakridgensis wild-type strain (Loa WT) and isogenic ΔfliA mutant strain harboring the flaA promoter-lacZ fusion (placZ = pKH12) grown in AYE medium at 30°C (left) and 37°C (right) with the indicated time points of sampling. (C) To determine the effects of growth phase and growth temperature on the expression of β-galactosidase from a fliA-dependent promoter, cells were grown in AYE medium and harvested at the exponential (E), postexponential (PE), and stationary (S) phases of growth (see also panel B) and β-galactosidase activity was measured and is given in Miller units. Error bars indicate the mean standard deviations of results of three independent experiments.

As the FliA activity was successfully measured using a pflaA reporter gene fusion, further investigations of FliA were started using the pflaA-lacZ fusion (pKH12) which was used earlier to investigate FliA-dependent expression of the flagellin gene in L. pneumophila Corby (31). The levels of expression of lacZ in strains L. oakridgensis (pKH12) and L. oakridgensis ΔfliA (pKH12) at the exponential (E), postexponential (PE), and stationary (S) phases of growth at 37°C and 30°C were additionally investigated. The growth curves and time points of sample collection are given in Fig. 3B. First, by using the lacZ gene as a reporter, we corroborated that there was no FliA activity detectable in the fliA mutant strain of L. oakridgensis, regardless of growth temperature or growth phase (Fig. 3C). Second, the activity of FliA is growth phase dependent; there was an ∼16-fold change (FC) in activity with respect to induction from the exponential to the stationary phase at 30°C. The main induction seems to have occurred after the postexponential phase of growth. Furthermore, FliA in L. oakridgensis is more active with respect to growth at 30°C than at 37°C (Fig. 3C). Interestingly, the activity of FliA in L. oakridgensis is thus similar to the activity of FliA in L. pneumophila (31), although L. oakridgensis does not exhibit a flagellum or a flagellar regulon.

The fliA mutant showed slightly reduced fitness.

Another issue was whether FliA in L. oakridgensis is also involved in the fitness of this strain, as demonstrated for the flagellated L. pneumophila strains (15, 17–19). We have recently shown that L. oakridgensis has the ability to multiply within amoebae (Acanthamoeba lenticulata) (4). Therefore, the ΔfliA mutant strain was investigated here to determine its ability to multiply within A. lenticulata and human macrophage-like U937 cells. It was found that the ΔfliA mutant strain had a reduced capacity to multiply in three different A. lenticulata strains but that the magnitude of this effect showed some variation (Fig. 4, Al 45, Al 118, and Al 72/2). However, no growth defect was detectable by using the human macrophage-like U937 cell line (Fig. 4, U937). Nevertheless, the results indicated that FliA is also necessary for the full fitness of L. oakridgensis.

FIG 4.

Replication of the L. oakridgensis wild-type strain (Loa Wt), ΔfliA mutant strains (ΔfliA clones 1, 2, 9, and 10), and a glycerol-3-phosphate dehydrogenase mutant (ΔgpsA) in infection assays performed with A. lenticulata strains ATCC 50703 (Al 45), ATCC 50706 (Al 118), and ATCC 50704 (Al 72/2) and a human macrophage-like cell line (U937). The ΔgpsA mutant was used as a control strain known to be unable to replicate in A. lenticulata (4). Cells were infected with bacteria at a multiplicity of infection (MOI) of 1 (Al 72/2, MOI = 5), washed after 2 h, and incubated for 4 days at 37°C. The number of CFU per well was determined by plating on BCYE agar plates. Data are means and standard deviations of results from duplicate samples and are representative of at least three independent experiments. Statistically significant differences in the levels of growth of ΔfliA mutant strain clones 2 and 9 compared with the wild-type strain (determined by Student's t test; *, P < 0.05; ***, P < 0.001) are indicated. Loa wt, L. oakridgensis wild type; cl., clones.

Identification of target genes of FliA.

As the ΔfliA mutant strain showed reduced intracellular replication in amoeba, we were interested in identifying putative target genes of FliA in L. oakridgensis. Therefore, we performed mRNA deep sequencing of samples of L. oakridgensis and its isogenic ΔfliA mutant strain at the exponential and postexponential phases of growth. The time points of RNA isolation are indicated in Fig. S1 in the supplemental material. Genes highly downregulated in the ΔfliA mutant strain of L. oakridgensis at PE in comparison to the WT strain are given in Table 1. The genes most highly downregulated were loa1456 (FC, 0.04) and loa0818 (FC 0.08). Gene loa1456 encodes a putative anti-sigma-28 factor (FlgM), and loa0818 encodes a hypothetical protein of 91 amino acids. Genes loa1456, loa0818, and loa2818 seemed to be induced in the PE phase in the wild-type strain (Table 1, WT PE/E ratio), which corresponds to the observed activity of FliA (see above). This was also true for loa2264, although we could not detect a sigma-28 consensus sequence in front of the gene (see below). Homologs of three of the fliA target genes (loa1456, loa0818, and loa2818) have been reported to also be fliA dependent in flagellated L. pneumophila strain Paris (14). A consensus sequence of fliA-dependent genes of L. pneumophila Corby was characterized earlier (see above) (8), and we screened the upstream region of FliA-dependent genes of L. oakridgensis for the presence of a similar consensus sequence. For five genes, we successfully identified a sigma-28 consensus promoter sequence (Fig. 5). This indicates that we had successfully identified at least five putative target genes of FliA of L. oakridgensis.

TABLE 1.

FliA target genes of L. oakridgensis^a

Gene	ΔfliA mutant/WT (PE) ratio	WT PE/E ratio	FliA-Pr.	Homolog	Function(s)
loa1456	0.04	7.9	+	lpp0969*	FlgM, anti-sigma28 factor
loa0818#	0.08	3.1	+	lpp2282*	HP
loa0928	0.15	0.5	+	lpp0941	HP
loa2818	0.20	2.9	+	lpp2998*	Pfam_YCCV-like, HspQ-like, protein turnover
loa2516#	0.44	1.1	+	n.p.	SP, Pfam_AMP binding
loa2199	0.11	0.5	−	n.p.	SP
loa2512	0.15	0.5	−	n.p.	HP, CheY-like receiver
loa1679#	0.15	1.6	−	lpp1496	Pfam_HPPK (FolK)
loa2264**	0.23	6.8	−	n.p.	SP, glycine zipper, putative surface antigen

^aFliA-Pr., binding of FliA to promoter; n.p., not present; HP, hypothetical protein; SP, secretion signal sequence; *, FliA dependent and also expressed in L. pneumophila Paris (Lpp) (20) (Table S7); #, probably part of an operon (loa0818, operon loa0818-0819*; loa2516, operon loa2516-2517; loa1679, operon loa1678-1679); **, highly expressed in the PE phase.

FIG 5.

Putative FliA-dependent promoter sequences. §, FliA-dependent promoter sequence of flaA (flagellin gene) of L. pnemophila Corby (8, 30). (−10) and (−35), −10 and −35 promoter sequence binding sites; RBS, ribosome binding site.

We used gene-specific DNA-kanamycin resistance (Km^r) cassette constructs (see Materials and Methods) to generate mutant strains of genes flgM (loa1426), loa0818, and loa2818 of L. oakridgensis. We were successful in generating mutant strains with deletion of genes loa0818 and loa2818 (Table 1), but the two mutant strains of A. lenticulata grew similarly to the wild-type strain (data not shown). However, we had not yet been successful in generating an flgM mutant strain.

FlgM of L. oakridgensis is a negative regulator of FliA activity.

Interestingly, the most downregulated gene in the fliA mutant strain was a putative homolog of FlgM, an anti-sigma-28 factor. Thus, we were interested in further characterization of this gene. flgM in L. oakridgensis codes for a protein of 92 amino acids (10.5 kDa) that is only 28% identical to FlgM of L. pneumophila Corby (Fig. 6A). The protein exhibits a Pfam-FlgM domain, a putative ArfGAP domain (putative GTPase activating proteins for the small GTPase ADP-ribosylation factor [ARF]), and a C-terminal region with similarity to those of members of the Structural Classification of Proteins (SCOP)-proteasome activator super family (glavo.1). The N-terminal parts of FlgM proteins of L. oakridgensis, L. adelaidensis, and L. londiniensis are shorter than those of FlgM proteins of L. longbeachae and L. pneumophila Corby (Fig. 6A). First, we determined if FlgM of L. oakridgensis is able to block the activity of FliA, a known general function of bacterial FlgM proteins. Therefore, we cloned the flgM gene into vector pBCSK and the resulting plasmid (pflgM) was then transformed into E. coli strain YK410 (see above) and L. pneumophila Corby. The resulting E. coli strain and the corresponding E. coli WT strain were inoculated into motility agar and incubated for 10 h at 37°C. FlgM of L. oakridgensis significantly reduced the swarming ability of E. coli (Fig. 6B) and also the expression of the flagellin (FlaA) in L. pneumophila Corby (Fig. 6C, Lpc pflgM_Loa). The results demonstrate that FlgM of L. oakridgensis is a negative regulator of FlaA expression and indicate that this may be due to the interaction of FlgM with FliA, since the corresponding motif (Pfam_FlgM) was identified by in silico analysis of FlgM.

FIG 6.

Functional analysis of FlgM of L. oakridgensis. (A) Amino acid sequence comparison of proteins encoded by the L. oakridgensis (Loa), L. adelaidensis (Lad), L. londiniensis (Lon), L. longbeachae (Llo), L. sainthelensi (Lsa), L. drancourtii (Ldr), L. pneumophila Corby (Lpc), L. hackeliae (Lha), and H. pylori (Hpy) flgM genes. Symbols in the Loa sequence are as follows: line above, Pfam-FlgM; dotted line, SCOP-proteasome activator SF (glavo.1); underlined, peptide successfuly used for anti-FlgM antiserum production. (B) Strains were inoculated into motility agar and incubated for 10 h at 30°C (left). Data represent motile E. coli wild-type strain YK410 harboring vector pBCSK (YK410 pBCSK) or the flgM gene of L. oakridgensis (YK410 pflgM). After 10 h of incubation, the swarming radius of the cells was measured (right). The error bars represent mean standard deviations of results from 5 experiments, and the inhibition of swarming activity by pFlgM was statistically significant (Student's test; P < 0.05). (C) Western blot analysis using an anti-FlaA antiserum. Lpc WT, L. pneumophila Corby; Lpc pBSC SK, L. pneumophila Corby vector control; Lpc pflgM_Loa, L. pneumophila Corby harboring FlgM of L. oakridgensis; Lpc ΔfliA, L. pneumophila Corby isogenic ΔfliA mutant; Lpc ΔfliA pfliA_Loa, L. pneumophila Corby ΔfliA mutant harboring FliA of L. oakridgensis; FlaA, flagellin.

Degradation of FlgM in L. oakridgensis is growth phase dependent.

In general, FlgM is secreted from the cell, leading to the activation of FliA. As L. oakridgensis does not exhibit a basal body (4), the protein may be secreted by another secretion system or it may be inactivated instead of being secreted, as was reported for Helicobacter pylori (29). A further hint regarding this hypothesis was the identification of a C-terminal region with similarity to those corresponding to proteasome activator proteins (see above). For further analysis, we generated an anti-FlgM antiserum using a peptide of the last 19 amino acids of FlgM (see Fig. 6A and Materials and Methods). We then used this antiserum to analyze the growth-phase-dependent presence of FlgM at growth temperatures of 30°C and 37°C. The same cells (cell lysates) were also used for the experiments mentioned above and shown in Fig. 3C. Results are shown in Fig. 7. The antiserum positively detected a protein of approximately 10 kDa (FlgM) and a second protein of about 37 kDa (Fig. 7A). The 37-kDa protein was also present in the ΔfliA mutant of L. oakridgensis (Fig. 7A, lane 5). We were able to corroborate the identity of the 10-kDa protein as FlgM by the expression of the flgM gene in E. coli (Fig. 7A, lane 7). Furthermore, the results demonstrated that the amount of FlgM was decreasing during growth (Fig. 7B) and with increasing temperature (from 30°C to 37°C) (Fig. 7B). In addition, we were not able to detect FlgM in the supernatant during growth of L. oakridgensis (Fig. 7B, supernatant). These results indicate that in L. oakridgensis, the degradation of FlgM, instead of its secretion into the medium, leads to the activation of FliA.

FIG 7.

Western blot analysis of FlgM expression in L. oakridgensis using an anti-FlgM antiserum. (A) Western blot of cell extracts of L. oakridgensis (Loa WT) or its isogenic ΔfliA mutant (Loa ΔfliA) grown to the exponential (E), postexponential (PE), or stationary (S) phase of growth and of E. coli (DH5α pBCSK, vector control) and E. coli expressing FlgM of L. oakridgensis (DH5α pflgM_Loa). (B) Western blot of whole-cell lysates of L. oakridgensis harboring pKH12 grown at 30°C or 37°C and of supernatant of L. oakridgensis harboring plasmid pKH12. Samples were harvested at the exponential (E), postexponential (PE), and stationary (S) phases of growth (Fig. 3B), and equal amounts of bacterial lysates were loaded onto the gel (see Materials and Methods). The supernatant was 25-fold concentrated, and 15 μl was loaded onto the gel.

Recently, a phylogenetic tree of the Legionella genus based on whole-genome sequences was published (32) that showed that L. oakridgensis, L. adelaidensis, and L. londiniensis cluster in a seperate clade. Therefore, we analyzed the published genomes of L. adelaidensis and L. londiniensis for the presence of flagellar genes and found that genes of the flagellar regulon are also missing in L. adelaidensis and L. londiniensis (data not shown), but we found homologs of FliA exhibiting 80% and 39% identity to FliA of L. oakridgensis and homologs of FlgM exhibiting 84% and 48% identity to FlgM of L. oakridgensis, respectively (Fig. 6A and Fig. S2). Earlier, we showed that L. israeliensis is also negative for the flagellin gene (8), and here we were able to corroborate this finding by analyzing the published genome sequence (32). We were also able to identify a homolog of FliA within the genome sequence of L. israeliensis (72% identity to FliA of L. oakridgensis) (Fig. S2). However, we have not yet been able to identify a homolog of FlgM in the genome sequence of L. israeliensis.

DISCUSSION

L. oakridgensis is one of the known unflagellated Legionella species that does not need additional l-cysteine for growth. In addition, the bacterium is known to lack a flagellar regulon but to be positive for FliA and its anti-sigma factor FlgM (2, 4, 8). The species is furthermore pathogenic for humans (3) and able to replicate inside human macrophage-like cell lines (5, 6, 33) as it does inside amoebae (4). However, L. oakridgensis is less virulent then L. pneumophila strains, and its ability to replicate inside amoebae is decreased in comparison to that to L. pneumophila strains (4, 6). Here we investigated the flagellar regulon and the function of FliA (9) that is needed for full fitness of L. pneumophila (15, 17–19). Therefore, we also investigated the contribution of FliA to other functions not associated with motility or the flagellar regulon. Thus, L. oakridgensis was chosen to address these issues.

It was demonstrated that the fliA gene of L. oakridgensis encodes a functional alternative sigma-28 factor which is able to complement the motility defect of a ΔfliA mutant strain of E. coli. Moreover, the flagellin expression of the ΔfliA mutant strain of L. pneumophila Corby (Fig. 6C, last lane) was restored. It is not surprising that the interspecies complementation of motility (E. coli), depending on the coordinated function of 40 to 50 genes, is not complete, as this had already been reported elsewhere (30). Furthermore, we demonstrated that FliA of L. oakridgensis binds to the sigma-28 consensus sequence of L. pneumophila Corby, inducing transcription of the downstream reporter gene. However, the overall identity of the two FliA proteins is only 74%, but the amino acid regions known to be responsible for the binding to the −10 region of the promoter are 100% identical in the two proteins.

In addition, by using a ΔfliA mutant strain of L. oakridgensis, we were able to demonstrate that the activity of this promoter is specifically FliA dependent, as also shown for L. pneumophila Corby (30). Moreover, although L. oakridgensis does not exhibit a flagellar regulon, the growth phase- and temperature-dependent activity of FliA is similar to that seen with the flagellated L. pneumophila Corby strain (31). High expression of the FliA-dependent promoter was observed at 30°C and at the PE phase of growth for both Legionella species, although no basal body or flagellum is present in L. oakridgensis (see also below).

Interestingly, L. adelaidensis and L. londiniensis, which form a separate phylogenetic cluster with L. oakridgensis (32), are also flagellar regulon negative and still exhibit the presence of fliA and flgM genes within their genomes. Since L. longbeachae and L. hackeliae are also nonflagellated and negative for most of the genes of the flagellar regulon (8, 10, 34), representing another major clade of the Legionella genus, the loss of genes of the flagellar regulon did not take place as a single event during the evolution of legionellae.

In addition, we were able to show that the ΔfliA mutant strain exhibited a slightly reduced capacity of intracellular replication in A. lenticulata. Thereby, the magnitude of the described effect showed variations. This result may have been related to the host amoeba used. A. lenticulata is known not to be the ideal host for L. oakridgensis (4). However, no other amoeba is known to support the growth of L. oakridgensis. The defect is not visible using a human macrophage-like cell line, in which L. oakridgensis replicates very well, as host (4, 6). Additionally, we tried to identify target genes of FliA by the help of mRNA deep sequencing. Five of the identified target genes exhibited a sigma-28 consensus sequence, and three of them (loa1456, loa0818, and loa2818) were induced in the PE phase in the wild-type strain, confirming the FliA-dependent expression. These three target genes (Table 1) were also detected by an analogous screen in L. pneumophila Paris (20). The most highly repressed gene identified was the flgM gene, coding for a putative anti-sigma-28 factor. Two identified targets, the operon loa0818-loa0819 (protein-binding PDZ superfamily; Zn metallo-proteinase domain) and the gene loa2818 (YccV-like domain; heat shock protein HspQ-like), code for proteins putatively involved in the degradation of proteins. Yet we do not know which of these target genes are responsible for the observed phenotype of the ΔfliA mutant. We generated specific mutant strains of FliA target genes loa0818 and loa2818; nevertheless, the analysis demonstrated that both A. lenticulata mutants replicated similarly to the wild-type strain. Further investigations are ongoing to generate a flgM mutant strain of L. oakridgensis.

As L. oakridgensis encodes a putative anti-sigma-28 factor exhibiting a Pfam-FlgM motif, we were not surprised that FlgM of L. oakridgensis was able to prevent expression of FlaA when overexpressed in L. pneumophila Corby and that it reduced the motility of E. coli. This indicates that FlgM is indeed a functional negative regulator of FliA activity, and it may indicate that FlgM binds to FliA as shown in other bacteria (25, 26, 28) and as illustrated in a schematic drawing (Fig. 8). For example, the activity of FliA in S. enterica serovar Typhimurium is negatively regulated by FlgM. The protein prevents the association of FliA with the RNA polymerase core by binding to it (24, 25). Thus, in general, FlgM binds to FliA and thereby prevents FliA-dependent transcription. When the flagellar basal body is assembled, FlgM is secreted through the complex and FliA is then ready to start the expression of FlaA (FliC) and of proteins encoded by other genes exhibiting a sigma-28 consensus sequence (21, 22, 35). However, there is no basal body present in L. oakridgensis (4). For H. pylori, it was published that the bacterium exhibits an unusual FlgM which is not secreted and that the interaction of FlgM with the basal body may determine the turnover of functional FlgM (29, 36). Our Western blot analysis using an anti-FlgM antiserum demonstrated that the FlgM protein in L. oakridgensis seems not to be secreted into the medium. It seems that the protein is degraded in a growth phase-dependent manner in parallel with increasing activity of FliA (Fig. 8). The proteasome of L. oakridgensis is probably involved in this process, as FlgM possesses a C-terminal region exhibiting a putative proteasome activation sequence. Figure 6A shows that FlgM of the phylogenetically related species L. adelaidensis and L. londiniensis exhibited a similar C-terminal amino acid sequence. In addition, the anti-FlgM antiserum identified a further protein band at ∼35 kDa. This band is not associated with the presence of FliA and seems to represent growth-phase-dependent degradation (Fig. 7A). So far, we have not accumulated further information about this protein, and since we were not successful in generating an flgM mutant strain, we could not determine if this was the consequence of a cross-reaction of the antiserum or if the identified protein band is associated with the presence of FlgM.

FIG 8.

Schematic overview of proposed growth-phase-dependent interactions among FliA, FlgM, and the proteasome. During the exponential-growth phase (E), FlgM binds FliA, preventing it from interacting with the sigma-28 promoter, which leads to a block of transcription of FliA-dependent genes (loa0818). In the postexponential phase (PE), FlgM seems to be degraded by the proteasome, freeing FliA to bind to the promoter and to induce transcription of loa0818.

Together, the experiments pointed out that in L. oakridgensis, FlgM seems not to be secreted and its degradation leads to the activation of FliA (Fig. 8). Interestingly, FlgM itself is expressed in a FliA-dependent manner, as shown for L. pneumophila Paris and H. pylori (20, 36).

MATERIALS AND METHODS

Strains and growth conditions.

The following Legionella strains were used in this study: L. pneumophila Corby (37) and its isogenic ΔfliA mutant strain (15); a cooling tower isolate, L. oakridgensis ATCC 33761 (2); and the ΔgpsA transposon mutant strain (4). Bacteria were cultivated either in N-(2-acetamido)-2-aminoethanesulfonic acid (ACES)-buffered yeast extract (AYE) broth (1% ACES, 1% yeast extract, 0.04% l-cysteine, and 0.025% ferric pyrophosphate, adjusted to pH 6.8 with 3 M KOH and subjected to sterile filtration) or on ACES-buffered charcoal-yeast extract (BCYE) agar plates at 37°C (38). The Escherichia coli strains used in this study were E. coli DH5α and E. coli YK410 (fliA⁺) and YK4104 (fliA mutant) (39). E. coli was grown in Luria-Bertani (LB) medium or on LB agar plates at 37°C for 1 day. When needed, the media were supplemented with antibiotics at final concentrations suitable for L. pneumophila or E. coli as follows: kanamycin at 8 or 40 μg/ml and chloramphenicol at 8 or 40 μg/ml, respectively; ampicillin at 100 μg/ml for E. coli. For the cultivation of L. pneumophila on agar plates, the final concentrations of kanamycin and chloramphenicol were 12.5 μg/ml and 20 μg/ml, respectively.

Acanthamoeba lenticulata 45 ATCC 50703, A. lenticulata 118 ATCC 50706 and ATCC 50704 (growth medium Chang; 1% casein hydrolysate, 0.1325% Na₂HPO₄, 0.25% glucose, 0.08% KH₂PO₄, 0.5% yeast extract, 0.02% penicillin, 0.02% streptomycin, 10% fetal calf serum [FCS]), and the U937 human macrophage-like ATCC CRL-1593.2 cell line (growth medium RPMI 1640–10% FCS [purchased from PAA, Pasching, Austria]) were used to investigate the multiplication of L. oakridgensis mutant strains. All amoebae were grown at room temperature, whereas the U937 cell line was cultivated at 37°C and 5% CO₂.

DNA techniques and mutant strain construction.

Genomic and plasmid DNAs were prepared accordingly to standard protocols and manufacturers' instructions (40). All PCRs were carried out using TRIO-Thermoblock (Biometra, Göttingen, Germany), and amplification was performed with a Taq DNA polymerase (Qiagen, Hilden, Germany). Vector pBCSK (Agilent Genomics, CA) was used for cloning of fliA and flgM genes of L. oakridgensis, resulting in plasmids pfliA_Loa and pflgM, respectively. All plasmids used in this study are listed in Table 2. Foreign DNA was introduced into bacterial strains by electroporation using a Gene Pulser system (Bio-Rad, Munich, Germany) according to the manufacturer's specifications. E. coli or L. pneumophila strains were electroporated at 1.7 or 2.3 kV, 100 Ω, and 25 mF, respectively. Oligonucleotides were obtained from Eurofins MWG Operon (Ebersberg, Germany). Restriction enzymes were purchased from New England BioLabs (Frankfurt a.M., Germany).

TABLE 2.

Oligonucleotide and plasmid sequences

Oligonucleotide or plasmid	Sequence (5′–3′) or descriptiona	Source or reference
Oligonucleotides
iLoa_FliA_1F	CCGCGGGAATTCGATATCGCCAGTGTTGGATCATGAAACC	This study
iLoa_FliA-2R	TAGAAGCTGACATTCATCAAACACTCGTCAATGCTATGGTTTA	This study
iLoa_FliA-3F	TGACGAGTGTTTGATGAATGTCAGCTTCTAGACTATCTGG	This study
iLoa_FliA-4R	ACAATGTGCTTCCTTACAATTCAGGCGGCCATCGTGTCTA	This study
iLoa_FliA-5F	TAGACACGATGGCCGCCTGAATTGTAAGGAAGCACATTGT	This study
iLoa_FliA-6R	GAATTCACTAGTGATATCGTATGACGACGATTCCCTGAATC	This study
FliA_1U	GCCAGTGTTGGATCATGAACC	This study
FliA-2R	GTATGACGACATTCCCTGAATC	This study
Mut_Test-5U	CTATCGTTGGTTGGCGCCG	This study
Mut_Test-6R	CCATAAACTGCCGGCAAGCG	This study
Loa_flgM_Mut1U	TCCAGTGATCATGAATGACGGTACCT	This study
Loa_flgM_Mut2R	ACAAGTGGTTATCGCCGGTATCGAA	This study
Loa0818_KmR_U	TTATACGTTTGCTGGCATTAAATACG	This study
Loa0818_KmR_R	TAGCCGGCCTGTAAAAAAGACTT	This study
M13U	GTAAAACGACGGCCAGT	41
M13R	GGAAACAGCTATGACCATG	41
Loa0818_Test_1U	CAGGCTGCTTGTCGTAAACCG	This study
Loa0818_Test_2R	GCCGTTCTTGAAGCTTATGAAGAG	This study
Loa2818_T1Uneu	CGGTACGTTCGTCTGGTA	This study
Loa2818Test_3R	GATGCTGAGATCCTGAAG	This study
Plasmids
pKH12	pMMB207 containing the pflaA-lacZ fusion	31
pKH14	pMMB207 containing the promoterless lacZ gene	31
pKH23	pMMB07 cntaining the pflaA-luxAB fusion	30
pKH24	pMMB207 containing the promotorless luxAB gene	30
pKH27	pUC containing the complete fliA gene of L. pneumophila Corby	30
pESfliA K3	fliA::Kmr in fusion construct (3.4 kb)	This study
pfliALoa	pBCSK containing the complete fliA gene of L. oakridgensis	This study
pflgM	pBCSK containing the complete flgM gene of L. oakridgensis	This study
LoaflgM-GmR	flgM::Gmr synthetic construct (4.4 kb)	This study
Loa0818-KmR	loa0818::Kmr synthetic construct (2.4 kb)	This study
Loa2818-KmR_Neu	loa2818::Kmr synthetic construct (3.4 kb)	This study

^aKm^r, kanamycin resistance; Gm^r, gentamicin resistance.

The Legionella loa2161 (fliA) knockout mutant of L. oakridgensis was constructed using an In-Fusion cloning kit (TaKaRa Clontech) according to the instructions of the manufacturers. To generate the construct for natural transformation, regions of 1,000 bp flanking the loa2161 gene and a kanamycin cassette were amplified by PCR. The amplification of the flanking regions (primer pairs iLoak_FliA-1F/iLoak_FliA-2R and iLoak_FliA-5F/iLoak_FliA-6R) was done with chromosomal DNA from L. oakridgensis. The kanamycin cassette was amplified from Tn1732 (42) using primers iLoak_FliA-3F/iLoak_FliA-4R. The primers were constructed with an overlap according to the instructions of the In-Fusion manual. The cloning enhancer-treated fragments were fused with pGEMTeasy open vector (Promega) and transformed into stellar competent cells (TaKaRa Clontech). Afterward, the cells were plated on LB plates containing kanamycin for selection. PCR amplification was used to identify colonies carrying the plasmids with flanking regions surrounding the kanamycin cassette in pGEMTeasy vector (control primers FliA-1F/FliA-2R). The plasmid pESfliAK3 was confirmed by sequencing the insertion DNA and used for the amplification of the kanamycin cassette with the flanking regions (primers M13F/M13R). The amplified and purified PCR product (3.4 kb) was used for two independent natural transformations of the L. oakridgensis WT strain as described below. The successful generation of the L. oakridgensis Δ2161 mutants was confirmed via PCR (primers FliAmut-Test-5F/FliAmut-Test-6R). Two independent mutant sets (clones 1 and 2 and clones 9 and 10) were generated. All primers are listed in Table 2.

PCR amplification cycling conditions were set to activation at 95°C for 5 min followed by 35 cycles of 94°C for 15 s, 60°C for 1 min, and 68°C for 8 min and a final extension step of 72°C for 10 min.

Constructs for natural transformation of the Legionella loa1456 (flgM), loa0818, and loa2818 knockout mutants of L. oakridgensis were synthetically generated by in vitro DNA synthesis by GeneCust (GeneCust, Dudelang, Luxembourg). For L. oakridgensis ΔflgM, the construct was composed of a 1,200-bp flanking region of loa1456 and a gentamicin (Gm) cassette instead of loa1456. The construct for the loaΔ0818 mutant included regions of 650 bp and 715 bp flanking the gene up- and downstream, respectively, and a kanamycin cassette. For the loaΔ2818 mutant, regions of 1,200 bp flanking the loa2818 gene and a kanamycin cassette were used. Sequence verifications for plasmids LoaflgM-GmR, Loa0818-KmR, and Loa2818-KmR_New by DNA sequencing and following cloning of the DNA fragments into multiple cloning sites of pUC57, pUC-SP, and pUC57, respectively, were performed by GeneCust (GeneCust, Dudelang, Luxembourg). For amplification of the kanamycin or gentamicin cassette with the flanking regions, primer pairs Loa_flgM_Mut1U/Loa_flgM_Mut2R, Loa0818_KmR_U/Loa0818_KmR_R, and M13U/M13R were used. Amplified and purified PCR products (4.4 kb, 2.4 kb, and 3.4 kb) were used for natural transformation of the L. oakridgensis WT as described below. The successful generation of LoaΔ0818 mutants was confirmed using primers Loa0818_Test_1U/Loa0818_Test_2R. For LoaΔ2818 mutants, primers Loa2818_T1Uneu and Loa2818Test_3R were used. The respective primers are listed in Table 2.

Natural transformation of L. oakridgensis was performed as previously described (43) with slight modifications. In brief, in a plastic tube, 3 ml of fresh medium was inoculated with 200 μl of L. oakridgensis overnight culture. When the cultures reached the exponential-growth phase, the PCR product was added and the cultures were incubated for 3 days at 37°C. Subsequently, bacteria were grown on antibiotic selective media for 4 additional days at 37°C and 5% CO₂. The screening for positive Km^r mutants (or Gm^r mutants for gene loa1456) obtained by homologous recombination was performed by PCR.

RNA isolation, mRNA deep sequencing, and mapping.

Bacterial growth in broth was monitored through the determination of the optical density at 600 nm (OD₆₀₀) of the culture using a Thermo Scientific Genesys 10 Bio spectrophotometer (VWR, Darmstadt, Germany). Growth phases were defined as follows: OD₆₀₀ of 0.8 to 1.0 corresponded to the exponential (E) growth phase, OD₆₀₀ of ∼1.5 to the postexponential (PE) growth phase, and OD₆₀₀ of ∼1.8 to the stationary (S) growth phase.

RNA preparation was performed as previously described (16) with slight modifications. In brief, bacteria were grown to a specific optical density at 600 nm (see above) and harvested, and total RNA was extracted using a High Pure RNA isolation kit (Roche) following the instructions of the manufacturer. The purified RNA was incubated with 5 μl of DNase I (RNase-Free DNase set; Qiagen, Hilden, Germany) at 37°C for 20 min and then repurified using an RNeasy minikit (Qiagen). Using standard PCR protocols (see above) and primers specific for the rpoN gene, the preparation was screened for possible chromosomal DNA contamination. If DNA was detected within the sample, the DNase I treatment procedure was repeated. RNA was isolated from the L. oakridgensis WT strain and the L. oakridgensis ΔfliA mutant strain at the E and PE phases of growth from three independent experiments for each strain and growth phase, respectively.

Reverse transcription-PCR (RT-PCR) experiments were performed using a OneStep RT-PCR kit (Qiagen) with gene-specific primers according to the manufacturer's instructions (Table 2). The RT reaction was performed at 50°C for 30 min with 10 ng of total RNA. PCR amplification was performed in the same tube using each primer at a concentration of 0.5 μM, deoxynucleoside triphosphates at a final concentration of 400 μM, 5× OneStep RT-PCR buffer containing 12.5 mM MgCl₂, and 1 μl of OneStep RT-PCR enzyme mixture in a final volume of 25 μl. After the reverse transcription, initial denaturation was performed at 95°C for 15 min (activation step), and final extension was performed at 72°C for 10 min. The cycling conditions were the following: 94°C for 1 min, 55°C for 1.5 min, and 72°C for 1 min. Depending on the respective gene products, different numbers of cycles (24, 27, and 30 cycles) were chosen (data not shown).

Equal amounts of total RNA from three independent experiments (for each strain and growth phase, respectively) were pooled, and the mRNA deep sequencing and mapping were performed by Eurofins (Ebersberg, Germany) using an Illumina Hiseq 2000 sequencer and HiSeq control software v. 1.4.8, RTA 1.12.4.2 software, and CASAVA 1.8.2 software. The genome sequence of L. oakridgensis was used as the reference sequence for the mapping.

Swarm assays and reporter gene activity.

The swarm assays were performed using stabbing of fresh bacteria onto semisolid agar plates (10 g of Bacto tryptone, 5 g of Bacto yeast extract, and 2 g of Bacto agar per liter) that were incubated either at 30°C overnight or at 37°C for 10 h. During growth of the strains in AYE medium, β-galactosidase activity was measured by quantifying the hydrolysis of o-nitrophenyl-β-d-galactopyranoside (ONPG) as described by Miller (44), and the results are given in Miller units (MU). The FlaA promoter fusion to the luxAB gene (pKH23) (30) and to the lacZ gene (pKH12) (31) was used to characterize FliA-dependent expression and functionality of FliA of L. oakridgensis.

Intracellular multiplication of L. oakridgensis in amoebae.

Infection assays using A. lenticulata for in vivo growth of L. pneumophila Corby and L. oakridgensis were performed as described previously (45, 46) with modifications. In brief, the 3-day-old amoeba cultures were washed in Acanthamoeba buffer (Ac buffer; 4 mM MgSO₄ · 7 H₂O, 0.4 M CaCl₂ · 2 H₂O, 0.1% sodium citrate dihydrate, 0.05 mM Fe(NH₄)₂(SO₄)₂ · 6 H₂O, 2.5 mM NaH₂PO₄, 2.5 mM K₂HPO₄) and the concentration was adjusted to 1 × 10⁵ cells per ml and further incubated in 24-well plates for 2 h at 37°C and 5% CO₂. Stationary-phase Legionella bacteria, grown for 3 days on BCYE agar, were also diluted in AC buffer and mixed with the amoeba solution at a multiplicity of infection (MOI) of 1 (if not otherwise stated). After bacterial invasion for 2 h at 37°C, the amoeba cell layer was washed with AC buffer, defining the start point of the time course experiment. The CFU of the legionellae was determined after 24, 48, 72, and 96 h by plating on BCYE agar. Unless otherwise stated, each infection was carried out twice and was repeated at least three times.

Antibody production.

We generated an anti-FlgM antiserum using two peptides (DLNTLKNMDAKSPRQMLDA-C and C-RIAQKLTEYNQSMKEPEIA) of FlgM of L. oakridgensis (localization within FlgM [Fig. 6A]) to immunize two rabbits with each. Peptides were conjugated to keyhole limpet hemocyanin (KLH) at the C terminus. The peptide synthesis and immunization were performed by LifeTein LLC (Hillsborough, NJ). Obtained sera were antigen affinity purified. The serum derived from rabbits by the use of the second peptide (C terminus of FlgM) was reactive to the FlgM protein of L. oakridgensis.

SDS-PAGE and immunoblotting.

Flagellin and FlgM detection was carried out by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting. The SDS-PAGE assay was performed as described previously (47). Legionella grown on BCYE agar plates or in AYE medium to the respective growth phases were diluted in phosphate-buffered saline (PBS) and adjusted to an OD₆₀₀ of 1. Equal amounts of these lysates were boiled for 10 min in Laemmli buffer. A total of 15 μl of the solution was loaded onto a 12% SDS polyacrylamide gel. Western blotting was carried out using polyclonal anti-FlaA or anti-FlgM antisera diluted in 1% milk–Tris-buffered saline (TBS) (1:1,000 or 1:100, respectively) (17). A horseradish peroxidase-conjugated goat anti-rabbit antibody was used as secondary antibody (1:1,000). FlaA was visualized by incubation of the blot with 50 ml color reaction solution (47 ml TBS, 3 ml 4-chloro-1-naphthol, 80 μl H₂O₂), and the reaction was stopped with distilled water. FlgM was visualized using ECL Western blotting substrate (Thermo Scientific) and X-ray film. Results were determined in at least two independent experiments.

Accession number(s).

The RNA-seq data have been entered into the European Nucleotide Archive (ENA) under study accession number PRJEB19613 and sample accession numbers ERS1568799, ERS1568800, ERS1568801, and ERS1568802.