BqsR/BqsS Constitute a Two-Component System That Senses Extracellular Fe(II) in Pseudomonas aeruginosa

Bacterial strains used in this study are listed in Table 1. Stationary-phase cultures of P. aeruginosa strains were grown aerobically at 37°C in 3 g/liter Bacto tryptic soy broth containing 100 mM KNO₃. The overnight cultures were diluted to an optical density at 500 nm (OD₅₀₀) of 0.01 in minimal metal medium (MMM). MMM consists of 0.3 g/liter Bacto tryptic soy broth, 50 mM glutamate, 1% glycerol, 100 mM KNO₃, and 50 mM MOPS (morpholinepropanesulfonic acid); the pH was adjusted to 7.0 with NaOH, and the medium was treated with 1% (wt/vol) Amberlite chelating resin (Sigma) for 1 h and filtered into acid-washed medium bottles prior to autoclaving. After autoclaving, MMM was immediately sparged to maintain anaerobic conditions and brought into a vinyl anaerobic chamber containing a 5% hydrogen, 15% carbon dioxide, and 80% nitrogen gas mixture (Coy Laboratory Products). Ten milliliters of medium was added to 25-ml anaerobic Balch tubes, which were acid-washed in 1 M HCl to remove residual metals. Unless otherwise specified, inoculations were performed anaerobically in the anaerobic chamber with the previously stated gas mixture and incubated and shaken at 37°C. Spectrophotometric measurements were performed on a Beckman Spec20 unless otherwise specified.

The bqsR unmarked deletion was created using a pMQ30-based construct, which was made by applying the Saccharomyces cerevisiae-based molecular tool kit (47) (Table 1). This procedure takes advantage of yeast homologous recombination to combine the vector with PCR-amplified regions upstream and downstream from the gene of interest. The 5′ region (∼1 kb in length) of the sequence flanking bqsR was amplified using primers JW1_F and JW1_R, and the 3′ region (∼1 kb in length) of the sequence flanking bqsR was amplified using primers JW2_F and JW2_R (see Table S1 in the supplemental material). Both primer pairs contained regions that were homologous to each other, and the 5′ ends of the JW1_F and the JW2_R primers contained 40-bp regions of homology to pMQ30.

S. cerevisiae strain InvSc1 (Table 1) was cotransformed with plasmid linearized with EcoRI and BamHI as well as with the upstream and downstream PCR fragments; recombinant plasmids were generated by yeast homologous recombination (47). Recombinants were selected on synthetic defined medium lacking uracil (SD−Ura). Plasmids were liberated from uracil-auxotrophic yeast using the QIAprep Spin miniprep kit (Qiagen) yeast protocol and electroporated into Escherichia coli DH5α. Plasmids obtained from DH5α were electroporated into E. coli BW29427. The deletion plasmid was mobilized from BW29427 into PA14 using biparental conjugation (64). PA14 single recombinants (merodiploid containing the intact bqsR gene and a deleted gene) were selected on LB agar containing 100 μg/ml gentamicin. These colonies were then grown in the absence of selection to resolve merodiploids. Potential bqsR deletion mutants (resolved merodiploids) were identified by selecting for colonies that grew in the presence of 10% sucrose. Strains with properties of double recombination were further analyzed by PCR and sequenced to verify bqsR deletion (ΔbqsR). The above protocol was performed for the bqsS deletion (ΔbqsS), substituting primers dbqsS 1 and dbqsS 2 for primers JW1_F and JW1_R and dbqsS 3 and dbqsS 4 for primers JW2_F and JW2_R (see Table S1 in the supplemental material for primer sequences). Because the reading frames in the bqs operon overlap, we were careful to construct our deletions in this region so as to both render them nonpolar and preserve the integrity of the remaining genes in the operon.

The GeneChip P. aeruginosa genome array (Affymetrix) contains probe sets for more than 5,500 genes from P. aeruginosa PAO1 and 117 additional genes from strains other than PAO1. Due to the strong similarity between strains PAO1 and PA14, the Affymetrix array has been used for gene expression analysis of both strains (12, 33, 63). P. aeruginosa PA14 was grown in anaerobic MMM as described in “Bacterial strains and growth conditions.” The cultures were grown anaerobically at 37°C until they reached exponential phase (OD₅₀₀ of 0.3), at which time they were removed from the incubator and were taken into the anaerobic chamber containing a gas mixture of 5% H₂ and 95% N₂ (Coy Laboratory Products). Water, 100 μM FeCl₃, or 100 μM FeCl₂ was added to the cultures. Iron concentrations were calculated by weight, assuming 100% purity of Fe(III) and Fe(II) stocks. The cultures were vortexed every 5 min over a 30-min period, after which 5 ml of culture was removed and mixed with 10 ml of bacterial RNAprotect (Qiagen). The cultures were incubated for 5 min at room temperature before cells were pelleted (10 min, 5,000 × g). DNA-free RNA was isolated from the cell pellet using the RNeasy minikit (Qiagen) that includes on-column DNase treatment. The preparation of the cDNA, as well as the processing of the P. aeruginosa GeneChip arrays, was performed by the BioMicro Center at the Massachusetts Institute of Technology (MIT) using an Affymetrix fluidics station. GeneChip arrays were performed in biological triplicate. Signal intensities were background corrected, normalized, and summarized using the robust multiarray averaging (RMA) method in the Bioconductor package affy (8, 14, 22). Linear models for each gene were fit using the program limma, and moderated t statistics were calculated with limma's empirical Bayes approach (50). P values were corrected for multiple testing using the Benjamini-Hochberg method to control the false-discovery rate (6); these adjusted values are referred to as q values. A threshold q value of 0.01 was used to identify differentially expressed genes, resulting in a false-discovery rate of 1% among genes called significant. Verification of GeneChip data was performed with quantitative reverse transcriptase PCR (qRT-PCR).

Fe(II) readily oxidizes in the presence of trace levels of oxygen. Different Fe(II) salts have different stabilities in water. For most experiments, the Fe(II) source was ferrous chloride; it was used because the anion, chloride, was expected to have little effect on the metabolism of PA14. However, this form is less stable in water than ferrous ammonium sulfate. To be certain of the total iron concentration and to distinguish between Fe(II) and Fe(III), the ferrozine assay (56) was used. Ferrozine binds to Fe(II) and provides a colorimetric readout of Fe(II) concentration compared to that of a standard. Ferrous ammonium sulfate in acid was used as a standard because it is the most stable form of Fe(II) and, under acidic conditions, all the iron is expected to remain in the Fe(II) state. Iron-bound ferrozine was measured at 562 nm. Total iron concentration was determined by reducing the sample with hydroxylamine hydrochloride to convert all Fe(III) to Fe(II).

Genes found to be strongly induced in the array experiment were examined using qRT-PCR (primers are listed in Table S1 in the supplemental material). P. aeruginosa strains listed in Table 1 were grown anaerobically in 10-ml cultures in triplicate as described in “Bacterial strains and growth conditions.” The Fe(II) shock was performed in the anaerobic chamber after the cultures grown in MMM reached early exponential phase. Five-milliliter samples of each culture were removed immediately before the Fe(II) shock and mixed with 2 volumes of bacterial RNAprotect (Qiagen), incubated for 5 min at room temperature, and centrifuged for 10 min at 5,000 × g. These served as the Fe-free control. A 50 μM concentration of Fe(NH₄)₂(SO₄)₂ was added to the cultures, mixed, and incubated for 30 min. After the 30-min shock, the remaining 5-ml culture was immediately added to 10 ml RNAprotect, incubated for 5 min at room temperature, and centrifuged for 10 min at 5,000 × g. The pellets were stored at −80°C until RNA isolation, and qRT-PCR analysis was performed.

PA14 was grown in MMM to early exponential phase and treated as described in “Fe(II) shock experiments,” substituting Fe(III) and various dipositive cations for Fe(II). The metals tested were 100 μM Fe(NH₄)₂(SO₄)₂ [containing 100 μM Fe(II), as determined by weight and the ferrozine assay], 100 μM FeCl₃, 5 mM CaCl₂, 100 μM MnCl₂, 10 mM MgCl₂, 100 μM ZnCl₂, and 100 μM CuCl₂. The concentrations of calcium and magnesium were selected to approximate physiologically relevant concentrations (27).

PA14 was grown in MMM aerobically to early exponential phase and shocked with the metal concentrations given above. The Fe(II) concentration in the Fe(NH₄)₂(SO₄)₂ was determined by the ferrozine assay before and after the aerobic shock to determine if any iron remained in the Fe(II) state after the 30-min exposure to oxygen. The initial Fe(II) concentration was 100 μM, and the concentration after the 30-min shock was 4.9 μM.

Triplicate samples of 50 ml MMM in 125-ml acid-washed serum vials were inoculated from cultures grown as described in “Bacterial stains and growth conditions” and sealed. When the cells reached early exponential phase, the Fe(II) shock was performed. Five milliliters of culture was treated with RNAprotect before the iron shock and used as the no-iron control. Subsequently, 5 ml of culture was added to tubes containing iron and incubated for 30 min. The Fe(II) concentrations assayed were 6 μM, 19 μM, 31.6 μM, 38 μM, 51 μM, and 63 μM (10 μM, 30 μM, 50 μM, 60 μM, 80 μM, and 100 μM FeCl₂ by weight, respectively). After the shock, the cultures were added to 10 ml RNAprotect for RNA stabilization.

Total RNA was extracted from the cell pellet using the RNeasy minikit (Qiagen), including the optional DNase treatment step, according to the manufacturer's instructions. cDNA was generated using the extracted RNA as a template for an iScript (Bio-Rad) random-primed reverse transcriptase reaction by following the manufacturer's protocol. The cDNA was used as a template for quantitative PCR (Real Time 7500 PCR machine; Applied Biosystems) using SYBR green with the ROX detection system (Bio-Rad). Samples were assayed in biological triplicate. The threshold cycle (C_T) values of recA and clpX were used as endogenous controls (12). Fold changes were calculated using the ΔΔC_T method. Briefly, all samples were normalized to each other by subtracting the C_T value for the control gene recA in the shock condition from the value in the control condition (ΔC_T recA). This value was then converted from the log₂ using the following formula, which accounts for machine error: a = 2^{ΔCT recA}. The change in the genes of interest was calculated by subtracting the C_T value of the gene in the shock condition from the C_T value for the gene in the control condition (ΔC_T gene). This value was then converted from the log₂ using the following formula: b = 2^{ΔCT gene}. The final relative fold change is calculated by the following formula: fold change = b/a. To ensure recA was constant under all conditions tested, the relative fold change for the internal control clpX, whose expression was also expected to remain constant across all our treatments, was calculated as described above. Only those samples with a clpX fold change between 0.5 and 2 were used. Primers (Integrated DNA Technologies) were designed using Primer3, with settings modified for qRT-PCR primers.

PA14 was grown aerobically on LB agar plates at 37°C. Two single colonies were picked, and each was transferred to a fresh tube containing 3 ml of sterilized MOPS minimal medium (50 mM MOPS, 2.2 mM KH₂PO₄, 43 mM NaCl, 93 mM NH₄Cl, 40 mM succinate, 1 mM MgSO₄, and 3.6 μM FeSO₄ · 7H₂O; pH 7.1) and grown overnight at 37°C. One milliliter of turbid inoculum was then added to 50 ml of sterilized MOPS medium with 500 μM FeCl₃ · 6H₂O and incubated at 37°C in triplicate, and a low-iron control culture was inoculated in MOPS minimal medium without additional Fe(III). At given time points, aliquots were removed from the culture for OD measurements, pyocyanin (a phenazine) concentration measurements, tests to distinguish the iron oxidation state, and qRT-PCR measurements of the bqs operon. A sample of 500 μl was taken for OD measurements, 1 ml was frozen for subsequent high-performance liquid chromatography (HPLC) analysis, 1.6 ml was taken to distinguish between iron oxidation states, and 2 to 5 ml was taken for RNA isolation. The OD₅₀₀ was measured in triplicate using the Beckman DU 800 spectrophotometer. Pyocyanin HPLC peaks were identified from experimental samples and quantified with known concentrations of a pyocyanin reference on a Beckman System Gold instrument with an acetonitrile gradient, a Waters Symmetry C₁₈ reverse-phase column (5-μm particle size; 4.6 by 250 mm), and a 168 diode array detector. The ferrozine assay was used to determine the iron concentrations and the oxidation states of the samples. All iron oxidation state distinctions were conducted in an anaerobic chamber to avoid reoxidation of ferrous ions, and measurements were made with a Biotek Synergy 4 plate reader. Relative fold changes for qRT-PCR were calculated by comparing samples from MOPS minimal medium plus Fe to samples with the same OD₅₀₀ from MOPS minimal medium without Fe.

Our microarray data have been deposited in EBI Array Express (accession number E-MEXP-3459).

To determine if exogenous Fe(II) and Fe(III) are perceived differently at the cellular level by P. aeruginosa strain PA14, we used Affymetrix P. aeruginosa GeneChips to identify transcriptional changes in response to Fe(II) or Fe(III). To ensure that the iron oxidation state remained stable, strain PA14 was cultured anaerobically as described in Materials and Methods. Upon reaching exponential phase, triplicate cultures were exposed to approximately 100 μM Fe(II), 100 μM Fe(III), or water for 30 min, and total RNA was extracted, labeled, and hybridized to P. aeruginosa Affymetrix GeneChips.

In both the Fe(II) and Fe(III) shock conditions, relative to the no-iron control, iron-scavenging genes (e.g., pyochelin biosynthesis genes) were downregulated, as expected (see the supplemental material). To determine if any genes specifically responded to Fe(II), we compared the Fe(II) and Fe(III) shock treatments. Cultures exposed to extracellular Fe(II) displayed an expression profile that was distinct from that of Fe(III)-exposed cultures (Table 2). We found 3 genes that were downregulated and 18 genes that were upregulated at least 2-fold by Fe(II) with q values of ≤0.01. Of the upregulated genes, a putative operon containing two short putative membrane proteins with PepSY domains and a putative two-component system showed the highest upregulation (Table 2).

To determine whether these genes constitute an operon, we performed RT-PCR. All of the genes from PA14_29710 to PA14_ 29750 appear to be transcriptionally connected (Fig. 1), and thus we refer to this operon as the bqsPQRST operon, because homologs of BqsS and BqsR (exhibiting 100% amino acid identity) were previously identified in P. aeruginosa strain PAO1 for their role in biofilm development and quorum sensing (13). Whether bqsT is a bona fide member of this operon is less clear, as qRT-PCR results (see Fig. 3) show lower fold change in this gene than in other members of the operon, possibly reflecting transcriptional bleed-through. Moreover, whereas all of the other members of the operon possess overlapping reading frames, there are 90 bp between bqsS and bqsT. The five-gene cluster is conserved in its entirety only among P. aeruginosa strains. Orthologs of the sensor BqsS are encoded in similar predicted operons with a response regulator and one or two PepSY domain-containing proteins across other Pseudomonas and Azotobacter species, but the fifth hypothetical gene is excluded.

It was striking that exogenous Fe(II) induced a two-component system, because such systems are responsible for transmitting extracellular signals (e.g., the presence of ferrous iron) to the intracellular environment (55). This suggested that our two-component system, comprising a cytoplasmic response regulator (bqsR) and a sensor histidine kinase (bqsS) and identified by its conserved kinase domain and its predicted localization to the inner membrane, was sensing extracellular Fe(II). To test this, we used qRT-PCR to measure the expression of bqsR and bqsS in response to Fe(II) shock in the wild-type (WT) strain and in strain feoB::MAR2xT7 [which has a severe defect in Fe(II) uptake across the cytoplasmic membrane (31, 32, 61)]. If the cells were sensing extracellular Fe(II), we anticipated that bqsR and bqsS expression would be the same in both the wild type and the feoB mutant. Alternatively, if the cells were sensing cytoplasmic Fe(II), we expected bqsR and bqsS to be upregulated in the wild type compared to their expression in the feoB mutant. P. aeruginosa strains PA14 and PA14 feoB::MAR2xT7 were cultured under anaerobic conditions until early exponential phase, at which time 63 μM Fe(II) was added to the medium. After 30 min, the RNA was stabilized and extracted for qRT-PCR. As seen in Fig. 2, bqsR and bqsS were upregulated in both the wild type and the feoB mutant upon Fe(II) shock, which indicates their expression is activated in response to extracellular Fe(II).

To gain insight into the mechanism of the extracellular Fe(II) shock transcriptional response, we examined expression of the nine genes most upregulated in the microarray (each having a fold change of ≥4) in strains lacking either the regulator (ΔbqsR mutants) or the sensor (ΔbqsS mutants) (Fig. 3). Four of these genes are components of the bqs operon (bqsPQRST). Under anaerobic conditions, the ΔbqsR mutant, ΔbqsS mutant, and wild type were grown to early exponential phase and exposed to Fe(II). RNA was extracted for qRT-PCR analysis before and after 30 min of exposure. Our results reveal that BqsR/BqsS is autoregulated. When both BqsR and BqsS are present, extracellular Fe(II) induces the upregulation of the entire bqs operon (Fig. 3). However, if either bqsR or bqsS is deleted, the bqs operon no longer responds to extracellular Fe(II). Complementation with bqsR expressed on pMQ64 resulted in an Fe(II) response similar to that of the wild type (Fig. 3). The exception to this is expression of bqsR itself, which does not exhibit the same dramatic increase in expression with Fe(II) shock; this can be explained by the fact that before Fe(II) shock, bqsR is upregulated 75-fold compared to wild-type levels because its expression is being driven off a multicopy plasmid. No data are presented for bqsR in both the ΔbqsR mutant and the vector-only control because no copies of bqsR are present and any values detected are within background error. The transcriptional responses of the five most highly upregulated genes outside this operon—PA14_04180, PA14_07070, PA14_01240, PA14_01250, and PA14_04270—also depended on bqsR and bqsS. Of these genes, PA14_04180 shows the highest upregulation by qRT-PCR (Fig. 3) and was the fourth most upregulated in the microarray. In general, the strength of the transcriptional response in our qRT-PCR experiments was significantly higher than what we observed in the array study (Table 2), which is not unexpected because qRT-PCR is a more sensitive technique than microarrays.

Intrigued by the differential fold changes we observed in our qRT-PCR experiments, we hypothesized that the BqsS/BqsR-mediated transcriptional response was tunable by the Fe(II) concentration. To test this, we performed an Fe(II) titration experiment, varying Fe(II) over an order of magnitude from ∼6 μM to ∼60 μM. A concentration of ∼20 μM was the lowest concentration of Fe(II) to elicit a response for bqsR or bqsS; the maximum response was seen for >60 μM Fe(II) (Fig. 4).

Because some two-component systems are induced by multiple external signals, we questioned whether BqsR/BqsS was specific for Fe(II). To determine this, we performed transcriptional shock experiments with five other dipositive cations that are commonly found in CF sputum (16). We also measured the transcriptional response to Fe(III) as a control. The wild type was cultured under either anaerobic or aerobic conditions and exposed to metal shock in early exponential phase. Fe(III), Ca(II), Cu(II), Mg(II), Mn(II), and Zn(II) do not significantly upregulate bqsS under either aerobic or anaerobic conditions (Fig. 5). From this, it appears that BqsR/BqsS is selective for ferrous iron. Notably, the extent of bqsS gene expression was attenuated in the aerobic Fe(II) shock experiment (Fig. 5). The lower degree of upregulation can be explained by rapid oxidation of Fe(II) to Fe(III): the initial concentration of Fe(II) was 100 μM, but after completion of the 30-min shock, the concentration was 4.9 μM.

To explore the connection between bqs operon transcriptional induction and biological iron reduction in a wild-type culture, a 27-h growth experiment was conducted in duplicate using PA14 grown in MOPS medium with an initial iron allocation of almost exclusively Fe(III) (Fig. 6). OD₅₀₀ measurements show a classical growth curve exhibiting exponential growth between approximately 10 and 18 h after inoculation. Pyocyanin, a type of phenazine, was detected in the culture starting in mid-exponential phase and continuing through the end of the experiment. The culture's iron oxidation state shifted from entirely Fe(III) at the beginning of the experiment to >85% Fe(II) by the end; this shift coincided with the exponential phase of growth and the presence of phenazines. qRT-PCR showed upregulation of bqsS during exponential phase, when extracellular Fe(II) was at a concentration of ∼125 μM (well above the concentration needed to induce the bqs operon), and a repression of transcription in stationary phase.