Proteomic Analysis of Growth Phase-Dependent Expression of Legionella pneumophila Proteins Which Involves Regulation of Bacterial Virulence Traits

Legionella pneumophila, which is a causative pathogen of Legionnaires' disease, expresses its virulent traits in response to growth conditions. In particular, it is known to become virulent at a post-exponential phase in vitro culture. In this study, we performed a proteomic analysis of differences in expression between the exponential phase and post-exponential phase to identify candidates associated with L. pneumophila virulence using 2-Dimentional Fluorescence Difference Gel Electrophoresis (2D-DIGE) combined with Matrix-Assisted Laser Desorption/Ionization–Mass Spectrometry (MALDI-TOF-MS). Of 68 identified proteins that significantly differed in expression between the two growth phases, 64 were up-regulated at a post-exponential phase. The up-regulated proteins included enzymes related to glycolysis, ketone body biogenesis and poly-3-hydroxybutyrate (PHB) biogenesis, suggesting that L. pneumophila may utilize sugars and lipids as energy sources, when amino acids become scarce. Proteins related to motility (flagella components and twitching motility-associated proteins) were also up-regulated, predicting that they enhance infectivity of the bacteria in host cells under certain conditions. Furthermore, 9 up-regulated proteins of unknown function were found. Two of them were identified as novel bacterial factors associated with hemolysis of sheep red blood cells (SRBCs). Another 2 were found to be translocated into macrophages via the Icm/Dot type IV secretion apparatus as effector candidates in a reporter assay with Bordetella pertussis adenylate cyclase. The study will be helpful for virulent analysis of L. pneumophila from the viewpoint of physiological or metabolic modulation dependent on growth phase.


Introduction
Legionella pneumophila is a causative agent of Legionnaires' disease that replicates in macrophages in humans [1]. In nature, the bacteria reside and replicate in protozoa [2]. With the inhalation of L. pneumophila-contaminated aerosols, the bacteria can invade the human body and be phagocytosed by phagocytic cells, such as alveolar macrophages. The nascent phagosomes supporting the intracellular survival of L. pneumophila represent unique forms in that they are associated with small vesicles, mitochondria and the rough endoplasmic reticulum [3,4]. L. pneumophila replicates vigorously within this unique compartment, evading lysosomal fusion [5]. After it has replicated enough in phagosomes, the bacteria lyse the macrophage membrane, and infect new host cells. As this bacterial infectious cycle expands, it becomes pathogenic to humans.
The Icm/Dot type IV secretion apparatus is known as a major virulence factor of L. pneumophila [6,7]. The icm/dot system consists of 26 genes located in two separate regions of the genome. Icm/Dot delivers effector proteins into host cells, forming the unique phagosomes in which bacteria can survive and replicate [8]. Recently, many L. pneumophila effector proteins have been identified using several techniques including the Cya (B. pertussis adenylate cyclase toxin) assay system, yeast expression system and b-lactamase reporter system [9][10][11][12][13][14][15][16]. So far, over 140 substrates of Icm/Dot have been identified, but the single mutation of most of these proteins has only minor effects on bacterial growth in host cells, except for LepAB which promotes nonlytic release from protozoa, SidJ which contributes to the trafficking of ER proteins to phagosomes containing bacteria, SdhA which protects macrophages from apoptotic cell death in the early stages of infection, and AnkB which promotes the decoration of the Legionella-containing phgosomes (LCPs) with polyubiquitinated proteins to establish a favorable replicative niche [9,[15][16][17][18][19][20].
Similar to other bacterial pathogens, L. pneumophila has a biphasic life cycle: one is a replicative form in which bacteria multiply in host cells, and another is a transmissive form in which they escape from infected cells and infect new host cells [21]. In broth culture, L. pneumophila at a post-exponential phase when the supply of nutrients become limited, not at an exponential phase in which the bacteria vigorously replicate, exhibit transmissive phenotypes (stress resistance, cytotoxicity for macrophages, high motility and evasion of bacteria-containing phagosomes from lysosomal fusion) [22]. Therefore, the L. pneumophila biphasic life cycle can be roughly modelled in vitro broth culture, which regards the bacterial form in the exponential phase and post-exponential phase as the replicative form and transmissive form during the infection, respectively [21]. To date, studies have identified virulent phenotypes of this pathogen in post-exponential phase [21,22], the stringent response enzymes RelA and SpoT which monitor bacterial amino acid levels and fatty acid biosynthesis, respectively [23][24][25], and RpoS, LetA/S, CsrA, and small noncoding RNAs RsmY and RsmZ as regulators of the virulent phenotypes in the post-exponential phase [26][27][28][29][30]. However, a comprehensive analysis of L. pneumophila proteins that show growth phase-dependent expression has not been attempted. The proteomic approach is applicable to a comprehensive analysis of bacterial virulence factors, because fluorescence 2-D DIGE can analyze multiple proteins in a single gel and has greater sensitivity, reproducibility and quantitative accuracy than conventional 2dimensional gel electrophoresis .
In this paper, we report the result of a proteomic analysis of the growth phase-dependent expression of L. pneumophila in vitro culture. Using 2-D DIGE and MALDI-TOF-MS, we identified 68 protein species which significantly differ in expression between the exponential phase and post-exponential phase. Most of the identified proteins were up-regulated at the post-exponential phase, including metabolic enzymes and proteins related to motility. Moreover, 9 uncharacterized proteins up-regulated at the post-exponential phase were found. The genes encoding 6 of these proteins (lpg0634, lpg0901, lpg1851, lpg2275, lpg2678, and lpg2874) were specific to L. pneumophila in several Legionella species, but the single knockout of these 6 genes did not influence bacterial intracellular replication within U937 macrophages and Acanthamoeba polyphaga. It was shown that Lpg0634 and Lpg0901 partially contributed to hemolysis of SRBCs. And Lpg1851 and Lpg2874 were found to be translocated into macrophages via the Icm/Dot secretion apparatus, suggesting them to be the effector candidates.

Bacterial strains, plasmids, primers and cell culture
The Legionella strains used in this study are listed in Table S1. The plasmids and primers are listed in Table S2 and Table S3, respectively. The Legionella strains were cultured on charcoal-yeast extract (CYE) agar plates or ACES-buffered yeast extract (AYE) broth with appropriate antibiotics as needed. The human monocytic cell line U937 [31] was maintained in RPMI1640 medium (Sigma, Tokyo, Japan) supplemented with 10% heatinactivated FBS (Hyclone Laboratories, Inc., U.S.A.). At 48 h prior to infection, the U937 cells were induced to differentiate with 50 ng/ml of phorbol 12-myristate 13-acetate (Sigma). Axenic A. polyphaga was cultured as adherent cells in PYG medium [32]. All cells were maintained under a humidified atmosphere of 5% CO 2 and 95% air at 37uC, as described previously [31].

Sample preparation for 2-D DIGE
L. pneumophila Philadelphia 1 JR32 [33] was grown until the exponential phase (OD 600 = 1.6) or post-exponential phase (OD 600 = 3.8) in AYE broth, and the bacteria (7.5610 9 cells of each) were harvested by centrifugation at 7,0006g for 10 min at 4uC. The bacterial pellets were solubilized with lysis buffer (7 M urea, 2 M thiourea, 4% CHAPS, and 30 mM Tris-HCl (pH 8.5)), and centrifuged at 360,0006g for 1 h at 10uC to eliminate genomic DNA. The supernatants were dialysed in a 7 M urea/ 2 M thiourea solution using a PlusOne mini dialysis kit (GE Healthcare) at 15uC for 16 h. Protein concentrations were determined using the Bio-Rad protein assay kit (Bio-Rad Laboratories).

2-D DIGE
2-D DIGE was performed as described previously [34,35]. Briefly, 50 mg of bacterial protein extract cultured in the exponential phase or post-exponential phase was labeled with 400 pmol of either Cy3 or Cy5 (GE Healthcare). An internal standard was prepared by combining 25 mg of each bacterial protein extract in the exponential phase and the post-exponential phase, followed by labeling with 400 pmol of Cy2 (GE Healthcare). The protein labeling was achieved by incubation on ice in the dark for 30 min. The reaction was quenched by the addition of 10 mM lysine, followed by incubation on ice for a further 10 min. The labeled samples were combined and mixed with an equal volume of lysis buffer containing 2% DTT and adjusted to 450 mL with rehydration buffer (7 M urea, 2 M thiourea, 4% CHAPS, 0.2% DTT, and 0.002% bromophenol blue) and used for 2-DE. The sample solutions were then loaded on non-linear 24 cm IPG strips (pH 3-10) (GE Healthcare) and rehydrated at 20uC for 12 h. First-dimension isoelectric focusing was carried out using the IPGphor IEF system (GE Healthcare) for 9 to 11 h (a total of 45 to 60 kVh). The IPG strips were equilibrated with equilibration buffer (6 M urea, 30% glycerol, 2% SDS, and 50 mM Tris-HCl (pH 8.8)) containing 10 mg/mL of DTT for 15 min, followed by equilibration buffer containing 25 mg/mL of iodoacetamide for a further 15 min. The equilibrated strips were loaded on 10% or 12.5% polyaclylamide gels and the gels were electrophoresed using the Ettan DALT 6 unit (GE Healthcare).

Image Analysis and spot picking
The 2-D gel images were scanned using a Typhoon 9410 scanner (GE Healthcare) by selecting adequate excitation and emission wavelengths (Cy2, excitation 488 nm, emission 522 nm; Cy3, excitation 532 nm, emission 580 nm; Cy5, excitation 633 nm, emission 670 nm). The statistical analysis was performed with Decyder software (version 5.02; GE Healthcare). Only those spots with over a 1.5-fold change in volume between two populations, with p values less than 0.05 (p,0.05) in Student's t-test for the variance of these ratios for each protein pair across three independent gels, were defined as significantly different, as described previously [34,35]. For the picking of protein spots of interest, pooled whole protein (500 mg) was separated by 2-DE, and stained with Deep Purple dye (GE Healthcare). The gel was imaged with a Typhoon 9410 scanner using excitation/emission wavelengths of 532/560 nm. The protein spots of interest were picked with an Ettan Spotpicker (GE Helthcare).

In-gel digestion and MALDI-TOF-MS analysis
The gel pieces were destained with 50 mM ammonium biscarbonate/50% acetonitrile and dehydrated with acetonitrile. The dried gel pieces were digested with 2.5% trypsin (Promega) in 100 mM ammonium biscarbonate overnight at 37uC. The peptide solutions were desalted and concentrated using m-C18 Zip Tip (Millipore, Bedford, MA, U.S.A.). The samples were mixed with a-cyano-4-hydroxycinnamic acid (CHCA) matrix and applied onto a target plate. MALDI-TOF-MS was performed using Ultraflex (Bruker Daltonics). Protein identification was carried out with Mascot software against the sequence databases of NCBInr (The National Center for Biotechnology Information nonredundant).

Southern blot analysis
The Southern blot analysis was performed as described previously [31]. Briefly, chromosomal DNAs of Legionella strains were digested with EcoRI, separated by agarose gel electrophoresis, and transferred to Hybond-N+ (GE Helthcare). Parts of each candidate gene were amplified by PCR with the following primer pairs (Table S3): lpg0390-F and lpg0390-R for lpg0390, lpg0483-F and lpg0483-R for lpg0483, lpg0584-F and lpg0584-R for lpg0584, lpg0634-F and lpg0634-R for lpg0634, lpg0901-F and lpg0901-R for lpg0901, lpg0992-F and lpg0992-R for lpg0992, lpg1647-F and lpg1647-R for lpg1647, lpg1851-F and lpg1851-R for lpg1851, lpg2275-F and lpg2275-R for lpg2275, lpg2678-F and lpg2678-R for lpg2678, and lpg2874-F and lpg2874-R for lpg2874, and were used as probes. Labelling of DNA probes and detection of signals were performed using the AlkPhos Direct Labelling and Detection System with CDP-Star (GE Helthcare, RPN3690).

Construction of L. pneumophila mutants and their complemented strains
The insertion mutants were constructed in JR32 using allelic exchange, according to a published procedure [36,37]. Briefly, part of each candidate gene was amplified by PCR with the following primer pairs (Table S3): lpg0634-F and lpg0634-R for lpg0634, lpg0901-F and lpg0901-R for lpg0901, lpg1851-F and lpg1851-R for lpg1851, lpg2275-F and lpg2275-R for lpg2275, lpg2678-F and lpg2678-R for lpg2678, and lpg2874-F and lpg2874-R for lpg2874, and ligated to the pGEMH-T Vector System (Promega). The kanamycin resistance cassette (kan-cassette) aphA-3 was cloned into the restriction sites (Eco47 III: lpg0634, Eco47 III: lpg0901, BsrG I: lpg1851, PshA I: lpg2275, BssH II: lpg2678, Bgl II: lpg2874) in the uncharacterized protein-coding genes. These plasmids (pMN21, pMN22, pMN23, pMN24, pMN25, and pMN26) were digested with Not I or Xba I, and the resultant fragments were cloned into the Not I or Xba I site of the allelic exchange vector pLAW344 [36]. These plasmids (pMN31, pMN32, pMN33, pMN34, pMN35, and pMN36) were introduced into L. pneumophila JR32 by electroporation with the Gene pulserH II system (BIO-RAD). Then, the bacteria were grown in AYE for 8 h, and plated on CYE plates containing kanamycin. Kanamycin-resistant (Km r ) colonies were patched onto CYE plates containing sucrose and CYE plates containing chloramphenicol. Then, Km r , sucrose-resistant and chloramphenicol-sensitive isolates in which the loss of pLAW344 by a double-crossover mutation occurred, were selected. The isolates (MN101, MN102, MN103, MN104, MN105, and MN106) of L. pneumophila mutants were analyzed by genomic Southern blotting to confirm that the correct allelic change occurred on the chromosome (data not shown).
Complimented strains of MN101 (JR32 lpg0634::aphA-3) and MN102 (JR32 lpg0901::aphA-3) were constructed by introduction of expression vector producing N-terminal M45 epitope tagged-Lpg0634 and Lpg0901, respectively. Coding region of lpg0634 or lpg0901 was amplified by PCR with the following primer pairs (Table S3): lpg0634-F2 and lpg0634-R2 for lpg0634, lpg0901-F2 and lpg0901-R2 for lpg0901. The PCR products were cloned into the pGEMH-T Vector System to generate pTH11 and pTH12. The Bam HI/Sph I fragments from pTH11 and pTH12 were ligated into Bam HI/Sph I fragments of pM45-ralF [38], which can produce M45 epitope-tagged RalF, kindly provided by Dr Hiroki Nagai (Osaka University, Japan). This procedure replaced M45 epitope-tagged RalF by the epitope-tagged Lpg 0634 and Lpg0901. Thus pTH21 and pTH22 for expression of the epitope-tagged Lpg0634 and Lpg0901, respectively, were produced. These plasmids were introduced into the corresponding L. pneumophila mutants by electroporation. The resultant strains were designated as TH101 and TH102. The expression of fusion proteins in these constructs was assessed by immunoblotting using an affinity-purified rabbit polyclonal antibody raised against the M45 epitope (MDRSRDRLPPFETETRIL) [39].

Intracellular growth assays
Intracellular growth assays were performed as described previously [31]. Briefly, differentiated U937 cells (1610 5 cells per well in 96-well plates (Falcon 353072; Becton Dickinson, Franklin Lakes, NJ, U.S.A.)) or A. polyphaga (1610 5 cells per well in 96-well plates) were infected with L. pneumophila strains cultured in the post-exponential phase at a multiplicity of infection (MOI) of 0.5 or 10, respectively. The plate was spun at 2506g for 20 min to synchronize the infection, and the time point at the end of this centrifugation was designated as 0 h. The infected cells were incubated for 1 h at 37uC, and washed three times with culture medium. The cells were incubated for 1 h at 37uC in culture medium containing 50 mg/mL of gentamicin to kill the extracellular bacteria. They were washed again, and then incubated in culture medium for the indicated period. At the appropriate time points, the cells were lysed with distilled water, diluted serially in distilled water, and plated on CYE plates to measure the number of colony forming units (CFUs) per well.

Construction of L. pneumophila producing Cya-fused proteins and producing Cya-fused RalF
Cya-fused proteins were constructed using pcya-ralF [38] that can express the Cya-RalF fusion protein kindly provided by Dr Hiroki Nagai. Coding region of lpg1851 and lpg2874 was amplified by PCR with the following primer pairs (Table S3): lpg1851-F2 and lpg1851-R2 for lpg1851, and lpg2874-F2 and lpg2874-R2 for lpg2874. The PCR products were cloned into the pGEMH-T Vector System to generate pTH13 and pTH14. The Bam HI/Sph I fragment from pTH12 which contain lpg0901, pTH13, and pTH14 was ligated into a Bam HI/Sph I fragment of pcya-ralF to generate pTH23, pTH24, and pTH25, respectively. This procedure was aiming at the production of M45 epitope-tagged Cya-fusion proteins with Lpg0901, Lpg1851 and Lpg2874, respectively. These plasmids were introduced into the JR32 or LELA3118 strain by electroporation. Construction of the fused proteins was assessed by immunoblotting using an affinity-purified rabbit polyclonal antibody raised against the M45 epitope [39].
Contact-dependent hemolysis assay using SRBCs As described previously [31], contact-dependent pore formation in membranes was determined by examining the hemolysis of SRBCs in contact with L. pneumophila at a MOI of 10 for 1 h.

Cya reporter assay
The Cya reporter assay was performed as described previously [10]. Briefly, U937 cells (3610 5 cells per well) were cultured and induced to differentiate in the wells of a 24-well culture plate (Costar 3526, Corning Incorporated, Amsterdam, Netherlands).
The differentiated U937 cells were infected with JR32 (MOI = 50) and LELA3118 (MOI = 100) containing different Cya-fused proteins or Cya-fused RalF. The plate was spun at 2506g for 10 min to synchronize the infection, then incubated for 1 h at 37uC. After removal of the supernatant, the infected cells were lysed in lysis buffer 1B provided in the cAMP Biotrak EIA System (GE Helthcare, RPN225), and cAMP levels were determined as instructed by the manufacturer.

Results and Discussion
Protein expression profiles of L. pneumophila depends on growth phase For the proteomic analysis of the growth phase-dependent expression of L. pneumophila whole-cell proteins, 2-D DIGE was performed three times. Representative images of 12.5% polyaclylamide gels are shown in Fig. 1A. To separate proteins with high molecular masses, we used a 10% polyaclylamide gel in other experiments. On this gel, SidE family proteins already known to be substrates of the Icm/Dot system [40] were clearly detected (Fig. 1B). Among 2145 spots detected on the 2-D gel, 105 were found to significantly differ in protein expression between the exponential phase and post-exponential phase. Among these, 82 spots were successfully identified as corresponding 71 protein species (Table S4), and the identified proteins were functionally classified into the category of metabolic enzyme, motility, substrate of the Icm/Dot type IV secretion system, transcriptional regulator, chaperone, toxin production, or unknown function (Table S5).

Identified proteins
Enzymes. Approximately 60% of the identified proteins were defined as enzymes. Most of the identified enzymes were upregulated at the post-exponential phase, and categorized as involved in carbohydrate metabolism, amino acid metabolism, and lipid metabolism. Moreover, the identified enzymes were sorted according to KEGG pathway maps of L. pneumophila Philadelphia-1 available at http://www.genome.jp/kegg/ [41]. (Fig. 2). The data showed that enzymes related to glycolysis (pyrubate kinase II (YP_094190) and malate dehydrogenase (YP_096964)), the TCA cycle (2-oxoglutarate ferredoxin oxidoreductase beta subunit (YP_094982)), ketone biogenesis (HMG-CoA lyase (YP_095856) and acetoacectate decarboxylase (ADC) (YP_094708)) and poly-3-hydroxybutyrate (PHB) biogenesis (acetoacetyl CoA reductase (YP_095092), acetyoacetyl CoA reductase (YP_094601, YP_094602), 3-hydroxyisobutyryl Coenzyme A hydrolase (YP_094904)) were up-regulated at the post-exponential phase. This is the first report that the expression of enzymes related to carbohydrate or lipid metabolism in Legionella species are upregulated at a post-exponential phase. L. pneumophila is thought to utilize only amino acids as carbon and energy sources [42], but a study of the whole-genome sequence study reported that this bacterium has genes encoding enzymes involved in carbohydrate metabolism, lipid metabolism, and the TCA cycle [43,44]. Moreover, it is known that L. pneumophila utilizes PHB as an energy source for survival in nutrient-poor environments such as tap water [45]. Therefore, L. pneumophila may utilize sugars and lipids as energy sources at the post-exponential phase or within host cells in which amino acids are scarce.

Motility-associated proteins
We identified two flagellar components (FliG:YP_095785, FlgA:YP_094942) and the type IV pili (Tfp) component PilT (YP_095020). FliG is a motor switch protein of flagella that regulates bacterial motility, and Salmonella typhimurium FliG is essential for the assembly, rotation, and switching of the flagellar motor [46]. And bacterial motility and flagella promote the infectivity of bacteria and macrophage cell death [22,47,48]. Two flagellar component proteins were up-regulated at the post-exponential phase, consistent with a previous report that bacterial motility was elevated at the post-exponential phase in vitro [22]. PilT is involved in Tfp-mediated motility, and Neisseria gonorrhoeae PilT is essential for Tfp-mediated twitching motility [49]. Recently, the presence of type IV pili-mediated twitching motility in L. pneumophila has been shown [50]. Originally, it has been shown that type IV pili of L. pneumophila is essential for attachment to mammalian and protozoan cells and associated with natural competence for DNA transfer [51,52]. Therefore, It is suggested that these 3 identified proteins associated with bacterial motility may function as virulence-associated factors [53]. Fig. 1B shows 2D-DIGE gel images of 3 substrates of the Icm/ Dot secretion system (SidE, SdeA and SdeC) cultured in the exponential phase and post-exponential phase. These proteins were identified as proteins transfered between bacterial cells, and SidE and SdeA were identified as proteins known to be translocated into macrophases via the Icm/Dot secretion apparatus [10,40,54]. As noted in Table S4, SdeA (Fig. 1B, spot no. 77) and SdeC (Fig. 1B, spot no. 77, 78, 79), which are paralogous to SidE, were up-regulated at the post-exponential phase, consistent with previous results [10]. Interestingly, in contrast to SdeA and SdeC, it was found that SidE (Fig. 1B, spot no. 80, 81, 82) was down-regulated in post-exponential phase. Many Icm/Dot substrates were up-regulated at the postexponential phase in vitro [10,[54][55][56], but translocating proteins up-regulated at the exponential phase should exist because of their diverse functions throughout the infection process. Therefore, we predict that SidE participates in different stages of bacterial infection, which differ from the case of translocating proteins such as SdeA and SdeC which are up-regulated at the post-exponential phase.

Uncharacterized proteins
We focused on the uncharacterized proteins which have no homology with any known proteins. In total, 12 proteins were annotated as hypothetical proteins. Eleven of these were upregulated, and one was down-regulated, at the post-exponential phase (Table S4, spot no. 51). Among the up-regulated proteins, it has been reported that Lpg0390 and Lpg0483 were identified as VipA and LegA12, respectively [11,12]. VipA is a substrate of Icm/Dot that interferes with organelle trafficking in yeast. Therefore, VipA may play a role in manipulating the host secretion pathway in macrophages [11]. LegA12 is an eukaryotelike protein that contains ankyrin repeats [12]. Eukaryotic ankyrin proteins are thought to act as a link between membrane proteins and the cytoskeleton [44]. Ankyrin repeats-containing protein may modify host cell function. Pan et al. identified ankyrin repeatscontaining proteins (AnkW, AnkX, AnkY, AnkZ) translocated into host cells, and revealed that AnkX prevented microtubedependent maturation of bacteria-containing phagosomes [57]. And Price et al. have recently shown that AnkB mimics the action of host cell F-box proteins promoting the decoration of LCPs with polyubiquitinated proteins to establish a intracellular replicative niche [20]. Table 1 lists 9 uncharacterized proteins up-regulated at the post-exponential phase in this study, except Lpg0390 and Lpg0483.
We performed homology research based on NCBI proteinprotien BLAST (Basic Local Alignment Search Tool). Five proteins had homology in other bacterial species. Lpg0584 and Lpg1647 showed high homology to PhoU (identity; 52%) that correlates with phosphate transport regulator [58] and YceI reported to be a base-induced periplasmic protein in Escherichia coli [59], respectively. The other three revealed ,35% homology to certain proteins; Lpg2678 had homology to UbiE, a methylase involved in ubiquinone/menaquinone biosynthesis [60]. Lpg0901 had homology to the hypothetical protein NMA0899 in Neisseria meningitides. Lpg2275 was homologous with the hypothetical Figure 3. Genomic southern blotting of the genes encoding uncharacterized proteins in L. pneumophila and other Legionella species. Bacterial genomic DNA was digested with EcoR I and hybridized with probes against lpg0390, lpg0483, lpg0584, lpg0634, lpg0901, lpg0992, lpg1647, lpg1851, lpg2275, lpg2678, and lpg2874, respectively. Eight genes (lpg0390, lpg0483, lpg0634, lpg0901, lpg1851, lpg2275, lpg2678, and lpg2874) are specific for L. pneumophila among several Legionella species, but 3 genes (lpg0584, lpg0992, and lpg1647) were detectable in other Legionella species except L. pneumophila. doi:10.1371/journal.pone.0011718.g003 proteins CBU0952 and FTT0975 in the intracellular pathogens Coxiella burnetii and Francisella tularensis, respectively, speculating that these proteins might be originally transferred horizontally for parasitism within host cells. Lpg0992 had homology to hypothetical proteins in Congregibacter litoralis, Nitrosococcus oceani and Aromatoleum aromaticum. In contrast, Lpg0634, Lpg1851, and Lpg2874 had almost no homology with any proteins in other bacterial species. These 9 uncharacterized proteins up-regulated at the post-exponential phase are possible to be novel virulenceassociated factors of L. pneumophila, and we therefore focused on them.
Analysis of the genes encoding the uncharacterized proteins in different Legionella species Virulence factors specific to L. pneumophila are thought to exist because 90% of clinical isolates were L. pneumophila among Legionella species. [61]. Actually, previous studies reported that the effector protein RalF and intracellular growth factor PmiA were specific for L. pneumophila among several Legionella species [31,55,62]. Therefore, we examined whether 9 uncharacterized proteins, Lpg0390 (VipA) and Lpg0483 (LegA12) were specific to L. pneumophila.
A Southern blot analysis was performed to demonstrate the presence or absence of the genes for the uncharacterized proteins in several Legionella strains. The results showed that three genes (lpg0992, lpg0584, and lpg1647) were distributed among several Legionella species. In contrast, 8 genes (lpg0390, lpg0483, lpg0634, lpg0901 lpg1851, lpg2275, lpg2678, and lpg2874) were only detected in L. pneumophila (all strains and serogroups examined) and were not present in other Legionella species (Fig. 3). This result suggested that these genes may be specific for L. pneumophila. As mentioned above, proteins encoded by two (lpg0390 and lpg0483) of 8 genes were identified as VipA and LegA12, respectively [11,12]. Except for these two proteins (Lpg0390 and Lpg0483), we focused on the other 6 L. pneumophila specific genes encoding the uncharacterized proteins.
Six uncharacterized proteins (Lpg0634, Lpg0901, Lpg1851, Lpg2275, Lpg2678, and Lpg2874) are not directly required for bacterial replication within macrophages and protozoa, but Lpg0634 and Lpg0901 are partially associated with the hemolytic activity on SRBCs To explore the function of six uncharacterized proteins (Lpg0634, Lpg0901, Lpg1851, Lpg2275, Lpg2678, and Lpg2874) in macrophages and protozoa, we constructed single mutants of each six encoding genes by insertion of the kancassette, and tested these strains for intracellular growth within host cells. Six individual mutants (MN101, MN102, MN103, MN104, MN105 and MN106) were compared with a wild type, JR32, and an intracellular growth-deficient dotA mutant, LELA3118. The data showed that all mutants grew robustly within U937 cells and A. polyphaga in a time-dependent manner, similar to JR32, suggesting that the single inactivation of each of the genes did not influence bacterial growth within U937 macrophages and protozoa (Fig. 4). This result indicates that not all differentially expressed proteins play the critical role for intracellular replication and survival of bacteria. There may be synergetic effects among their proteins.
It is known that the pore-forming activity of L. pneumophila is not sufficient for or partially independent of their intracellular growth [63]. Therefore, we next examined the pore-forming activity of the mutants by conducting contact-dependent hemolysis assay using SRBCs. It was found that Lpg0634 and Lpg0901 are partially associated with hemolysis of SRBCs (Fig. 5). So far, IcmQ and IcmT are known as pore-forming factors of L. pneumophila [64][65][66][67]. The single knockout of each impedes intracellular bacterial replication dependent on the Icm/Dot system, suggesting that the pore-forming activity of IcmQ and IcmT would be probably involved in the function of Icm/Dot as the secretion apparatus. On the other hand, MN101 and MN102, which are single knockout mutants of lpg0634 and lpg0901 respectively, retain all of the intact icm/dot genes, however, the hemolytic activity on SRBCs

Lpg1851 and Lpg2874 are translocated into macrophages via Icm/Dot secretion apparatus
It is reported that many known effectors or Icm/Dot substrates localize in the bacterial cytoplasm and they are up-regulated at a post-exponential phase in broth culture [10,54,55], suggesting that translocation of many effectors into the host cells on an early stages of L. pneumophila infection would be necessary for causing specific phagocytosis of bacteria and beginning the formation of LCPs. Therefore, we investigated whether the uncharacterized proteins identified in our proteome analysis were translocated into the host cells in an Icm/Dot-dependent manner. We chose three (Lpg0901, Lpg1851 and Lpg2874) of the 9 uncharacterized proteins, because it is predicted on Legionella Genome Project database that these proteins locate in the bacterial cytoplasm and do not have similarity to other known bacterial proteins, as shown in Table 1.
To determine whether Lpg0901, Lpg1851 and Lpg2874 are translocated into host cells, we used a reporter assay system that takes biochemical advantage of the adenylate cyclase domain (Cya) of Bordetella pertussis. Cya is activated by calmodulin as a cofactor in the eukaryotic cell cytosol to synthesize cAMP from ATP. Therefore, the translocation of Cya-fused proteins was detected by monitoring the accumulation of cAMP in infected cells [68].
We constructed forms of each of 3 uncharacterized proteins (Lpg0901, Lpg1851 and Lpg2874) fused with Cya, expressed in L. pneumophila (JR32 and LELA3118). Then, U937 cells were infected with L. pneumophila producing these Cya-fused proteins. After 1 h, levels of cAMP on infection with the wild type JR32 strain producing Cya-Lpg1851 and Cya-Lpg2874 were over 100000 fmol/well (Fig. 6). The level of cAMP resembled that on infection with the JR32 strain producing a Cya-fused form of RalF, an effector protein previously shown to be translocated into host cells, as a positive control [38]. In contrast, when U937 cells were infected with the dotA mutant LELA3118 strain producing The hemolytic activity of L. pneumophila MN101 (lpg0634 2 ), MN102 (lpg0901 2 ), TH101 (lpg0634 2 /lpg0634 + ) and TH102 (lpg0901 2 /lpg0901 + ) was tested by contact-dependent hemolysis of SRBCs. Bacterial contact with SRBCs was performed at a MOI of 10 for 1 h. Wild type JR32 and LELA3118 (dotA 2 ) strains were used as positive and negative controls, respectively. PBS (2) was also used as a negative control for the absence of bacteria. The experiment was done in triplicate, and the error bars represent standard deviations. The data shown are representative of at least two independent experiments. Asterisks indicate statistically significant differences (* p,0.01, compared with JR32 samples by ANOVA followed by Dunnett's test). Sharp indicates statistically significant differences (# p,0.01 by Student t test). doi:10.1371/journal.pone.0011718.g005 any Cya-fused proteins, the levels of cAMP were similar to that on infection with JR32 producing only Cya or uninfected U937 as a negative control. However, cAMP on infection with JR32 producing Cya-fused Lpg0901 was similar in level to that on infection with LELA3118 producing any Cya-fused proteins (Fig. 6A). Therefore, this result clearly showed that Lpg1851 and Lpg2874 are translocated into U937 cells via the Icm/Dot secretion apparatus. Regarding the translocation of Lpg1851, it has recently been reported elsewhere [15]. As mentioned above, we showed in this study that the inactivation of the genes encoding these 2 proteins had no influence on the intracellular growth of L. pneumophila in U937 macrophage-like cells and A. polyphaga (Fig. 4). However, these results are not surprising, because effectors may have redundant or synergistic functions [8].
In the previous section, we described that Lpg0634 and Lpg0901 are partially involved in hemolysis of SRBCs. One might imagine that the partial defect of hemolytic activity of MN101 and MN102 toward SRBCs was caused by defects in the translocation of some effectors or in the construction of the translocation apparatus. Thus we constructed the MN101 and MN102 strains producing Cya-fused RalF, and we performed the Cya reporter assay as well. As a result, it was shown that MN101 and MN102 can normally translocate the RalF to macrophages and the ability of effector translocation for these mutants is not impaired (Fig. 6B). Therefore, it is likely that the partial defect of hemolytic activity for MN101 and MN102 is independent of effector translocation and the function of the Icm/Dot system.

Concluding remarks
In this study, we comprehensively identified proteins upregulated in the post-exponential phase in which L. pneumophila shows virulent phenotypes. They included proteins related to sugar utilization, motility, and Icm/Dot-substrates. Regarding high motility of L. pneumophila in the post-exponetial phase, previous papers reported an increase of FlaA expression [69][70][71]. It was confirmed in this study that the expression of several flagella and pili components increased as well. The up-regulation of proteins related to sugar utilization would indicate that the proteins are functional. In addition to circumstances in which the supply of amino acid becomes limited, L. pneumophila might use a pathway of sugar metabolization within intracellular environments in which the source of nutrients is restricted. On the screening of L. pneumophila Icm/Dot substrates based on a comparison of specific amino acid sequences in silico, it had been estimated that there exist ,100 substrates [72]. Recently, a large number of (,140) Icm/ Dot substrates have been identified, and some of which were characterized as effectors [8,15]. RalF and DrrA/SidM function as a guanine nucleotide exchange factor of ARF, leading to the interception of host vesicle transport and establishment of bacterial replication-permissive phagosomes [13,55,73]. SidF neutralizes the proapoptotic function of BNIP3 and Bcl-rambo, which is followed by the prevention of macrophage death caused in an early stage of infection [74]. Although few Icm/Dot substrates have been characterized in terms of their biological function, it is supposed that almost all substrates should be injected into host cells to perform certain functions for permitting bacteria replication intracellularlly. Further functional analyses of two Icm/Dot/substrates, Lpg1851 and Lpg2874, identified in this study are indispensable.
Since the genome sequencing of L. pneumophila has been completed and the post-genome era for the research into this bacteria has begun, several researchers have performed comprehensive analyses with transcriptomics or proteomics [75][76][77][78]. In our study, proteomics was shown to be powerful tool for identifying potential virulence factors of L. pneumophila. A comprehensive analysis will help to determine the entire molecular network for L. pneumophila pathogensis.  (b) Lpg0634 and Lpg0901 are not involved in translocation of RalF into macrophages. U937 cells (3610 5 cells/well) were infected with wild type JR32 (MOI = 50) or LELA3118 (dotA 2 ) (MOI = 100) producing the indicated proteins fused with Cya or MN101 (lpg0634 2 ) or MN102 (lpg0901 2 ) producing RalF fused with Cya (MOI = 50) for 1 h. After infection, the cells were lysed and the cAMP levels were measured as described in experimental procedure. Asterisks indicate statistically significant differences (p,0.005, by Student t test) between JR32 and LELA3118 expressing the indicated proteins. Sharp indicates statistically significant differences (# p,0.05, compared with JR32 pCya-RalF by ANOVA followed by Dunnett's test). The experiments were performed in triplicate, and the data are shown as means6S.D. The data are representative of one (Cya-Lpg0901, Cya-RalF) or three (Cya-Lpg1851 and Cya-Lpg2874) independent experiments. NS: not significant. doi:10.1371/journal.pone.0011718.g006