A Pseudomonas aeruginosa EF-Hand Protein, EfhP (PA4107), Modulates Stress Responses and Virulence at High Calcium Concentration

Pseudomonas aeruginosa is a facultative human pathogen, and a major cause of nosocomial infections and severe chronic infections in endocarditis and in cystic fibrosis (CF) patients. Calcium (Ca2+) accumulates in pulmonary fluids of CF patients, and plays a role in the hyperinflamatory response to bacterial infection. Earlier we showed that P. aeruginosa responds to increased Ca2+ levels, primarily through the increased production of secreted virulence factors. Here we describe the role of putative Ca2+-binding protein, with an EF-hand domain, PA4107 (EfhP), in this response. Deletion mutations of efhP were generated in P. aeruginosa strain PAO1 and CF pulmonary isolate, strain FRD1. The lack of EfhP abolished the ability of P. aeruginosa PAO1 to maintain intracellular Ca2+ homeostasis. Quantitative high-resolution 2D-PAGE showed that the efhP deletion also affected the proteomes of both strains during growth with added Ca2+. The greatest proteome effects occurred when the pulmonary isolate was cultured in biofilms. Among the proteins that were significantly less abundant or absent in the mutant strains were proteins involved in iron acquisition, biosynthesis of pyocyanin, proteases, and stress response proteins. In support, the phenotypic responses of FRD1 ΔefhP showed that the mutant strain lost its ability to produce pyocyanin, developed less biofilm, and had decreased resistance to oxidative stress (H2O2) when cultured at high [Ca2+]. Furthermore, the mutant strain was unable to produce alginate when grown at high [Ca2+] and no iron. The effect of the ΔefhP mutations on virulence was determined in a lettuce model of infection. Growth of wild-type P. aeruginosa strains at high [Ca2+] causes an increased area of disease. In contrast, the lack of efhP prevented this Ca2+-induced increase in the diseased zone. The results indicate that EfhP is important for Ca2+ homeostasis and virulence of P. aeruginosa when it encounters host environments with high [Ca2+].


Introduction
Pseudomonas aeruginosa is a facultative pathogen and a leading cause of severe nosocomial infections in both immunocompetent and immunocompromised patients [1] [2], including patients in intensive care units. P. aeruginosa is one of the primary organisms that forms biofilms on airway mucosal epithelium of patients with cystic fibrosis (CF) where it contributes to airway blockage and cellular damage. P. aeruginosa also causes infective endocarditis and device-related infections with high morbidity and mortality rates [3], [4], [5]. P. aeruginosa biofilm infections are increasingly difficult to treat with traditional antibiotic therapy, and are often not eradicated by host defensive processes [6], [7].
Calcium (Ca 2+ ) is a well-known signaling molecule that regulates a number of essential processes in eukaryotes [8]. Slight abnormalities in cellular Ca 2+ homeostasis have been implicated in many human diseases, including diseases associated with bacterial infections, such as CF pulmonary infections and endocarditis. Ca 2+ metabolism is recognized to be central to the pathology of CF [9]. Ca 2+ is a part of a hyperinflamatory host response to bacterial infection, and accumulates in airway epithelia, pulmonary and nasal liquids of CF patients [10], [11]. There is growing evidence suggesting that Ca 2+ also plays a significant role in the physiology of certain bacteria, affecting maintenance of cell structure, motility, chemotaxis, cell division and differentiation, gene expression, transport, and spore formation [12], [13], [14], [15], [16]. Several bacteria including P. aeruginosa have been shown to maintain intracellular Ca 2+ at sub-micromolar levels and produce Ca 2+ transients in response to environmental and physiological factors [17][18][19]. However, molecular mechanisms of Ca 2+ regulation in prokaryotes are not well defined.
In eukaryotes, the regulatory effects of Ca 2+ are carried out by Ca 2+ -binding proteins (CaBPs), which may function as Ca 2+ sensors, signal transducers, Ca 2+ buffers or Ca 2+ -stabilized proteins. Prokaryotic genomes also encode CaBPs with different Ca 2+ -binding motifs, including the EF-hand motif [12,20], which typically consists of a Ca 2+ -binding loop flanked by two a-helices. Acidic amino acids in the loop are preferentially bound by Ca 2+ [21], and are responsible for Ca 2+ -induced conformational changes required for function [22]. Bacterial EF-hand proteins constitute a majority of all studied CaBPs [23], and include the Ca 2+ transducer calsymin CasA from Rhizobium etli [24], putative Ca 2+ buffers and stabilizers calerythrin from Saccharopolyspora erythraea [25] and CabC [26] CabB [27] from Streptomyces coelicolor, and Escherichia coli transglycosylase MltB [28]. Most bacterial proteins containing Ca 2+ -binding motifs are classified as hypothetical proteins with unknown physiological functions.
Previously, we showed that Ca 2+ modulates proteome profiles of P. aeruginosa, influences biofilm architecture, and alters production of several secreted virulence factors [29], [30]. Michiels et al [31] screened 31 bacterial genomes for CaBPs, and predicted that the P. aeruginosa hypothetical protein, PA4107, contains EF-hand Ca 2+ -binding motifs. We also identified PA4107 as having sequence similarity within the EF-hand motif to CasA of R. etli. Here, we further analyzed PA4107, which we designate EfhP (EF hand protein) and hypothesized that it plays roles in P. aeruginosa response to Ca 2+ . We generated deletion mutants lacking efhP in P. aeruginosa strain PAO1 and in the CF clinical isolate, FRD1, and studied their Ca 2+ -dependent phenotypes by global quantitative proteomics. Since certain stress response proteins had reduced abundances in the proteomes of the efhP deletion strains, we assayed the mutants for resistance to chemical and oxidative stresses. We also characterized the effect of the mutation on pyocyanin production and survival at low iron. Finally, we tested Ca 2+ -induced virulence of the mutant and wild-type strains in a plant infection model. The results indicate that EfhP is involved in plant virulence and in Ca 2+ -enhanced adaptation of P. aeruginosa to host environments.
For quantitative growth assays, cultures were pre-cultured in 5 ml tubes to mid-log phase, diluted to obtain optical density OD 600 of 0.3, and used to inoculate 100 ml of BMM medium in 250-ml flasks. Growth data are based on at least 3 biological replicates. Biofilm formation was quantified using abiotic solid surface assay in 96-well plates as described previously [35]. Prior to inoculation, overnight cultures were diluted to obtain equal optical densities. Dilutions 1:100 of the normalized cultures were inoculated into BMM medium and incubated for 24 h at 37 uC, the medium was replaced after 12 h. Nonattached cells were removed by washing the plates three times with saline solution (0.85% NaCl). Biofilms were stained with 1% crystal violet, nonbound crystal violet was removed by washing with water. Bound crystal violet was extracted with 80% ethanol/20% acetic acid, and the absorbance was measured at 595 nm using a microtiter plate reader (Tecan Instruments Inc.).
Cells from planktonic and biofilm cultures were obtained as described previously [30]. Cell pellets were washed twice with saline solution and resuspended in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0, containing 0.3 mg/ml of phenylmethylsulfonyl fluoride (PMSF)). Cells were disrupted by sonication (12 times for 10 s, 4W, 4uC), and the cell debris and unbroken cells were removed by centrifugation (12000 g, 60 min, 4uC). The protein concentrations of the supernatants were determined by using the modified Lowry assay (Pierce, Thermo Scientific). Protein samples were stored as aliquots at -80uC. Two-dimensional gel electrophoresis was performed as described previously [30]. Briefly, proteins were loaded by in gel rehydration into 18 cm immobilized pH gradient (IPG) strips with pH range of 4-7. Solubilization buffer consisted of 9 M urea, 2 M thiourea, 4% CHAPS, 2% w/v carrier ampholytes, 0.037 M DTT, and a trace amount of bromphenol blue. Isoelectric focusing was conducted using a Multiphor II (Pharmacia) at 20uC. Proteins were focused for a total of 28 kVh. Second dimension electrophoresis was carried out on 11% polyacrylamide gels (230620061 mm) using vertical Hoefer Dalt System (Pharmacia) at 10uC using 10 mA/gel for 4 h, then 40 mA/gel for 14-16 h. Proteins were detected by Colloidal Coomassie staining.

Proteomic Analysis
The gels were imaged, and the digital images were analyzed by using Progenesis Workstation Software (Nonlinear Dynamics, Durham, NC). A minimum of two biological replicates were performed for each condition, and the signal intensity of each spot was averaged over the replicates. The signal intensities (volumes) of protein spots were normalized against total signal intensity detected on a gel (normalized volumes). For protein identification we targeted proteins that met two criteria: (1) the protein had a normalized volume (NV) greater than 0.03, and (2) the proteins were at least threefold more or less abundant in the planktonic or biofilm proteomes cultivated with 10 mM Ca 2+ versus their matched counterparts cultivated without added Ca 2+ . Proteins of interest were excised from the gels and identified based on peptide mass fingerprint analyzed on 4700 MALDI-TOF/TOF Mass spectrometer (University of Texas, Biomolecular Resource facility, Mass Spec Lab), using the Applied Biosystems GPS (version 3.6) software, MASCOT search engine and NCBInr database. A protein was considered identified if the hit fulfilled four criteria: (1) it was statistically significant (a MASCOT search score above 75), (2) the number of the matched peptides was at least five, (3) the protein sequence coverage was above 20%, and (4) the predicted molecular mass and pI were consistent with the experimentally determined values.

Pyocyanin and Alginate Assays
For pyocyanin analysis, chloroform extraction of the pigment followed by a spectrophotometric assay was used as described in [47]. Briefly, BMM grown mid-log cultures of P. aeruginosa (100 ml) were grown on BMM agar for 24 h, and collected using 3-5 ml of saline. The samples were extracted with 3 ml of chloroform followed by extraction with 1 ml of 0.2 N HCl. The absorbance was measured at 520 nm, and pyocyanin concentrations (mg/ml) were calculated by multiplying the absorbance by extinction coefficient 17.072 [47]. The data were normalized by total cellular protein. For alginate assays, BMM-NI grown mid-log cultures of P. aeruginosa were grown on BMM-NI agar plates for 24 h. To deprive the cells of iron, seven passages onto fresh BMM-NI agar were performed. The cells were collected using 3-5 ml of saline. Alginate was precipitated with 2% cetyl pyridium chloride then isopropanol, dissolved in saline, and detected by using the modified carbazole method described in [46]. The concentration of alginate was determined using sodium alginate (Spectrum) as a standard and normalized by total cellular protein. The measurements were obtained for at least three independent replicates, and the averaged values with standard deviation are presented.

Response to Chemical and Oxidative Stresses
The sensitivity of cells to stress was tested by exposing cultures to 1% ethanol, 10% DMSO, 1 mM H 2 O 2 for 1 h or heat shock at 50uC for 30 min. The concentrations and the duration of treatments were optimized to reach approximately 50% cell survival in wild type FRD1. Prior to exposure, the cultures were grown to mid-log phase in BMM containing 10 mM CaCl 2 or no added Ca 2+ , then diluted to obtain optical density of 0.2 at 600 nm. Viable cells were determined after serial dilution as colony form units (CFUs) before and after exposure. Results are expressed as the mean survival percentages from three independent experiments.

Plant Virulence Assays
The lettuce infection model with the following modifications was used to assess strains pathogenicity [48]. Organic romaine lettuce was purchased fresh from market. Healthy leaves were detached, washed in 0.1% bleach and rinsed twice with distilled water and once with DI water. Midribs were cut and placed in Petri dish containing Whatman No.1 filter paper soaked in 10 mM MgSO 4 . P. aeruginosa wild type and mutant strains were incubated for 18 hours in BMM with 5 mM CaCl 2 or with no added Ca 2+ . Cells were harvested, washed and resuspended in 10 mM MgSO 4, containing the same amount of Ca 2+ as the original culture to obtain an OD 600 of 0.2. One end of the midrib was inoculated with the resuspended culture and the other end of the midrib was inoculated with sterile 10 mM MgSO 4 with 5 mM Ca 2+ or no added Ca 2+ , as controls. The Petri dishes were placed in a clear plastic bin with water to maintain humidity. The bins were incubated at room temperature with ambient sunlight for six days after which the zones of disease were measured. The experiments were performed in at least three independent biological replicates, and the averaged values with standard deviation are presented.

Estimation of Free Intracellular Calcium ([Ca 2+ ] in )
PAO1 and its efhP lacking mutant PAO1043 were transformed with pMMB66EH (courtesy of Dr. Delfina Dominguez), carrying aequorin [49] and carbenicillin resistance genes, using a heat shock method described in [50]. The transformants were selected on Luria bertani (LB) agar containing carbenicillin (300 mg/ml) and verified by PCR using aequorin specific primers (For: 59CTTACATCAGACTTCGACAACCCAAG, Rev: 59CGTA-GAGCTTCTTAGGGCACAG). Aequorin was expressed and reconstituted as described in [19]. Briefly, mid-log phase cells were induced with IPTG (1 mM) for 2 h for apoaequorin production, and then harvested by centrifugation at 6000 g for 5 min at 4uC. Aequorin was reconstituted by incubating the cells in the presence of 2.5 mM coelenterazine for 30 min.
Luminescence measurements and estimation of free [Ca 2+ ] in was performed as described in [19] with slight modifications. Briefly, 100 ml of cells with reconstituted aequorin were equilibrated for 10 min in the dark at room temperature. Luminescence was measured using Synergy Mx Multi-Mode Microplate Reader (Biotek). For basal level of [Ca 2+ ] in , the measurements were recorded for 1 min at 5 sec interval, then the cells were challenged with 1 mM Ca 2+ , mixed for 1 sec, and the luminescence was recorded for 20 min at 5 sec interval. Injection of buffer alone was used as a negative control, and did not cause any significant fluctuations in [Ca 2+ ] in. [Ca 2+ ] in was calculated by using the formula pCa = 0.612(2log 10 k)+3.745, where k is a rate constant for luminescence decay (s -1 ) [51]. The results were normalized against the total amount of available aequorin as described in [19]. The discharge was performed by permeabilizing cells with 2% Nonidet 40 (NP40) in the presence of 12.5 mM CaCl 2. The luminescence released during the discharge was monitored for 10 min at 5 sec interval. The estimated remaining available aequorin was at least 10% of the total aequorin. The experimental conditions reported here were optimized to prevent any significant cell lysis.

Results and Discussion
P. aeruginosa PA4107 Contains Two EF-hand Domains that are Predicted to Localize to the Cell Periplasm Based on 17% amino acid sequence identity (over the entire length) with Ca 2+ -binding calsymin (CasA, YP_472788) and the presence of two EF-hand domains with Ca 2+ -binding loops, we predict that the hypothetical protein, PA4107, is a Ca 2+ -binding protein (Fig. 1a) and we designate it EfhP. EfhP contains Ca 2+binding loops at positions 88-99, and 115-126. By using PROSITE algorithms, EfhP is also predicted to contain a transmembrane region from amino acids 13-32 (Fig. 1b). Sequence analysis by PrediSI predicted that EfhP is secreted with the signal peptide cleaved at the position 29. However, SignalP 4.0, specifically designed to discriminate signal peptides from transmembrane regions [43], gave low probability that the transmembrane domain is a signal peptide. Analysis by TMHMM indicated that EfhP is an integral membrane protein, with a short N-terminal region located in the cytoplasm, and the majority of the protein, including the C-terminal region and the EF-hands oriented to the periplasm. Based on these analyses, we predict that EfhP spans the inner membrane of P. aeruginosa with EF hand domains possibly located in the periplasm.
Sequence analyses of all P. aeruginosa proteins indicate that EfhP does not have EF-hand paralogs in the PAO1 genome. Ten completed genomes of P. aeruginosa, P. putida, P. fluorescens, P. syringae, and P. entomophila each contain one homolog of EfhP with EF-hand domains. These EF-hand proteins have a wide range of amino acid sequence identity over the length of the proteins, ranging from 19 to 99% (Fig. S1). Multiple sequence alignment revealed 30 highly conserved amino acid residues, 12 of which are within the EF-hand domains. Nine of the homologs are predicted to contain N-terminal transmembrane region. The EF-hand protein of P. syringae 1448A has no transmembrane region, and contains four predicted EF-hands domains.
The genome context of efhP in P. aeruginosa PAO1, suggests that it is operonic with two genes encoding hypothetical proteins PA4106 and PA4105. These proteins contain the DUF692 and DUF2063 domains, respectively, with PA4106 having structural similarity to sugar isomerases, and PA4105 having structural similarity to a predicted transcriptional regulator from Neisseria gonorrhoeae (PDB ID: 3DEE). We grouped the genomic environments of efhP homologs into 4 groups (Fig. S2). All three P. aeruginosa and P. putida W619 genomes contain efhP homologs adjacent to the DUF692 and DUF2063 domain proteins (Group I). The other three P. putida strains and P. entomophila L48 genome form group II, where efhP homologs are adjacent to genes for the DUF692 and DUF2063 proteins, but with a gene for a hypothetical protein with a DUF4174 domain. The genome context of the P. fluorescens Pf0-1 EfhP homologs (Group III) differs from the other Pseudomonads, in that it does not contain downstream genes for DUF692 or DUF2093 protein. Rather, the P. fluorescens EfhP homolog is adjacent to upstream genes coding for hypothetical proteins with DUF1780 and DUF3094 domains. The EfhP homolog in P. syringae 1448A, in addition to being the most divergent in amino acid sequence, also has a very different genome context, with upstream genes for a histidine kinase and response regulator. Overall, these results indicate that proteins with EF-hand domains are found in all sequenced pseudomonads, but that they have diverged extensively, and therefore may be involved in different Ca 2+ -dependent physiological processes.

EfhP Contributes to Maintenance of Intracellular Ca 2+ ([Ca 2+ ] in ) Homeostasis in P. aeruginosa PAO1
The role of EfhP in maintaining intracellular Ca 2+ homeostasis was studied using recombinant Ca 2+ -binding luminescence protein aequorin (Fig. 2). The lack of efhP did not affect the basal level of [Ca 2+ ] in (0.1960.01 mM) or the initial increase in response to 1 mM Ca 2+ (1.9960.14 mM). In contrast, however, this increase was not followed by the recovery of [Ca 2+ ] in to nearly basal WT level, as observed in WT PAO1 cells. Instead, the [Ca 2+ ] in continued to raise and in 20 min reaching 3.6660.41 mM, which is almost seven fold higher than in PAO1. This suggests that EfhP is involved in maintaining intracellular Ca 2+ homeostasis, which supports its predicted Ca 2+ -binding capabilities.

EfhP Influences the Proteomes of P. aeruginosa Cultured at High [Ca 2+ ]
In order to determine the role of EfhP in the physiology of P. aeruginosa, we performed global quantitative proteomics on wildtype P. aeruginosa strains PAO1 and the CF isolate FRD1. We compared those proteomes to the proteomes of their respective efhP deletion mutants, designated here as PAO1043 and FRD1043, respectively. Since our previous proteomics studies showed an effect of both [Ca 2+ ] and mode of growth (biofilm versus planktonic) on proteome profiles of the wild-type strains [29], we cultured the strains at low [Ca 2+ ] (no added Ca 2+ ) and high [Ca 2+ ] (10 mM), planktonically and in silicon tubing biofilms [52]. Table 1 shows the proteins that were affected by the mutations in at least one strain when the cells were cultured at 10 mM Ca 2+ . Table 1 also indicates whether the protein is found in increased abundance in the wild-type strains during biofilm growth and/or with added Ca 2+ , based on our prior study [29]. Selected proteins are shown in Figure 3.
Neither efhP mutant strain showed any significant difference in proteome profiles compared to the respective wild-type strain, when cultivated in medium with no added Ca 2+ . However, when cultured with 10 mM Ca 2+ , eight proteins that were abundant in PAO1 were absent in the proteomes of PAO1043 biofilm and/or planktonic cultures: PvdNO, FptA, PA1069, Pfp1, and PhzB1B2D1 (Table 1). The FRD1043 proteome profiles had additional differences compared to the FRD1 profiles when cells were cultured at high [Ca 2+ ]. In planktonic cultures, differences included four proteins (HitA, PA1127, PA5217, and Piv) that were absent from the efhP mutant. In tubing biofilms, twenty-nine protein differences were observed in the FRD1043 mutant versus wild-type strains ( Table 1).

The efhP Mutation Influences the Abundance of Proteins Involved in Iron Acquisition
The role of iron in P. aeruginosa pathogenicity is well characterized [53]. Iron limitation in the host is an important signal for inducing biofilm formation [54] and enhancing the expression of virulence factors [55]. Here, we identified five proteins involved in iron acquisition and storage that were less abundant in the mutants cultured at high Ca 2+ (Table 1, Fig. 3).    These include three proteins involved in the two iron-acquisition systems, PvdNO (the high-affinity pyoverdine system) and FptA (the low-affinity pyochelin receptor protein) [56] as well as two iron binding proteins HitA and PA5217. PvdNO and FptA were not expressed in planktonic PAO1, and were essentially absent in both PA1043 and FRD1043 mutants. HitB and PA5217 were present in both wild type strains (biofilm and planktonic), but were not detected in FRD1043 (biofilm and planktonic). Earlier we demonstrated that all of these proteins were highly induced by Ca 2+ in PAO1 or FRD1 strains [29]. Although the lack of iron did not affect growth of FRD1043 (data not shown), the mutant's ability to produce alginate was abolished during its biofilm growth when deprived of iron at high Ca 2+ (Fig. 4). The efhP complementation in trans fully restored the wild type phenotype. This observation interlinks the regulatory circuits of alginate production with responses to iron starvation and Ca 2+ , and suggests the mediating role of EfhP. We also showed that iron limitation enhances alginate production about twofold in FRD1, which agrees with earlier observations in PAO1 [57] [58] [59].

The efhP Mutations Influence Stress Response Proteins as Well as Oxidative Stress at High [Ca 2+ ]
We identified a number of stress response related proteins that were less abundant in the efhP mutants than in the wild-type strains ( Table 1, Fig. 3). The abundances of these proteins were more impaired in the CF strain FRD1043 than in PAO1043, and more changes occurred in biofilm cultures than in planktonic cells. Stress response proteins that were less abundant in FRD1043 than in FRD1, included the heat shock protein HtpG (PA4352), and the universal stress protein UspA (PA4352). In addition, two proteins PckA and a hypothetical protein, PA1069, were less abundant or not detected in biofilms of both mutants. The genes for the latter two proteins appear operonic with the genes encoding heat shock proteins Hsp33 and Hsp90, respectively. Previously, we showed that all of these proteins were induced by elevated [Ca 2+ ] and/or biofilm growth in P. aeruginosa [29]. It is possible that Ca 2+ is recognized as a signal of environmental stress or stress associated with a host environment, and EfhP plays role in the response. To test this hypothesis, we investigated the effect of chemical, oxidative, and heat induced stresses on the mutant strain, FRD1043 in comparison to wild-type, FRD1. Exposing cells to DMSO, ethanol, and heat did not reveal any differences in survival between mutant and wild type strains at either [Ca 2+ ] (data not shown). However, the lack of efhP impaired the resistance of FRD1 to H 2 O 2 treatment when cells were cultured in the presence of 10 mM [Ca 2+ ], and FRD1043 showed a 60% decrease in survival when exposed to 1 mM of H 2 O 2 for 60 min (Fig. 5). No significant effect on survival to H2O2 was observed in the efhP mutant cultured with no added Ca 2+ . In contrast, wildtype FRD1 cells were equally sensitive to H 2 O 2 with no added Ca 2+ and with 10 mM Ca 2+ . Resistance to H 2 O 2 was partially restored in the efhP mutant containing the complementing efhP gene in trans. The results suggest the importance of EfhP in the ability of P. aeruginosa to withstand oxidative stress at elevated Ca 2+ . This resistance to H 2 O 2 -induced oxidative stress may be crucial for survival of P. aeruginosa in airways, where H 2 O 2 is produced in response to infection [60].

EfhP Affects Production of Virulence Factors, Biofilm Formation, and Virulence in a Plant Infection Model at High [Ca 2+ ]
Pyocyanin is a redox-active virulence factor produced by P. aeruginosa. It imposes oxidative stress on airway epithelial cells and mediates tissue damage and necrosis during lung infection [61,62] [63]. Phenazine biosynthesis proteins PhzB1B2D were at least sixfold less abundant or not detected in the proteomes of strains PA1043 and FRD1043 as compared to PAO1 and FRD1 (Table 1, Fig. 3). These proteins were detected primarily in biofilms of the wild-type strains cultured at high [Ca 2+ ] [30] [29], and the mutation-induced effects detected here occurred primarily in biofilm cultures. In agreement, pyocyanin production was not affected in the mutant planktonic cultures (data not shown), but was completely abolished in FRD1043 biofilm cells grown at high [Ca 2+ ] (Fig. 6). Biosynthesis of pyocyanin is controlled by the RhlI/RhlR [64] as well as PQS quorum sensing systems [65]. Furthermore, pyocyanin regulation is also affected by iron depletion [66], therefore the effect of EfhP seen here may be related to its effect on iron acquisition enzymes (Table 1) or its potential interaction with one of several major regulatory systems in P. aeruginosa (reviewed in [67]).
We previously showed that several proteases are induced by high [Ca 2+ ], and that extracellular proteases accumulate in the matrix material of FRD1 biofilms [30]. Two proteases Piv, PfpI were at least fourfold less abundant in PA1043 than in PAO1, and were not detected in FRD1043 planktonic and biofilm cells (Table 1). In addition, a trigger factor (PA1800), encoded on the Figure 4. Alginate production at no iron in FRD1 and FRD1043. To deprive the cells of iron, seven passages on no-iron BMM (BMM-NI) agar were performed. The cells were collected using saline. The concentration of alginate was determined using sodium alginate (Spectrum) as a standard and normalized by total cellular protein. Fold difference was calculated between the alginate produced after seventh and first passages. The measurements were obtained for at least three biological replicates, and the mean values with standard deviation are presented. doi:10.1371/journal.pone.0098985.g004 same operon as the clpP, clpX, and lon proteases had reduced abundance in the PA1043 and FRD1043 mutant strains (Fig. 3). Since secreted proteases are harbored in the biofilm matrix of P. aeruginosa [30], the decreased abundance of proteases in the mutant strain biofilms may be related to the mutant's reduced ability to grow a biofilm. Earlier we have shown that Ca 2+ enhances biofilm formation in P. aeruginosa [30], [29]. To test whether EfhP plays a role in this induction, FRD1, its efhP mutant, and complemented strain were grown in biofilms using 96-well plates. The lack of efhP caused 35% decrease in the organism's ability to form biofilm, however complementation of the gene did not restore the wild type phenotype (Fig. S3). The latter may be due to the different level of efhP expression in the complemented strain compared to the wild type, which may be particularly influential since the protein plays role in maintaining intracellular Ca 2+ homeostasis. We also cannot rule out that the mutant strain acquired a compensatory mutation that cannot be complemented. Finally, the reducing effect of the efhP deletion on biofilm growth coincided with a similar effect on H 2 O 2 resistance and pyocyanin production. Both of the latter phenotypes are induced during biofilm growth [29] [68] and were restored by the efhP complementation.
To determine if EfhP plays a role in P. aeruginosa virulence when the bacteria are exposed to high [Ca 2+ ], we used the lettuce infection model [48] and compared infection zones in the wildtype and the efhP mutant strains. In both PAO1 and FRD1, preincubation with 5 mM CaCl 2 resulted in increased zone of disease, compared to bacteria cultured with no added CaCl 2 (Fig. 7). This effect of Ca 2+ was more pronounced in FRD than in PAO1. When cultured at low [Ca 2+ ], little effect of the efhP mutation was observed for both strains. However, when the strains were cultured with 5 mM CaCl 2 , the zone of disease reduced at least two fold in the mutants compared to the wild-type strains. The disease zone was partially restored in the mutant cells  complemented with efhP in trans. These results suggest that EfhP is required for Ca 2+ -induced virulence of P. aeruginosa in a plant model, likely due to the effect of EfhP on virulence factor production at high [Ca 2+ ].
The P. aeruginosa EfhP has sequence similarity to the EF-hand protein, CasA, produced by the plant symbiont, Rhizobium etli. R. etli casA is expressed during plant host invasion and is required for symbiotic nitrogen fixation [24]. Therefore we hypothesize that EfhP of P. aeruginosa may also play a role in Ca 2+ -dependent interaction between P. aeruginosa and its host during infection. Based on the microarray expression data available from the GEO database, the transcription of efhP increases more than 150 fold in P. aeruginosa isolates from CF sputa grown both in vitro and in vivo (GDS2869 and GDS2870). Considering that Ca 2+ accumulates in pulmonary liquids of CF patients [10], this provides further evidence for a role of EfhP in P. aeruginosa adaptation to a host environments with elevated [Ca 2+ ].

Conclusions
Sequence analyses predicted that EfhP is a Ca 2+ -binding protein spanning the inner membrane, with the two EF-hand domains facing the periplasm. Deletions of efhP in two P. aeruginosa strains causes multiple changes in the cytosolic proteome of both strains, but with more changes occurring in the CF pulmonary isolate FRD1. The effects of the efhP deletions only occurred when the cells were exposed to elevated [Ca 2+ ] and included reduced abundance of virulence factors and stress response proteins. The lack of efhP abolished production of pyocyanin, and reduced the degree of infection, biofilm formation, and resistance to oxidative stress in FRD1 at high [Ca 2+ ]. The mutant also lost the ability to produce alginate at no iron and high [Ca 2+ ]. Finally, the lack of EfhP abolished the ability of P. aeruginosa to maintain intracellular Ca 2+ homeostasis. These findings suggest that EfhP is important for Ca 2+ homeostasis and plays role in Ca 2+ -triggered virulence and resistance of P. aeruginosa in high Ca 2+ environments. Supporting Information Figure S1 Sequence alignment of EfhP and its homologs from nine Pseudomonas strains. EF-hands are shown in boxes, transmembrane regions are shown in grey. Conserved amino acids are in bold. Sequences were aligned using ClustalW. (TIF) Figure S2 Genome context of the efhP homologs from Pseudomonads. The organization of the efhP genomic neighborhood in P. putida strains KT2440, F1, W619, and GB-1, P. entomophila L48, P. aeruginosa strains LESB58, PAO1, and PA7, P. fluoresecens Pf0-1, and P. syringae 1448A. Genes are colored according to their predicted functional domains: black, EfhP and homologs; wavy lines, DUF domain proteins; horizontal lines, transcriptional regulators; diagonals, two component system; light grey, hypothetical proteins. (TIF) Figure S3 Biofilm assessment. Biofilm formation was quantified using 96-well plates. Cells were cultured on BMM medium containing either 10 mM CaCl2 or no added calcium for 24 h with medium replacement at 12 h. Biofilms were stained with crystal violet, and the absorbance was measured at 595 nm using a microtiter plate reader. (TIF)