Calcium homeostasis in Pseudomonas aeruginosa requires multiple transporters and modulates swarming motility

Abstract

Pseudomonas aeruginosa is an opportunistic human pathogen causing severe acute and chronic infections. Earlier we have shown that calcium (Ca²⁺) induces P. aeruginosa biofilm formation and production of virulence factors. To enable further studies of the regulatory role of Ca²⁺, we characterized Ca²⁺ homeostasis in P. aeruginosa PAO1 cells. By using Ca²⁺-binding photoprotein aequorin, we determined that the concentration of free intracellular Ca²⁺ ([Ca²⁺]_in) is 0.14±0.05 μM. In response to external Ca²⁺, the [Ca²⁺]_in quickly increased at least 13 fold followed by a multi-phase decline by up to 73%. Growth at elevated Ca²⁺ modulated this response. Treatment with inhibitors known to affect Ca²⁺ channels, monovalent cations gradient, or P-type and F-type ATPases impaired [Ca²⁺]_in response, suggesting the importance of the corresponding mechanisms in Ca²⁺ homeostasis. To identify Ca²⁺ transporters maintaining this homeostasis, bioinformatic and LC-MS/MS-based membrane proteomic analyses were used. [Ca²⁺]_in homeostasis was monitored for seven Ca²⁺-affected and eleven bioinformatically predicted transporters by using transposon insertion mutants. Disruption of P-type ATPases PA2435, PA3920, and ion exchanger PA2092 significantly impaired Ca²⁺ homeostasis. The lack of PA3920 and vanadate treatment abolished Ca²⁺- induced swarming, suggesting the role of the P-type ATPase in regulating P. aeruginosa response to Ca²⁺.

INTRODUCTION

Calcium (Ca²⁺) is a well-known signaling molecule that regulates a number of essential processes in eukaryotes [1]. Abnormalities in cellular Ca²⁺ homeostasis have been implicated in many human diseases, including diseases associated with bacterial infections, for example, cystic fibrosis (CF) and endocarditis. In addition, Ca²⁺ plays a regulatory role in innate immune response [2], and its intracellular ([Ca²⁺]_in) and extracellular ([Ca²⁺]_ex) concentrations fluctuate in response to inflammation. For example, the levels of [Ca²⁺] in pulmonary fluid and nasal secretions of CF patients are increased [3, 4]. Altogether the evidence suggests that cellular Ca²⁺ balance in a host may provide an environmental cue for opportunistic pathogenic bacteria and trigger their virulence. In support, in prokaryotes, Ca²⁺ has been implicated in various physiological processes such as spore formation, motility, cell differentiation, transport, and virulence (reviewed in [5]). It has also been shown that Ca²⁺ modulates bacterial gene expression [6-8], suggesting its regulatory role in prokaryotes. Furthermore, there is growing evidence that Ca²⁺ plays a signaling role in prokaryotes, which requires a tight control of cellular Ca²⁺ homeostasis. Several bacteria including Escherichia coli [9], Propionibacterium acnes [10], Streptococcus pneumoniae [11] Bacillus subtilis [12] and cyanobacteria [13] have been shown to maintain intracellular Ca²⁺ at sub-micromolar levels, and produce Ca²⁺ transients in response to environmental and physiological conditions [14, 15]. Such responses may play a key role in Ca²⁺-regulated bacterial physiology and virulence, however, the molecular mechanisms of bacterial Ca²⁺ homeostasis have not been well characterized. Several studies suggest that bacteria control their [Ca²⁺]_in by using multiple mechanisms of transporting or chelating Ca²⁺ (reviewed in [5]).

Three major types of Ca²⁺ transport systems have been described in prokaryotes: gradient driven Ca²⁺ exchangers, ATP-ases, and non-proteinaceous polyhydroxybutyrate-polyphosphates (PHB-PP) channels. Ca²⁺ exchangers have been identified in a number of bacterial genera and are thought to serve as a major mechanism for Ca²⁺ transport in prokaryotes [16]. They are low-affinity Ca²⁺ transporters that use the energy stored in the electrochemical gradient of ions, and, depending on the gradient, can operate in both directions. The specificity of the transporters may vary. For example, YftkE (ChaA) from B. subtilis [17] as well as ApCAX and SynCAX from cyanobacteria [18] are Ca²⁺- specific, whereas ChaA from E. coli exhibits Na⁺/H⁺ and K⁺/H⁺ antiport activity in addition to Ca²⁺/H⁺ [19]. Ca²⁺ exchangers may also play role in cell sensitivity to Ca²⁺ and salt tolerance, as exemplified by cyanobacterial ApCAX and SynCAX [18]. ATP-ases are mostly high-affinity pumps that export cations from the cytosol by using the energy of ATP. They include P-type and F-type ATPases. Ca²⁺- translocating P-type ATPases belong to P2A and P2B subgroups, as classified in [20]. The former are similar to mammalian sarco(endo)plasmic reticulum (SERCA) Ca²⁺ pumps exporting Ca²⁺ against steep transmembrane gradients, and the latter are similar to plasma membrane (PMCA) calmodulin-binding ATPases. Five characterized prokaryotic P2A-ATPases include PacL from cyanobacteria [21], LMCA1 from Listeria monocytogenes [22], YloB from Bacillus subtilis [23], CaxP from Streptococcus penumoniae [11], and PacL from Flavobacterium odoratum [24]. Most of them were shown to export Ca²⁺ in membrane vesicles and proposed to play a role in cell protection against high Ca²⁺. LMCA1 from L. monocytogenes [22] and PacL from F. odoratum [21] were shown to undergo Ca²⁺-dependent phosphorylation required to transport Ca²⁺. F-type ATPases, or ATP synthases, are known to synthesize ATP at the expense of transmembrane electrochemical gradient of protons (most commonly). So far, only one F-type ATPase AtpD in E. coli was shown to play role in Ca²⁺ homeostasis, most likely due to its role in ATP synthesis [25]. Overall, although several prokaryotic gradient- and ATP- driven transporters were shown to translocate Ca²⁺ in-vitro, only few were tested for their role in cellular Ca²⁺ homeostasis in-vivo, of which only ion-exchanger SynCAX in Synechocystis sp. PCC6803 was shown to play role in cellular Ca²⁺ efflux [18]. The difficulty of identifying the roles of Ca²⁺ transporters in-vivo is likely due to their functional redundancy, the molecular basis of which requires further studies.

Pseudomonas aeruginosa is an opportunistic human pathogen, and a major cause of nosocomial infections and severe chronic infections in endocarditis and in CF patients. Earlier, we showed that growth at high Ca²⁺ enhances P. aeruginosa biofilm formation and induces biosynthesis of several secreted virulence factors including alginate, extracellular proteases and pyocyanin [6, 7]. However, the molecular mechanisms of Ca²⁺ regulation are not defined. To enable studies required to uncover such mechanisms, it is necessary to first characterize cellular Ca²⁺ homeostasis in this organism. Therefore, the aim of this work was to measure the intracellular Ca²⁺ concentration ([Ca²⁺]_in) in P. aeruginosa cells and characterize its responses to external Ca²⁺. We employed a recombinant photoprotein aequorin-based reporter system, which has been successfully used to measure [Ca²⁺]_in in live prokaryotic cells [12, 25], and to monitor both short- and long-term [Ca²⁺]_in responses to external Ca²⁺ transients [9, 12] as well as other environmental and physiological determinants [12, 15]. We also aimed to identify Ca²⁺ transporters that play role in maintaining cellular Ca²⁺ homeostasis. The strategy combined bioinformatic and proteomic approaches, followed by characterization of transposon insertion mutants obtained from the University of Washington Genome Center. This study presents the first evidence of Ca²⁺ homeostasis in P. aeruginosa and identifies several mechanisms and proteins required for maintaining Ca²⁺ homeostasis and regulating Ca²⁺-induced swarming motility. In addition, the results provide a basis and excellent tools for further studies of the roles of cellular Ca²⁺ homeostasis in the regulation of Ca²⁺-modulated physiology and virulence of this important human pathogen.

MATERIALS AND METHODS

Chemicals used in this study are listed in the supplementary information. Primers were obtained from Integrated DNA Technologies Inc.

Bacterial strains, plasmids, and media

P. aeruginosa strain PAO1, the non-mucoid strain with genome sequence available was used in the study. Biofilm minimal medium (BMM) was made as described in [6]. When required, CaCl₂.2H₂O was added to final concentration of 1 or 5 mM. For proteomic studies, PAO1 cells were first grown in 5 ml tubes for 16 h (mid-log), and then used to inoculate (0.1 %) 100 ml fresh medium in 250 ml flasks. The cultures were grown to mid-log growth phase and harvested by centrifugation. Transposon insertion mutants were obtained from the University of Washington Two - Allele library and are listed in Table 1. The mutants contained either ISphoA/hah or ISlacZ/hah insertions with tetracycline resistance cassette that disrupted the genes of interest. The mutations were confirmed by two-step PCR: first, transposon flanking primers were used to verify that the target gene is disrupted, and second, transposon-specific primers were used to confirm the transposon insertion. The primer sequence is available at www.gs.washington.edu. For convenience, the mutants were designated as PA:IS, where PA is the identifying number of the disrupted gene from P. aeruginosa PAO1 genome (www.pseudomonas.com).

Table 1.

Protein prediction and identification. Transposon mutants.

PA ORFa	Protein name, gene nameb	Predicted domain(s)c	Best characterized hit in the nr NCBI, Accession #, Organism, % identityd	Transposon mutant PW#e Mutant Genotype	Mutant Designated Name
ATP-driven transporters
PA1429	Probable cation transporting P-Type ATPase	E1-E2 ATPase	P-type Na+ ATPase, BAF91372.2, Exiguobacterium aurantiacum, 45%	PW3596 PA1429-E05::ISlacZ/hah	PA1429::IS
PA4825	Probable Mg2+ transport ATPase, P-Type, mgtA	E1-E2 ATPase	Ca2+/Mn2+ P-type ATPase PMR1, CAB87245, Candida albicans, 25%	PW9116 PA4825-F12::ISphoA/hah	PA4825::IS
PA3690	Probable metal-transporting P-type ATPase	E1-E2 ATPase; Heavy metal associated domain	Cd2+ /Zn2+ transporting ATPase, HMA3 P0CW78.1, Arabidopsis thaliana, 32%	PW7241 PA3690-C04::ISlacZ/hah	PA3690::IS
PA1634	K+-transporting ATPase, beta subunit, kdpB	E1-E2 ATPase	NA	PW3911 PA1634-E05::ISlacZ/hah	PA1634::IS
PA2435	Heavy metal translocating P-type ATPase, hmtA	E1-E2 ATPase		PW5099 PA2435-A02::ISphoA/hah	PA2435::IS
PA3920	Probable metal transporting P-type ATPase	E1-E2 ATPase; Heavy-metal-associated domain	Cation transporting ATPase, NP_440588, 44%, Synechocystis PCC6803, 44%	PW7626 PA3920-G01::ISphoA/hah	PA3920::IS
PA1549	Probable cation-transporting P-type ATPase	E1-E2 ATPase; Heavy-metal-associated domain	CtpA, AAW66130, Rubrivivax gelatinosus, 36%	PW3788 PA1549-G12::ISphoA/hah	PA1549::IS
PA5554	Probable ATPase synthase, beta subunit, atpD	F1 ATP synthase	F1F0-ATP synthase β subunit, Q587Q4, Acidithiobacillus ferroxidans, 77%	PW10412 PA5554-B01::ISlacZ/hah	PA5554::IS
PA4496	Probable ABC transporter, substrate binding subunit	Substrate binding component	Dipeptide binding protein chain A, 1DPP A, E. coli, 50%	PW8565 PA4496-F02::ISphoA/hah	PA4496::IS
PA3400	Hypothetical protein	ABC type transporter	NA	PW6735 PA3400-H11::ISphoA/hah	PA3400::IS
Ion gradient driven exchangers
PA3963	Probable Transporter	Cation Efflux Superfamily	Zinc transporter, YiiP, 3H90_A, E.coli K-12, 45%	PW7707 PA3963-E02::ISphoA/hah	PA3963::IS
PA2092	Probable major facilitator superfamily transporter	Major Facilitator Superfamily	Purine efflux pump PbuE, Q0GQS6.1, Bacillus amyloliquefaciens, 35%	PW4602 PA2092-F01::ISlacZ/hah	PA2092::IS
PA4292	Probable Phosphate transporter	PO43-/SO42- permease	NA	PW8230 PA4292-A03::ISphoA/hah	PA4294::IS
PA0397	Probable cation efflux system protein	Cation efflux family	Znt-like transporter 2, Q8NEW0.1, Homo sapiens, 27%	PW1733 PA0397-E03::ISlacZ/hah	PA0397::IS
Other transporters
PA2999	Probable Na+-translocating NADH:ubiquin one oxidoreductase subunit, nqrA	NADH-quinone reductase domain	Na+ translocating NADH ubiquinone oxidoreductase subunit A , ZP_08756125.1, Haemophilus pittmaniae HK85,58%	PW6021 PA2999-D10::ISlacZ/hah	PA2999::IS
PA4614	Probable conductance mechanosensitive channel, mscL	Large-conductance mechanosensitive channel	Mechanosensitive channel large, MscL chain A, 3HZQ_A, Staphylococcus aureus, 36%	PW8772 PA4614-B11::ISphoA/hah	PA4614::IS
PA5167	Dicarboxylic acid transporter, dctP	Bacterial extracellular solute binding		PW9688 PA5167-F06::ISphoA/hah	PA5167::IS
PA4016	Hypothetical protein	Membrane lipoprotein lipid attachment site	NA	PW7791 PA4016-E06::ISphoA/hah	PA4106::IS

^aGene identifier in the PAO1 genome available at www.pseudomonas.com.

^bAs annotated in the PAO1 genome.

^cDomains were predicted by the algorithms in CDD and PROSITE.

^dThe % identity was calculated over the full length of the proteins using transporter proteins encoded in PAO1 genome as a query.

^eThe mutant strain identifier in the UW library of transposon mutants available at www.gs.washington.edu.

^f Proteins were identified by using LC-MS/MS. Protein identification was accepted if the probability threshold was greater than 99% and at least three peptides were identified, each with 95% certainty. Spectral count (SC) was used to estimate the differential protein abundance. Fold difference was calculated as a ratio of the number of the peptides identified at 5 mM vs. 0 mM Ca²⁺. ND, protein was detected at 0 mM and was not detected at 5 mM Ca²⁺. New, protein was detected at 5 mM, but was not detected at 0 mM Ca²⁺. PA number and fold difference (in brackets) are provided for the identified proteins that are encoded within the operonic gene clusters of the predicted proteins.

NA. No characterized hits with greater than 25% amino acid sequence identity were found.

Sequence analyses

To predict Ca²⁺ transporters in P. aeruginosa PAO1 genome, we applied BLASTp sequence alignments, using the National Centre for Biotechnology Information (NCBI) non-redundant database (GenBank release 160.1), as well as functional domains search using Conserved Domain Database (CDD) and PROSITE. Homologous proteins were selected based on at least 25 % identity over the full length of amino acid sequence. For phylogenetic analyses, amino acid sequences of the functionally characterized transporters of Ca²⁺ and other cations were aligned with P. aeruginosa putative Ca²⁺ transporters using ClustalW. Thus obtained multiple sequence alignments were then used to build unrooted phylogenetic tree using Neighbor-joining algorithm in the MEGA 5.1 software.

Proteomic analysis

Membrane proteins were isolated by carbonate extraction as described in [26]. The details are described in supplementary information. Protein concentration was determined using the 2D Quant kit (GE Healthcare). LC-MS/MS-based spectral counting was performed at the OSU Proteomics Facilities using LTQ-OrbitrapXL mass spectrometer. Proteins were identified using Mascot (v.2.2.2 from Matrix Science, Boston, MA, USA) and a database generated by in silico digestion of the P. aeruginosa PAO1 proteome predicted from the genome. Search results were validated using Scaffold (v.3 from Proteome Software Inc., Portland, OR). Proteins were considered identified if the protein probability threshold was greater than 99 % and at least three peptides were identified, each with 95 % certainty.

Expression and reconstitution of aequorin

PAO1 and transposon mutants were transformed with pMMB66EH (courtesy of Dr. Delfina Dominguez), carrying aequorin [27] and carbenicillin resistance genes, using a heat shock method described in [28]. The transformants were selected on Luria bertani (LB) agar containing carbenicillin (300 μg/ml) and, in case of transposon mutants, tetracycline (60 μg /ml) and verified by PCR using aequorin specific primers (For: 5’CTTACATCAGACTTCGACAACCCAAG, Rev: 5’CGTAGAGCTTCTTAGGGCACAG). Aequorin was expressed and reconstituted as described in [9] with modifications. The details are described in supplementary information.

Luminescence measurements and estimation of free intracellular calcium

Cells with reconstituted aequorin were aliquoted (100 μl) in 96 well plate (Griener bio one, Lumitrack 600) and, when required, treated with inhibitors (2, 4 di-nitrophenol at 0.5, 1, or 2 mM; LaCl₃ at 300 or 600 μM; gramicidin D at 1 or 10 μg/ml; calcimycin at 5 μM; or vanadate at 2 mM) for 10 min in the dark at room temperature without shaking. Gramicidin D and calcimycin were dissolved in 50 % and 3 % DMSO, respectively. To ensure penetration of gramicidin D and calcimycin through the bacterial outer membrane, we added 10 μg/ml of compound 48/80, known to permeabilize the outer membrane of gram negative bacteria without affecting the cytoplasmic membrane [29]. In this case, the corresponding amounts of DMSO and 48/80 were used to treat cells as a negative control. Luminescence was measured at 25 °C using Synergy Mx Multi-Mode Microplate Reader (Biotek). To measure the basal level of [Ca²⁺]_in, the measurements were recorded for 1 min at 4 sec interval, then the cells were challenged with 1 or 5 mM Ca²⁺ (final concentration) injected by using the internal Synergy injectors, mixed for 1 sec, and the luminescence was recorded for 20 min at 4-5 sec interval. Injection of buffer alone was used as a negative control, and did not cause any significant fluctuations in [Ca²⁺]_in. [Ca²⁺]_in was calculated by using the formula pCa= 0.612 (−log₁₀k) + 3.745, where k is a rate constant for luminescence decay (s⁻¹) [9]. The excel-based template incorporating the formula and the aequorin standard curve was generously shared by Dr. Anthony Campbell. The results were normalized against the total amount of available aequorin, which was estimated by summing the light detected during an entire experiment and the light detected during discharge. The discharge was performed by permeabilizing cells with 2 % Nonidet 40 (NP40) in the presence of 12.5 mM CaCl₂. The luminescence released during the discharge was monitored for 5 min at 4-5 sec interval. The estimated remaining available aequorin was at least 10 % of the total aequorin, unless mentioned otherwise. To control possible cell lysis and aequorin leakage during the procedure, following the luminescence measurements, the cells were collected by centrifugation, and luminescence was reported for both the supernatants and the cell pellets resuspended in buffer (25 mM HEPES, 1 mM MgCl₂, 125 mM NaCl; pH 7.5). The experimental conditions reported here were optimized to prevent any significant cell lysis. The responses to Ca²⁺ challenges were characterized and validated by using curve fitting analysis with IGOR PRO software v.6.3.1.2. (WaveMetrics).

Ca²⁺ Tolerance Assay

To test cell tolerance to high Ca²⁺, PAO1 and mutants were grown at 5 or 100 mM Ca²⁺. Mid-log cultures grown in 5-ml BMM were inoculated (1 %) into 200 μl of fresh BMM (no added, 5 or 100 mM Ca²⁺) in 96 well plates. The plates were incubated for 12, 24 h or 48 h, and the OD₆₀₀ was measured using Synergy Mx Multi-Mode Microplate Reader (Biotek). The tolerance to Ca²⁺ was calculated as a ratio of OD₆₀₀ of the cultures grown at elevated Ca²⁺ to the cultures grown at no added Ca²⁺. For the wild type PAO1, this ratio equals one.

Swarming Assay

Swarming motility was assayed as described in [30]. PAO1 and mutants were grown in BMM at 0 mM or 5 mM Ca²⁺. 2 μl of the mid log cultures normalized to the OD600 of 0.3 were spot inoculated onto the surface of swarming agar [30]. When needed, inhibitors (10 μg/ml gramicidin D, 2 mM vanadate, or 600 μM LaCl₃) were added. Gramicidin D was dissolved in DMSO and added with compound 48/80 (10 μg/ ml). After inoculation, the plates were incubated for 16 h and the colony diameters were measured. The effect of Ca²⁺ was calculated as a fold difference (ratio) between the diameters of the colonies grown at 5 mM and 0 mM Ca²⁺. The mutants and treatments were compared to their corresponding controls using the mean percentage of fold difference from at least three independent experiments.

RESULTS

P. aeruginosa maintains [Ca²⁺]_in homeostasis

To study Ca²⁺ homeostasis in P. aeruginosa PAO1, we first measured the basal level of free intracellular Ca²⁺ ([Ca²⁺]_in), and then monitored the changes in [Ca²⁺]_in in response to externally added 1 and 5 mM Ca²⁺. Cells grown at no added Ca²⁺ (naïve) or 5 mM Ca²⁺ (induced) were compared. In naïve cells, the [Ca²⁺]_in was 0.14 ± 0.05 μM (Fig. 1A). The addition of 1 mM Ca²⁺ caused a 13 fold increase of [Ca²⁺]_in to 1.86 ± 0.53 μM within 0.6 min (2.92 μM/min) followed by a two-phase decline, validated by the curve fitting analysis. The first, fast, phase proceeded at a rate 0.75 μM/min, lasted for ~2 min, and accounted for 38 % of [Ca²⁺]_in reduction. During the second, slow, phase, [Ca²⁺]_in reduced for another 35% to 0.5 μM at a rate 0.04 μM/min. When the cells were transferred to a fresh buffer containing no Ca²⁺, a complete recovery to the basal [Ca²⁺]_in level was observed (data not shown). We also monitored [Ca²⁺]_in for additional 70 min and detected a very slow decrease at a rate 1.43 nM/min (data not shown). Considering the minor changes during this extended incubation, [Ca²⁺]_in was monitored during 21 min in all further experiments. Challenging naïve cells with 5 mM Ca²⁺ caused 28 fold increase in [Ca²⁺]_in to 3.93 ± 0.37 μM within 0.92 min (3.88 μM/min) followed by a multi-phase decrease. First, [Ca²⁺]_in dropped similarly to naïve cells by 38 %, but at a lower rate 0.46 μM/min. Then it increased to 3.01 ± 0.34 μM and slowly declined to 2.22 μM, showing a total decrease by 44 %.

Figure 1.

The effect of externally added Ca²⁺ on [Ca²⁺]_in in P. aeruginosa PAO1. Cells were grown in BMM with no added Ca²⁺ (A) or in the presence of 5 mM Ca²⁺ (B), and challenged with of 0 mM (black), 1mM (dark grey), and 5 mM (dash) Ca²⁺. The basal level of luminescence was monitored for 1 min. 1 mM or 5 mM CaCl₂ was added at the time indicated by the arrow, followed by luminescence for 20 min. Changes in free [Ca²⁺]_in were calculated as described in the Methods section. The averages of at least three independent experiments are plotted.

In induced cells, the basal [Ca²⁺]_in was 0.22 ± 0.04 μM, which is 57 % higher than in naïve cells (Fig. 1B). Upon exposure to 1 mM Ca²⁺, the induced cells increased their [Ca²⁺]_in at a rate (3.12 μM/min) similar to naïve cells, but only fivefold (to 1.19 ± 0.11 μM) and during a shorter period of time (0.3 min). The following decrease proceeded with a much lower rate (0.16 μM/min), lasted for ~1.5 min, and accounted for 21 %. In contrast to naïve cells, during the second phase, [Ca²⁺]_in slowly increased with a rate 0.04 μM/min to 1.23 ± 0.45 μM. The response to 5 mM Ca²⁺ in induced cells also showed a multi-phase pattern. Although the first rapid increase brought [Ca²⁺]_in to a similar to naïve cells level (3.66 ± 0.44 μM), it was only 17 fold higher than the basal level. This increase occurred very quickly within 0.08 min at a rate 42.62 μM/min. The following decline was threefold faster than in naïve cells (1.6 μM/min). The second increase was similar to the one in naïve cells elevating [Ca²⁺]_in to 2.9 ± 0.45 μM, followed by a decline to 2.13 ± 0.27 μM with a total decrease by 42 %.

The PAO1 genome contains multiple homologs of Ca²⁺ transporting ATPases and gradient driven exchangers

Homology searches in the PAO1 genome (www.pseudomonas.com) using amino acid sequences of the characterized prokaryotic Ca²⁺ transporters revealed 18 genes encoding different types of transporters (Table 1). Based on sequence similarity and predicted conserved domains, ten were predicted to encode P-, F-type, and ABC-type ATPases, and eight to encode ion exchangers and other types of transporters. They include two earlier characterized proteins: heavy metal translocating P-type ATPase (PA2435) and dicarboxylic acid transporter (PA5167) [31, 32]. The comparative sequence analyses revealed that seven predicted P-type ATPases share 13 - 39 % sequence identity and do not have other paralogs in the PAO1 genome. Phylogenetic analysis using functionally characterized prokaryotic P-type ATPases showed that the PAO1 proteins can be grouped into four clades, which coincide with ion specificity: 1) Ca²⁺ (PA1429); 2) Mg²⁺ (PA4825); 3) K⁺ (PA1634); and 4) Pb²⁺, Cd²⁺, Zn²⁺, CO²⁺, Cu²⁺ (PA2435, PA3920, PA1549, and PA3690) (Fig. S1). Further sequence comparison revealed that PA1429 together with five prokaryotic Ca²⁺-translocating P-type ATPases in clade 1 are more closely related to the human Ca²⁺ translocating ATPase SERCA than to the human calmodulin-binding ATPase PMCA1 (data not shown). This supports their classification into the subgroup of P2A-ATPases [20]. Interestingly, clade 2, including PA4825 and Mg²⁺-translocating MgtB from S. typhimurium, is closely related to PMCA1 (data not shown). Furthermore, considering that PA2435 was shown to uptake both Cu²⁺ and Zn²⁺ [32], we combined Cu²⁺ and heavy metal- translocating proteins into one clade (4). As a result, PA1549, earlier grouped into functionally uncharacterized subfamily FUPA27 [33], was clustered with clade 4 proteins, suggesting its possible involvement in translocation of these ions. Other predicted transporters have one to four paralogs in the PAO1 genome and share 28 % – 82 % amino acid sequence identity. All the predicted transporters are conserved among ten sequenced pseudomonads and share 19 % – 100 % identity with their corresponding homologs.

Growth at elevated Ca²⁺ alters abundance of PAO1 membrane proteins

The modulated [Ca²⁺]_in response in Ca²⁺ induced cells suggested the involvement of transporters, whose expression is affected by Ca²⁺. To identify such transporters, membrane proteins from PAO1 cells grown at no added or 5 mM Ca²⁺ were extracted and subjected to LC-MS/MS spectral counting - based comparative analyses. In total, about 500 proteins were identified, of which more than 80 % were transmembrane or membrane-associated proteins. These included seven bioinformatically predicted putative Ca²⁺ transporters, and ten proteins encoded within the same apparent operons (Table 1). To estimate the effect of Ca²⁺, for every protein we calculated a ratio of the total number of fragmentation spectra that map to the peptides of the protein in the samples collected at 5 mM vs. no added Ca²⁺. The results suggest that six predicted Ca²⁺ transporters (shown in bold in Table 1) increased abundance at least twofold in response to growth at elevated Ca²⁺. Considering that apparent operonic genes may have similar expression profile, the results also suggest Ca²⁺ induction for PA3400, PA5167, PA4016, and PA1549. However, PA4496 and PA4497 showed decreased abundance in response to 5 mM Ca²⁺.

Ca²⁺ homeostasis in PAO1 involves multiple transporters of different types

To study the role of the predicted transporters in Ca²⁺ homeostasis, we obtained 18 mutants, each with one predicted transporter encoding gene disrupted with either ISphoA/hah or ISlacZ/hah transposon insertion. The mutants were monitored for their [Ca²⁺]_in response to 1 mM Ca²⁺ and compared to the wild type (WT) PAO1 cells. Based on the results they were grouped into four groups (Fig 2). Group I (four mutants) showed the initial rapid increase in [Ca²⁺]_in at least twofold higher than in WT cells followed by the recovery to approximately the WT level [Ca²⁺]_in. Group II (five mutants) failed the recovery to the WT level [Ca²⁺]_in during 21 min monitoring, and had the remaining [Ca²⁺]_in at the levels at least twofold higher than in PAO1 cells. Group III (four mutants) showed two peaks of [Ca²⁺]_in increase and the remaining [Ca²⁺]_in at the level at least twofold higher than in WT cells. The [Ca²⁺]_in changes less than twofold vs. PAO1 were considered not significant (five mutants in group IV).

Figure 2.

Free [Ca²⁺]_in profiles of transposon mutants with disrupted ATP-dependent transporters (A) or ion exchange transporters (B). The mutants were obtained from the University of Washington Two - Allele library. Cells were grown in BMM media with no added Ca²⁺. The basal level of luminescence was monitored for 1 min. 1 mM CaCl₂ was added at the time indicated by the arrow, followed by luminescence measurements for 20 min. Changes in free [Ca²⁺]_in were calculated as described in the Methods section. PA numbers represent the open reading frames in PAO1 genome. Black, PAO1 wild type; grey, transposon mutant. The data is an average of at least three independent experiments. Upward and downward arrows indicate that the protein abundance was increased and decreased, correspondingly, during growth at 5 mM CaCl₂.

Among the ten examined mutants with disrupted putative ATP-ases, seven showed significant changes in [Ca²⁺]_in response (Fig 2A). They included mutants lacking six P-type ATPases: PA3690, PA4825 (Group I), PA1429, PA1549 (Group II), PA2435, PA3920 (Group III), and ABC transporter PA3400 (Group II). The group III mutants showed the most drastic changes in [Ca²⁺]_in response, and together with PA1549-lacking mutant failed to significantly decrease the elevated [Ca²⁺]_in. The remaining three ATP-ases: P-type PA1634, F-type PA5554, and ABC transporter PA4496 showed no role in maintaining [Ca²⁺]_in homeostasis (Group IV). Among the eight tested putative ion exchangers and other transporters, disruption of six showed significant changes in [Ca²⁺]_in response (Fig. 2B). They included: three ion exchangers PA3963, PA4292 (Group I), and PA2092 (Group III), dicarboxylic acid transporter PA5167, hypothetical protein PA4016 (Group II), and mechanosensitive channel PA4614 (Group III). Among these proteins, the most significant effect was observed for the mutant lacking a major facilitator type transporter PA2092, that was unable to reduce [Ca²⁺]_in after the initial increase. Ion exchanger PA0397 and probable Na⁺ translocating oxidoreductase PA2999 showed no role in Ca²⁺ homeostasis.

Multiple mechanisms are involved in Ca ²⁺ homeostasis in PAO1

To identify the role of different types of transporters in Ca²⁺ homeostasis and better understand the mutants [Ca²⁺]_in profiles, we examined the effect of several inhibitors specifically affecting different mechanisms associated with ion transport, on the maintenance of [Ca²⁺]_in. For this, PAO1 cells were treated with the selected inhibitors and monitored for their [Ca²⁺]_in response to 1 mM Ca²⁺. First, calcimycin, a Ca²⁺ ionophore known to carry Ca²⁺ into the cells [34], was used to test the response of PAO1 cells to Ca²⁺ influx (Fig. 3A). Treatment with 5 μM calcimycin caused a rapid (36 μM/min) uptake of Ca²⁺ and increase of [Ca²⁺]_in to 9.54 ± 0.15 μM followed by an immediate but slower (4.1 μM/min) decline to 2.36 ± 0.33 μM. The second increase in [Ca²⁺]_in , although appearing 4 min earlier than in the DMSO control, is most likely due to the presence of DMSO (used to dissolve calcimycin).

Figure 3.

Effect of inhibitors on free [Ca²⁺]_in in P. aeruginosa PAO1. Cells grown in BMM media containing no added Ca²⁺ were treated with inhibitors for 10 min at room temperature. The basal level of luminescence was monitored for 1 min. 1 mM CaCl₂ was added at the time indicated by the arrow, followed by luminescence measurements for 20 min. Changes in free [Ca²⁺]_in were calculated as described in the Methods section. A. Calcimycin. Black, 0 mM; dashed black, 5 μM; small dotted grey, solvent control. B. Cells challenged with 2, 4 dinitrophenol. Black, 0mM; grey, 0.5 mM; dashed grey, 1 mM; dashed black, 2 mM. C. Gramicidin D. Black, 0 μg/ml; small dotted grey, 1 μg/ml; dashed black, 10 μg/ml; and grey, solvent control. Vertical lines on the plots indicate the points, at which the remaining available aequorin reached 10 % of the estimated total aequorin. D. LaCl₃. Black, 0 μM; grey, 300 μM, and dashed black, 600 μM. E. Vanadate (2 mM, pH 7.5). The data were averaged from at least three independent experiments.

To test the role of a proton gradient across the cytoplasmic membrane and intracellular ATP in [Ca²⁺]_in response, 2,4-Dinitrophenol (DNP), a proton ionophore known to disrupt a proton gradient and uncouple oxidative phosphorylation [25], was used (Fig. 3B). In a concentration dependent manner, DNP caused a second peak of [Ca²⁺]_in to increase, which at higher levels of DNP appeared to merge with the first peak and reached 7.28 ± 0.74 μM. This level of [Ca²⁺]_in was reached at the rate of 22 fold slower than in calcimycin-treated cells, and was followed by a 32 % decrease. The remaining [Ca²⁺]_in was higher with increasing dose of DNP, and at 2 mM DNP, reached 8.39 ± 1.35 μM.

To examine the role of monovalent cation exchangers in PAO1 Ca²⁺ homeostasis, gramicidin D treatment was used (Fig. 3C). Gramicidin D is known to form channels across the cytoplasmic membrane and dissipate the gradients of H⁺, Na⁺, and K⁺ [35], which may also affect the intracellular ATP pool [36]. In a concentration-dependent manner, gramicidin D treatment elevated the initial increase in [Ca²⁺]_in up to 9.06 ± 1.15 μM, which was followed by a decline to 2.49 ± 0.23 μM. The rates of [Ca²⁺]_in increase and decrease were at least twofold higher at 10 μg/ml than at 1 μg/ml gramicidin D. The [Ca²⁺]_in recovery levels were similar in the treated cells and at least fivefold higher than in the untreated cells. The effect of gramicidin D on [Ca²⁺]_in also included the effect caused by DMSO.

Lanthanum (III) is a Ca²⁺ antagonist, known to block Ca²⁺ channels [37]. It may also arrest Ca²⁺-binding ATPases in a phosphorylated form and prevent further conformational changes required for Ca²⁺ translocation [38]. The effect of LaCl₃ on the level of [Ca²⁺]_in showed two distinct concentration-dependent patterns: a decrease in the initial rise in [Ca²⁺]_in and appearance of the second peak of [Ca²⁺]_in (Fig. 3D). The remaining level of Ca²⁺_in was at least fivefold higher in the LaCl₃ treated cells than in the untreated.

We also tested the effect of vanadate on Ca²⁺ homeostasis in PAO1 (Fig. 3E). Vanadate is known to inhibit P-type [39] and ABC [40] ATPases, and therefore is expected to block Ca²⁺ uptake or efflux mediated by these transporters. Vanadate treatment increased the initial rise of [Ca²⁺]_in by twofold, and the remaining level of [Ca²⁺]_in by fourfold.

Ca²⁺ homeostasis is not involved in PAO1 tolerance to external Ca²⁺

To examine whether the putative Ca²⁺ transporters play a role in cell tolerance to high Ca²⁺, thirteen mutants from groups I - III were grown in the presence of 5 or 100 mM Ca²⁺ and compared to WT. The results showed that at 5 mM Ca²⁺, neither PAO1 nor mutants showed any growth defects. Similarly, the addition of 100 mM Ca²⁺, although introduced a 13 h lag phase, did not significantly affect growth of the wild type or the mutants (data not shown).

Ca²⁺ homeostasis is involved in regulating Ca²⁺-induced swarming motility in PAO1

To determine the role of Ca²⁺ homeostasis in Ca²⁺- induced swarming motility in PAO1, the group III mutants with abolished ability to maintain Ca²⁺ homeostasis were tested for swarming motility at 5 mM Ca²⁺ vs. no added Ca²⁺ (Fig. 4 A). The presence of Ca²⁺ induced swarming in PAO1 by sixfold. When this induction was taken as 100 %, two mutants PA3920:IS, with disrupted P-type ATPase, and PA2092:IS, with disrupted major facilitator protein, showed only 27 % and 67 % of Ca²⁺ induction, respectively. PA2435:IS, with disrupted P-type ATPase, showed no significant difference in swarming, and PA4614:IS, with disrupted mechanosensitive channel, swarmed 26 % further in the presence of Ca²⁺ than PAO1. We also tested the effect of inhibitors targeting P-type and ABC ATPases (vanadate), Ca²⁺ channels (LaCl₃), or dissipating H⁺, Na⁺, and K⁺ gradients (gramicidin D) on Ca²⁺-induced swarming motility (Fig. 4 B). All three inhibitors reduced the effect of Ca²⁺ on swarming. The most significant reduction was observed in the presence of 2 mM vanadate, where Ca²⁺ induction reached only 17 % of the induction in untreated PAO1. The presence of 10 μg/ml gramicidin D or 600 μM LaCl₃ decreased Ca²⁺ induction by 34% and 63%, respectively. The results for LaCl₃ should, however, be taken with a caution, since it precipitated during preparation.

Figure 4.

Swarming motility. Cells were grown on swarming agar containing 0 mM or 5 mM Ca²⁺. A. Group III mutants. B. PAO1 cells. Inhibitors: 2 mM vanadate, 600 μm LaCl₃, or 10 μg/ml gramicidin D were added to the medium prior to plating. Colony diameters were measured, and fold differences (5 mM vs. 0 mM) were calculated using the corresponding controls. The percentage of fold changes was calculated considering that the fold difference in untreated PAO1 was 100 %. The averages of at least three biological replicates were used to calculate the percentage of the fold changes. To avoid bias, in case of mutants, colony diameters were first averaged, and then the fold differences between the mutants and WT were calculated. The standard deviation between replicates was below 10 %.

DISCUSSION

This study establishes that Pseudomonas aeruginosa PAO1, a human pathogen with Ca²⁺- induced virulence, maintains free intracellular Ca²⁺ at sub-micromolar level, which rapidly increases and slowly restores in response to external Ca²⁺. This process is impaired by blocking monovalent cation gradient or modulating Ca²⁺ uptake, and requires several transporters from the superfamilies of P-type ATPases and ion exchangers. These proteins will make an excellent tool for studying Ca²⁺ regulation and signaling in this organism. Finally, the study showed the role of Ca²⁺ homeostasis in Ca²⁺- induced swarming motility, suggesting its regulatory role in P. aeruginosa response to Ca²⁺.

The intracellular level of [Ca²⁺]_in (0.1 – 0.2 μM) and the magnitude (5-28 fold) of the initial response to external Ca²⁺ (1 - 5 mM) places P. aeruginosa among three other bacteria E. coli JM109, Anabaena sp. PCC7120, and B. subtilis [9, 12, 13], whose intracellular Ca²⁺ levels have been characterized. In contrast, Propionibacterium acne, although has a similar basal level of [Ca²⁺]_in, responds to 1mM Ca²⁺ by only twofold increase in [Ca²⁺]_in [10]. The results illustrate that P. aeruginosa is able to maintain sub-micromolar levels of intracellular Ca²⁺ in the presence of millimolar levels of external Ca²⁺, suggesting its ability to (1) block Ca²⁺ uptake, (2) pump out Ca²⁺ from the cells against the steep concentration gradient, or (3) chelate Ca²⁺ inside the cells. The first two abilities in prokaryotes were attributed to the function of several mechanisms including ion channels, P- and F-type ATPases and ion gradient driven transporters that were identified using inhibitors [13, 25] or individual proteins [22].

The detected two-phase [Ca²⁺]_in recovery (short-rapid and long-slow) following the challenge with 1 mM Ca²⁺ may suggest the function of distinct efflux mechanisms of high and low efficiency. The following fluctuations of [Ca²⁺]_in in the presence of 5 mM Ca²⁺ may be a result of continuous cumulative effect of both influx and efflux mechanisms eventually resulting in the recovery of the [Ca²⁺]_in basal level. Comparison of Ca²⁺-induced and naïve cells revealed multiple differences in [Ca²⁺]_in responses. Lower initial [Ca²⁺]_in responses with elevated rates of the [Ca²⁺]_in initial increase and decrease in the presence of 5 mM Ca²⁺ in induced cells suggest that Ca²⁺ induces adaptation mechanisms. However, the elevated basal level of [Ca²⁺]_in in induced cells and their inability to significantly reduce the increased [Ca²⁺]_in at 1 mM Ca²⁺ indicate that growth at elevated Ca²⁺ possibly sensitizes cells to external Ca²⁺, as was proposed in E. coli [9]. Overall, these observations suggest the presence of multiple mechanisms for controlling Ca²⁺ homeostasis in P. aeruginosa and their complex regulation in response to different levels of Ca²⁺ in the environment.

Among several types of transporters determined to be involved in the maintenance of [Ca²⁺]_in homeostasis, the most numerous were P-type ATPases. This superfamily of transporters has long been established to transport metals, but was not initially considered to play a major role in Ca²⁺ transport in prokaryotes. Later, several reports described both Ca²⁺-dependent P-type ATPase activity and ATP-dependent Ca²⁺ transport or transporters, most of which were shown to efflux Ca²⁺ in-vitro in membrane vesicles [21, 41]. However their role in maintaining cellular Ca²⁺ homeostasis in-vivo has not been tested. In this study, out of the seven P-type ATPases predicted in the PAO1 genome, four were induced by Ca²⁺, and six played a role in [Ca²⁺]_in maintenance. These six were phylogenetically related to divalent cation translocating ATPases, whereas PA1634 clustered with K⁺-translocating KdpB from E. coli, showed no effect on [Ca²⁺]_in homeostasis. P-type ATPases are known to be inhibited by vanadate, functioning as a phosphate analog [39]. In agreement, four mutants with disrupted P-type ATPases showed a [Ca²⁺]_in profile similar to the profile of vanadate treated PAO1 cells with increased [Ca²⁺]_in or lowered ability of its recovery. The disruption of two other P-type ATPases PA2435 and PA3920 showed the most significant effect and abolished the ability of PAO1 to reduce [Ca²⁺]_in below 1.5 μM in the presence of 1mM Ca²⁺. PA2435 has been shown earlier to translocate Cu²⁺and Zn²⁺ [32], but in our phylogenetic analysis, it clustered closer to heavy metal-translocating ATPases, whereas PA3920 was more closely related to Cu²⁺-translocating transporters. Noteworthy, these two proteins were not detected to be induced by Ca²⁺, and therefore, likely do not contribute to the increased rate of Ca²⁺ efflux in Ca²⁺-induced cells. Since the identified P-type ATPases contributing to the detected changes in the intracellular Ca²⁺ do not contain typical Ca²⁺-binding domains, they are not likely to play a direct role in translocating Ca²⁺. The ATPases may be involved in controlling transition of metals important for Ca²⁺ translocation or generating an ion gradient that serves as an energy source for Ca²⁺ transporters, as was proposed in A. vinelandii [42].

To characterize the contribution of F-type ATPases (ATP synthases) in Ca²⁺ homeostasis, we treated cells with a protonophore 2,4 DNP, known to inhibit ATP synthesis by uncoupling oxidative phosphorylation [25]. This treatment raised the level of [Ca²⁺]_in and impaired the ability of PAO1 to decrease it, most likely due to the lack of ATP. The role of ATP in Ca²⁺ efflux has been described in E. coli, where the lack of F-type ATP synthase AtpD decreased cellular ATP content and impaired [Ca²⁺]_in efflux [25]. However, the disruption of PA5554, the AtpD homolog with 82 % amino acid sequence identity, although induced by Ca²⁺, did not affect Ca²⁺ homeostasis. It is possible that two paralogs of PA5554 in the genome: PA1697 and PA1104, with 24 and 28 % amino acid sequence identity, correspondingly, could compensate for the absence of PA5554 and provide ATP required for Ca²⁺ efflux.

Gradient driven Ca²⁺ exchangers have been considered as a major mechanism in bacterial Ca²⁺ efflux that most commonly employs H⁺ and Na⁺ as coupling ions. In agreement, treating PAO1 with gramicidin D, known to dissipate monovalent ion gradients across the cytoplasmic membrane [34], caused a significant initial accumulation of [Ca²⁺]_in, which decreased over time, suggesting the function of alternative efflux mechanisms. We predicted and tested the role of four putative gradient driven exchangers in [Ca²⁺]_in homeostasis, three of which showed a significant impact on Ca²⁺ homeostasis. The disruption of PA2092 showed the most prominent effect, as the mutant was not able to decrease the elevated [Ca²⁺]_in in the presence of 1mM Ca²⁺. None of the predicted ion exchangers were induced by Ca²⁺, suggesting that, similarly to P-type ATPases PA2435 and PA3920, they do not play role in the differences between naïve and induced cells. The results suggest that Ca²⁺ translocation may be coordinated by the combined action of both P-type ATPases and ion exchangers as was illustrated for SERCA or PMCA ATPases and Na⁺/Ca²⁺ exchangers in mammalian systems [43, 44].

Treating PAO1 cells with a Ca²⁺ channel blocker LaCl₃ [37] reduced the initial uptake of Ca²⁺, confirming the role of La³⁺ sensitive channels in PAO1 Ca²⁺ influx, as has been shown in other bacteria [9, 13]. The antagonistic effect of La³⁺ on Ca²⁺ influx in E. coli was associated with poly-3-hydroxybutyrate/polyphosphate (PHB-PP) channels [9]. However, search of the PAO1 genome for homologs of the functionally defined PHB-PP genes from Ralstonia [45] returned only scattered homologous genes. This may indicate the presence of a non- or low-homology PHB-PP biosynthetic pathway or a different type of La³⁺ sensitive channels in PAO1. Interestingly, in addition to blocking Ca²⁺ influx, LaCl₃ treatment caused a secondary [Ca²⁺]_in increase and a plateau at about 3 μM. This suggests that La³⁺ also inhibits Ca²⁺ efflux, which is consistent with the known inhibition effect of La³⁺ on P-type ATPases via blocking phosphorylation events [20, 38]. Furthermore, we detected that the disruption of PA4614 encoding Ca²⁺-induced probable mechanosensitive channel MscL caused a very quick uptake of Ca²⁺ followed by a second increase and a slow recovery of [Ca²⁺]_in. In E. coli, although the expression of the homologous gene was also affected by Ca²⁺, its disruption showed no effect on Ca²⁺ homeostasis [25]. Mechanosensitive channels are known to open in response to membrane tension and protect cells against hypoosmotic shock as reviewed in [46]. It is not clear why the lack of the MscL protein caused the increase in Ca²⁺ uptake. We hypothesize that Ca²⁺-induced MscL may contribute to limiting Ca²⁺ uptake when in a closed state under the tested physiological conditions, and when mscL is disrupted, cells become more permeable to Ca²⁺ influx.

Finally, we determined that elevated Ca²⁺ induced swarming motility in PAO1, whereas treatment with vanadate, LaCl₃ and gramicidin D decreased this induction. The disruption of P-type ATPase PA3920 and ion exchanger PA2092 also significantly diminished the inducing effect of Ca²⁺ on swarming. The effect of Ca²⁺ on swarming motility varies in different bacteria, for example, Ca²⁺ has been shown to induce swarming in Vibrio parahaemolyticus [47], but inhibited this type of motility in fluorescent pseudomonads [48]. This study shows for the first time that Ca²⁺ homeostasis plays an important role in P. aeruginosa Ca²⁺ induced swarming, a complex physiological phenomenon known to modulate virulence and antibiotic resistance in this organism [49].

Overall, this study reports that P. aeruginosa PAO1 maintains a sub-micromolar basal level of [Ca²⁺]_in using multiple transport mechanisms that most likely require ATP and monovalent cation gradient. Involvement of multiple transport systems in Ca²⁺ influx and efflux has been suggested in E. coli [9, 50], and may be differentially regulated by growth conditions. This functional redundancy was also illustrated by the high tolerance of PAO1 cells to external Ca²⁺, and reflects the physiological importance of the controlled cellular Ca²⁺ homeostasis. The two major aspects of such importance include protecting cells against the toxicity of high [Ca²⁺] and maintaining low basal level of [Ca²⁺]_in required for Ca²⁺ to play a signaling role. The transient changes in [Ca²⁺]_in (magnitude, length, and frequency) may function to relate external signal(s) to cellular response(s), as has been established in eukaryotes [1] and proposed in prokaryotes [5]. We identified at least three transporters that play a major role in [Ca²⁺]_in homeostasis, and will be used in further studies required for experimental confirmation of signaling role of Ca²⁺ in P. aeruginosa.