Introduction

The pathogen Pseudomonas aeruginosa produces several life-threatening infections in humans, especially in immunocompromised, cancer, burn and cystic fibrosis patients and it is one of the primary reasons of hospital-acquired infections1,2,3,4. Importantly, antibiotic resistance among this pathogen has escalated globally over the past three decades and several outbreaks in hospitals have highlighted the need of controlling multi-drug resistant P. aeruginosa infection and spread5. Indeed, the World Health Organization has declared this bacterium the second priority pathogen demanding research and development of new treatment strategies. Therefore, there is an enormous research need to identify new molecular targets that permit the inhibition or elimination of this pathogen.

P. aeruginosa is highly metabolic versatile and harbors multiple virulence factors that enable this pathogen to infect essentially any mammalian tissue3,6. Central to the infectious process is the ability of the pathogen to adapt to changing environments and P. aeruginosa produces many global regulators and signal transduction systems that facilitate its adaptation7,8. Regulation of gene expression in bacteria occurs initially at the transcription initiation level through the modulation of the affinity of the RNA polymerase (RNAP) for the DNA. Such affinity can be modified through the replacement of the sigma (σ) subunit of the RNAP, which is the subunit responsible of promoter recognition and thus of the specificity of the RNAP, and/or by transcriptional regulators that enhance or repress RNAP binding and activity9. P. aeruginosa contains a plethora of these regulatory proteins, which often function in response to specific cues. Among them, sigma factors of the extracytoplasmic function sigma (σECF) factor family are important signal-responsive regulatory proteins in P. aeruginosa10,11. The σECF-mediated signaling in this bacterium generally involves the function of an anti-σ factor10,11. Most P. aeruginosa anti-σ factors are single-pass transmembrane proteins that contain a cytosolic N-terminal domain that binds the σECF factor and occludes the RNAP binding determinants, and a periplasmic C-terminal domain required for signal transduction. In response to a specific inducing signal, the anti-σ factor usually undergoes regulated proteolysis12,13,14,15, which leads to the release of an active σECF factor that binds to the RNAP and promotes transcription of the signal response genes.

P. aeruginosa contains between 19 and 21 σECF factors that cluster into nine different phylogenetic groups10. Most belong to the iron starvation (IS) group and are expressed in iron limiting conditions together with an anti-σ factor. Post-translational activation of IS σECF factors often occurs in response to the presence of an iron chelating compound (i.e. siderophores, heme/hemoglobin, iron-citrate) by a signal transduction cascade known as cell-surface signaling (CSS) that also involves an outer membrane-located TonB-dependent transducer (TBDT)10,16,17. IS σECF factors promote transcription of iron acquisition functions and regulate iron homeostasis, which are essential processes for P. aeruginosa to spread and colonize the host. Besides, several P. aeruginosa IS σECF factors stimulate the transcription of virulence determinants10,11,16. The second most abundant σECF group in P. aeruginosa is formed by the RpoE-like σECF factors. These σ factors are activated in response to cell envelope stress and trigger expression of functions that mitigate stress and maintain the integrity of the bacterial cell envelope, thus ensuring pathogen survival10,11. While required during infection to cope with stresses produced by the host immune response (e.g. increased temperature, formation of oxygen reactive species or osmotic changes), P. aeruginosa stress-responsive σECF factors also promote expression of important virulence determinants (i.e. the exopolysaccharide alginate)10,11. The signaling cascade activating these σECF factors usually involves an anti-σ factor but not an outer membrane TBDT10,11.

The P. aeruginosa σVreI factor was initially classified within the IS group18. However, our recent analyses showed that expression of this σ factor is not regulated by iron, but by inorganic phosphate (Pi)19,20. This was in agreement with our initial observations showing that σVreI does not promote expression of iron acquisition functions21. σVreI is encoded by the vreAIR operon together with a CSS-like receptor protein (VreA) and a transmembrane anti-σ factor (VreR)19,21. While the anti-σ role of VreR has been demonstrated19, the function of VreA in the σVreI signaling cascade, if any, is at present unknown. The N-terminal domain of VreA resembles that of CSS receptors21, which is the domain that interacts with the anti-σ factor upon signal recognition triggering activation of the CSS cascade and the IS σECF factor16. However, VreA lacks the C-terminal β-barrel domain of CSS receptors, which is the domain required for the uptake of the CSS ligand (i.e. siderophore, heme)21. We initially hypothesized that VreA could be involved in signaling but not in transport21; however, our recent analyses showed that, in vitro, VreA is not required for σVreI activation19. Transcription of the vreAIR operon takes place in Pi limited conditions and requires the phosphate transcriptional regulator PhoB19. Besides promoting vreI transcription, PhoB is also required for transcription of the σVreI regulon genes by recruiting the σVreI-RNAP complex to the promoter region of these genes19. Our earlier microarray analyses showed that the σVreI regulon includes several P. aeruginosa virulence determinants (Fig. 1)21. In accordance, constitutive activation of σVreI increases P. aeruginosa pathogenicity21. Antibodies directed against secreted proteins of the σVreI regulon (i.e. PdtA, Fig. 1) are detected in the serum of P. aeruginosa infected patients21. Moreover, interaction of P. aeruginosa with host cells promotes transcription of σVreI regulon genes22,23. Together this suggests that transcription of the σVreI regulon occurs during infection and thus that σVreI is active in this condition. However, there is no direct proof yet demonstrating the activation of σVreI in vivo.

Figure 1
figure 1

Genetic organization of the σVreI regulon. Transcriptional organization of the vreAIR locus (black) and the downstream σVreI-regulated genes (colored). Block arrows represent the different genes, their relative sizes, and their transcriptional orientation, with the name of the gene or the PA number (http://www.pseudomonas.com/) indicated below the arrow. The promoters and regulatory boxes identified within this locus are indicated19,20,62. Numbers indicate the fold-change in the expression of the gene in cells overproducing σVreI as determined earlier by microarray21. The hxc genes (dark green) encode a type II secretion system involved in the secretion of the low molecular weight alkaline phosphatase LapA (light green)62. pdtA and pdtB (blue) encode a functional two-partner secretion (TPS) system involved in P. aeruginosa virulence in the C. elegans model63. phdA (yellow) encodes a homologue of the prevent-host-death (Phd) protein family and is required for biofilm formation and eDNA release64. exbB2-exbD2-tonB4 genes (orange) encode a still uncharacterized putative TonB system. The function of the PA0696-PA0700 gene products (purple) is still unknown. PA0701 (dark grey) encodes a putative LysR-like transcriptional regulator and PA0701a (light grey), which is not annotated in the PAO1 genome but it is in the P. aeruginosa PA14 genome, encodes a putative AraC-like transcriptional regulator.

The aim of this work was to investigate whether the σVreI factor is active during infection. Moreover, because vreI is expressed under Pi starvation, a condition often encountered by pathogens in the host environment that is known to induce a virulent phenotype in P. aeruginosa24, we also aimed at determining the contribution of this σ factor to the Pi starvation-induced virulence of this pathogen. Using zebrafish embryos and a human respiratory epithelial cell line as P. aeruginosa hosts we show for the first time that σVreI is indeed activated during infection and that lack of the σVreI/VreR signaling proteins diminishes the Pi starvation-induced virulence of this pathogen.

Results

σVreI and VreR are required for P. aeruginosa virulence in zebrafish embryos

The virulence of P. aeruginosa was assayed using zebrafish (Danio rerio) embryos as host. P. aeruginosa is able to lethally infect zebrafish embryos when the number of bacterial cells injected exceeds the phagocytic capacity of the embryo21,25,26. Because the expression of the vreAIR operon is induced under phosphate (Pi) starvation19,20, we first determined the effect of the Pi concentration in P. aeruginosa virulence. Thus, P. aeruginosa cells previously grown either in low or high Pi conditions were injected into the blood stream of one-day post-fertilization embryos to generate a systemic infection and embryo survival was monitored during five days. Survival of the embryos injected with PAO1 wild-type cells grown in low Pi was considerably lower than that of the embryos injected with PAO1 cells grown in high Pi conditions (P < 0.001) (Fig. 2A), which indicates that Pi starvation induces a P. aeruginosa virulent phenotype. The contribution of σVreI to this phenotype was determined using a null ΔvreI mutant. The virulence of this mutant was significantly lower than that of the PAO1 wild-type strain (P < 0.05) (Fig. 2B), which suggests that σVreI contributes to the low Pi-induced virulence of P. aeruginosa. Unexpectedly, a ∆vreR anti-σ factor mutant in which σVreI is highly active19 also showed attenuated virulence (P < 0.001) (Fig. 2B). This prompts that VreR may have more functions in P. aeruginosa than solely controlling σVreI activity or alternatively, that the timing and/or quantity of the σVreI response is crucial. In contrast, the absence of the CSS-like receptor protein VreA did not have any effect on P. aeruginosa virulence, which was similar for the ΔvreA mutant and the PAO1 strain (P = 0.35) (Fig. 2B). This suggests that this protein is not involved in the σVreI/VreR-mediated virulence of this bacterium.

Figure 2
figure 2

Survival of zebrafish embryos upon infection with P. aeruginosa. One day post-fertilization embryos were injected with 1000 CFU of the P. aeruginosa PAO1 wild-type strain grown either in Pi-restricted or Pi-sufficient conditions (A) or with the indicated PAO1 isogenic mutant grown under Pi starvation (B). Uninfected control (non-injected) is shown. Data represents the mean ± SD of four biologically independent replicates (N = 4) with 20 embryos/group in each replicate. P-values were calculated by log-rank (Mantel-Cox) test.

σVreI and VreR are required for P. aeruginosa cytotoxicity

P. aeruginosa commonly affects the respiratory tract in humans. Therefore, we used the A549 human respiratory epithelial cell line as P. aeruginosa host. First we assayed the cytotoxicity of P. aeruginosa against the eukaryotic cells by determining A549 cell viability after co-incubation with the bacteria. Similar to the results obtained in the zebrafish embryos, we observed that growth of P. aeruginosa under Pi starvation slightly increases the bacterial cytotoxicity as the eukaryotic cells were more damaged by bacteria grown in low rather than in high Pi medium (Fig. 3A). Both σVreI and VreR contribute to this phenotype, because mutants lacking these proteins were significantly less efficient in damaging the A549 cells than the PAO1 wild-type strain (Fig. 3A).

Figure 3
figure 3

P. aeruginosa infections in the A549 cell line. (A) A549 cell viability. The P. aeruginosa PAO1 wild-type strain and the indicated isogenic mutant were grown in Pi-restricted (−) or Pi-sufficient (+) conditions prior to infection. Formazan production upon addition of the MTT tretazolium salt was determined spectrophotometrically at 620 nm. Uninfected cells (white bar) were used as control. (B) P. aeruginosa internalization into A549 cells. A549 cells were infected with the indicated P. aeruginosa strain previously grown in Pi-restricted conditions. Internalization is reported as the ratio between bacteria CFU inside (in) the A549 cells and CFU in the culture supernatant (out). In both panels data are means ± SD from three biological replicates (N = 3). P-values were calculated by unpaired two-tailed t-test as described in Materials and Methods and brackets indicate the comparison to which the P-value applies.

We also measured internalization of P. aeruginosa into A549 cells. Although considered an extracellular pathogen, P. aeruginosa is able to enter into non-phagocytic host cells such as epithelial cells27,28. The internalization efficiency of a given strain depends on several factors including the T3SS profile29,30. Strains that are more efficient in internalizing are less cytotoxic while less invasive strains kill the eukaryotic cells more rapidly. Therefore, there is an inverse correlation between internalization and cytotoxicity, and thus between internalization and acute virulence31. Upon P. aeruginosa infection of A549 cells, we observed a 3-fold increase in the internalization of the ∆vreI and ∆vreR mutants as compared to the PAO1 wild-type strain (Fig. 3B). This is consistent with the lower cytotoxicity displayed by these mutants (Fig. 3A) and with the reduced capacity of these mutants to produce a systemic infection in zebrafish embryos (Fig. 2B).

VreR influences P. aeruginosa gene expression also in a σVreI-independent manner

The phenotype of the ΔvreR mutant, in which reduced instead of increased virulence was observed, prompted us to analyze gene expression in this mutant by RNA-seq. We compared the transcriptome of the P. aeruginosa PAO1 wild-type strain with that of the ΔvreR mutant upon bacterial growth in Pi starvation, a condition known to induce vreR expression19. A total of forty-four transcripts were more abundant in the ΔvreR mutant than in the PAO1 wild-type strain, and nine were less abundant (including the vreR transcript) (Table 1). Increased expression of some of these genes in the ΔvreR mutant was confirmed by qRT-PCR (Fig. S1). Of the forty-four upregulated genes, nineteen are located immediately downstream the vreAIR locus and most of them belong to the σVreI regulon previously identified by microarray of cells overproducing σVreI (Fig. 1)21. Other upregulated transcripts belong to genes located in different loci in the P. aeruginosa PAO1 genome (Table 1). These include genes encoding metabolic and energy obtaining functions, transport, and several regulators of gene expression (Table 1).

Table 1 Differentially expressed P. aeruginosa genes in ΔvreR versus PAO1a.

We then wondered whether the increased expression of these genes in the ΔvreR mutant was due to σVreI, which is highly abundant and active in this mutant19. Therefore, we compared the relative expression of these genes in the ΔvreR mutant with that of the ΔvreI mutant in which vreR is also not produced due to a polar effect of the vreI deletion on the expression of the downstream vreR gene19 (referred as a ΔvreI vreR mutant in Fig. 4). A total of twenty-five genes were selected, eight known to belong to the σVreI regulon (Fig. 1). Indeed, these eight genes were expressed considerably less in the ΔvreI vreR mutant than in ΔvreR (Fig. 4). This shows that the higher expression of these genes in the ΔvreR mutant directly depends on σVreI. However, the relative expression of eight genes that were not previously identified as part of the σVreI regulon (PA0141, ccoO2, PA1673, PA1746, PA3880, PA4348, dadA and adhA) was similar in both mutants (Fig. 4). This indicates that the expression of these genes is not affected by the absence of σVreI and therefore that this σ factor is not involved in their transcription. Finally, the expression of nine other genes (PA0200, ddaR, PA1414, hemN, phrS, rfaD, oprG, feoA and PA5475) was higher in the ΔvreI vreR mutant than in ΔvreR (Fig. 4). This suggests that σVreI not only does not mediate the transcription of these genes, but also that presence of this σ factor impairs their expression. Overall, this analysis suggests that VreR has a broader role in gene regulation beyond its direct involvement in the σVreI signaling pathway.

Figure 4
figure 4

Differential gene expression in the P. aeruginosa ΔvreR and ΔvreI vreR mutants. mRNA levels of the indicated genes were obtained by qRT-PCR upon growth of the P. aeruginosa mutants in low Pi medium. The 2−ΔΔCT method was used to determine the fold-change range in gene expression in ΔvreI vreR versus ΔvreR. Data are means ± SD from three biological replicates (N = 3) each one including three technical replicates. P-values were calculated by one-sample t-test to a hypothetical value of 1 as described in Materials and Methods.

σVreI is activated in vivo during infection

To analyze whether σVreI is activated in vivo, we used zebrafish embryos and A549 cells as hosts and a σVreI-dependent fluorescent construct in which the promoter of the σVreI regulated gene pdtA (Fig. 1) was cloned in front of a red fluorescent protein (rfp) gene (Table S1). P. aeruginosa PAO1 and ΔvreI strains were grown in high Pi conditions to avoid σVreI expression and activation prior to infection19. Bacteria were injected in the hindbrain ventricle of the zebrafish embryos (Fig. 5A) to produce an infection that remains initially localized, which facilitates fluorescence measurements. Directly after injection with either strain [0 hours post-infection (hpi)], red fluorescence was undetectable, indicating that there was no transcription from the σVreI-dependent pdtA promoter and thus that σVreI is not active (Fig. 5B, red channel). At this early infection point there was no neutrophil recruitment observed inside the hindbrain ventricle (Fig. 5B, green channel). However, at 12 hpi embryos injected with the PAO1 wild-type strain or the ΔvreI mutant showed recruitment of neutrophils in the head, indicative of an ongoing infection (Fig. 5B, green channel). Importantly, at this time point red fluorescence in bacterial aggregates was observed in the embryos infected with the wild-type but not with the ∆vreI mutant (Fig. 5B, red channel). This shows that pdtA expression occurs during infection in a σVreI-dependent manner and therefore suggests that this σ factor is expressed and active in these conditions.

Figure 5
figure 5

σVreI activation during P. aeruginosa infection in zebrafish embryos. (A) Dorsal view of the head of a two days post-fertilization zebrafish embryo. The hindbrain ventricle where P. aeruginosa was injected is highlighted. (B) Confocal images of the head of embryos injected in the hindbrain ventricle with 2000 CFU of the P. aeruginosa PAO1 wild-type strain or its isogenic ΔvreI mutant bearing the σVreI-dependent pdtA::rfp transcriptional fusion (pMP0690mCherry plasmid, Table S1) (red channel) at 0 and 12 hpi. P. aeruginosa was grown in high Pi medium prior injection. Neutrophils expressing constitutively a green fluorescence protein (gfp) are also visualized (green channel). Images are representative of three independent experiments (N = 3).

We also analyzed σVreI activation upon interaction of P. aeruginosa with the A549 human lung epithelial cell line. Green fluorescence-labelled P. aeruginosa PAO1 and ΔvreI strains bearing the σVreI-dependent pdtA::rfp transcriptional fusion were inoculated in A549 cultures. As control, bacteria were also inoculated in cell-free cultures. At 0 hpi, red fluorescence in either extracellular or internalized P. aeruginosa cells was undetectable (Fig. 6A). However, at 10 hpi red fluorescence was observed in the PAO1 wild-type cells and was considerably higher in the A549-containing cultures than in the A549-free cultures (P < 0.01) (Fig. 6A,B). In contrast, fluorescence remained undetectable in the ΔvreI mutant independently of the presence or absence of the A549 epithelial cells (Fig. 6A). Together this suggests that the expression of pdtA is induced by the presence of the human cells and that this induction depends on σVreI. To confirm this result, we measured pdtA expression by qRT-PCR after PAO1 inoculation in A549-free and A549-containing cultures. The cycle threshold (ct) value of a total of eight biologically independent co-incubations was determined (Fig. 7). In A549-containing cultures the expression of pdtA increased between 3- and 40-fold compared to the cell-free cultures (Fig. 7). This confirms that expression of pdtA is induced upon contact of P. aeruginosa with the eukaryotic cell. Altogether, these results indicate that σVreI is active during infection.

Figure 6
figure 6

Activation of σVreI upon interaction of P. aeruginosa with A549 eukaryotic cells. (A) Confocal images of P. aeruginosa-A549 co-cultures at 0 and 10 hpi. P. aeruginosa PAO1 wild-type strain and its isogenic ∆vreI mutant expressing a green fluorescent protein (gfp) constitutively (from the pBBRmEos3.1 plasmid, Table S1) (green channel) and containing the σVreI-dependent pdtA::rfp transcriptional fusion (pMP0690mCherry plasmid, Table S1) (red channel) were grown in high Pi medium and inoculated in A549-containing and A549-free cultures. A549 cells DNA was stained with DAPI (blue channel). Images are representative of three independent experiments (N = 3). (B) Quantification of the red fluorescence intensity observed in (A) was performed as described in Materials and Methods, and the total corrected cellular fluorescence (TCCF) is given. Data are means ± SD from three biological replicates (N = 3). P-value was calculated by unpaired two-tailed t-test.

Figure 7
figure 7

pdtA mRNA levels upon P. aeruginosa interaction with A549 cells. The P. aeruginosa PAO1 wild-type strain was grown in high Pi medium and inoculated in A549-containing and A549-free cultures. At 3 hpi, total RNA was extracted and pdtA mRNA levels determined by qRT-PCR. Data plotted are the result of eight biologically independent replicates, each bar representing means ± SD of the three technical replicates performed on each biological replicate. The cycle threshold (ct) average and the standard deviation (SD) of each condition is indicated. The 2−ΔΔCT method was used to determine the fold-change range in pdtA expression in A549-containing versus A549-free cultures. The fold-change range taking into account the SD is shown between brackets.

Discussion

Pi starvation has been described as an important signal for pathogens and indeed Pi amounts are partially linked to the immune status of the host. During major surgical operations and also in patients with severe burns, the reabsorption of Pi by the kidneys is reduced and exudative losses are higher than normal leading to hypophosphatemia, which can intensify upon treatments with bisphosphonates or antivirals32,33,34. During respiratory alkalosis caused by sepsis or mechanical respiration, a redistribution of the Pi into cells occurs and this reduces the extracellular content of Pi35. Increased levels of circulating catecholamines, which occurs in CF patients36, has been also associated with hypophosphatemia35. Evolution has benefited pathogens that are able to recognize and react to this marker37, thus exploiting the lower activity of the host immune system in this condition38. P. aeruginosa recognizes low Pi environments through the phosphate-specific ABC transport Pst system, which in Pi limiting conditions transports phosphate and triggers activation of the PhoR/PhoB two-component system37. Activated PhoR histidine kinase phosphorylates the transcriptional response regulator PhoB, which in turn promotes expression of the so-called pho regulon that in P. aeruginosa comprises several virulence factors37. In accordance, Pi starvation enhances P. aeruginosa virulence in zebrafish embryos (Fig. 2A), as also observed in mice and nematodes24,39,40,41, whereas excess of Pi reduces virulence (Fig. 2A). The pho regulon includes the vreAIR operon and the σVreI regulon (Fig. 1)19, and we have shown here that ΔvreI and ΔvreR mutations impair the low Pi-induced P. aeruginosa virulence (Figs. 2B and 3). Although the attenuated virulence of the ΔvreI mutant could be due to the lack of vreR expression in this mutant19, its phenotype is in line with previous results showing that overproduction of σVreI considerably increases P. aeruginosa pathogenicity21. In fact, it was surprising to find that the absence of the anti-σ factor VreR also led to attenuated virulence, despite the fact that σVreI is considerably active in the ΔvreR mutant19. Our transcriptomic and qRT-PCR analyses have revealed that VreR influences gene expression not only in a σVreI-dependent but also in a σVreI-independent manner. This includes several functions that could affect P. aeruginosa virulence (e.g. the PA0141/PPK2 required for alginate synthesis, the non-coding phrS RNA involved in quorum sensing regulation, the universal stress response dadA), and several regulators of gene expression (e.g. ddaR, PA3458/MarR-like regulator, PA4577/DksA-like regulator) (Table 1). How VreR modulates gene expression independently of σVreI is at present unknown but likely involves some of these regulatory proteins. In contrast, the third component of the operon, VreA, does not seem to play a role in the low Pi-induced virulence of P. aeruginosa. In fact, VreA does not seem to be involved in σVreI activation either in vitro19 or in vivo (Fig. 2B), which suggests that the σVreI/VreR signaling cascade does not require this protein.

Furthermore, we have shown in this work that σVreI is active in vivo during infection. pdtA expression, which in vitro completely depends on the σVreI/VreR signaling cascade and the PhoB regulator19, occurs during P. aeruginosa infections in zebrafish embryos and upon interaction with human epithelial cells, and in these in vivo situations expression of pdtA also completely depends on σVreI. The activation of σVreI in vivo is in accordance with the increased expression of the vreAIR operon and the σVreI regulon (including pdtA) in mouse models of acute and chronic infections42, and with the presence of antibodies against the PdtA protein in the serum of P. aeruginosa infected patients21. Presence of an active σVreI protein requires Pi limitation conditions in order for vreI to be expressed. However, the σVreI inhibitor, the VreR anti-σ factor, is also expressed in this condition19. A basal activity of σVreI is observed under Pi limitation; however, maximal activity of this σ factor requires the removal of VreR19. Because activation of P. aeruginosa σECF factors in response to the inducing signal usually occurs through the regulated proteolysis of the anti-σ factor12,13,14,15, another signal is expected to be required for complete σVreI activation. The signal could be host-derived, as described for the σV factor of the opportunistic pathogens Clostridioides difficile and Enterococcus faecalis, which is activated by lysozyme, an important component of the innate immune system of many organisms43. Alternatively, the activating signal could be produced by P. aeruginosa itself in response to the host environment. An example of such a mechanism is the activation of σPvdS in P. aeruginosa, which in response to the iron starvation conditions encountered in the host44,45 produces pyoverdine that in turn increases σPvdS activation and P. aeruginosa virulence46,47. Another example is the activation of the P. aeruginosa σAlgT factor in response to the oxidative stress generated by the oxygen radicals produced by leucocytes or in response to the elevated temperatures that are often produced in infected hosts48,49. Both situations lead to cell envelope stress and the accumulation of misfolded proteins in the bacterial periplasm, which trigger σAlgT activation50. Further research will be conducted to clarify how activation of σVreI in response to the host occurs.

Methods

Bacterial strains and growth conditions

Strains used are listed in Table S1. Bacteria were routinely grown in Luria-Bertani (LB) medium51 at 37 °C in a rotary shaker at 200 rpm. For differential expression of the vreAIR operon low and high phosphate media were used19. When necessary, antibiotics (Sigma-Aldrich) were used at the following final concentrations (μg ml−1): ampicillin (Ap), 100; gentamicin (Gm), 20; hygromycin B (Hg), 100; nalidixic (Nal), 10; and tetracycline (Tc), 20.

Plasmids construction and molecular biology

Plasmids used are described in Table S1 and primers listed in Table S2 and S3. PCR amplifications were performed using Phusion® Hot Start High-Fidelity DNA Polymerase (Finnzymes). All constructs were confirmed by DNA sequencing and transferred to P. aeruginosa by electroporation52.

Zebrafish maintenance, embryo care and infection procedure

Transparent adult casper mutant zebrafish (mitfaw2/w2; roya9/a9) and adult labelled Tg(mpx:GFP)i114 casper zebrafish producing green fluorescent neutrophils53,54 were maintained at 26 °C in aerated 5 L tanks with a 10/14 h dark/light cycle. Zebrafish embryos were collected during the first hour post-fertilization (hpf) and kept at 28 °C in E3 medium (5.0 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl·2H2O, 0.33 mM MgCl2·6H2O) supplemented with 0.3 mg/L methylene blue. Prior to infection, 1 or 2 days post-fertilization (dpf) embryos were mechanically dechorionated and anaesthetized in 0.02% (w/v) buffered 3-aminobenzoic acid methyl ester (pH 7.0) (Tricaine, Sigma-Aldrich). Zebrafish embryos were individually infected by microinjection with 1 nl of P. aeruginosa either in the hindbrain ventricle (localized infection) or in the caudal vein (systemic infection) as described elsewhere26,55. All procedures involving zebrafish embryos were according to local animal welfare regulations and in accordance with the EU Animal Protection Directive 86/609 EEC.

Virulence assay in infected zebrafish embryos

Zebrafish embryos were injected in the caudal vein with 1000 colony forming units (CFU) of exponentially grown P. aeruginosa cells in low or high phosphate conditions previously resuspended in phosphate-free physiological salt containing 0.5% (w/v) of phenol red. After infection, embryos were kept in 6-well plates containing 60 μg/mL of Sea salts (Sigma-Aldrich) at 32 °C with 20 individually injected embryos in each group per well. Embryo survival was determined by monitoring live and dead embryos at fixed time points during five days. Four biologically independent experiments were performed and the data given are the average. P-values were calculated by log-rank (Mantel-Cox) test.

Confocal fluorescence imaging of zebrafish embryos

For confocal imaging, zebrafish embryos were injected in the hindbrain ventricle with 2000 CFU of P. aeruginosa cells containing a σVreI-dependent red fluorescence transcriptional fusion. At 0 and 12 hpi, embryos were fixated overnight in 4% (v/v) paraformaldehyde in phosphate buffered saline (PBS). Before imaging, fixated embryos were embedded in 1.5% (w/v) low-melting-point agarose using an open uncoated 8-well microscopy µ-slide (Ibidi®). Confocal images were generated with a Leica TCS SP8 Confocal Microscope. Leica Application Suite X and ImageJ software was used to process the confocal images, specifically for brightness/contrast enhancements as well as for creating merged images.

Cytotoxicity assay in A549 human lung epithelial cells

P. aeruginosa cytotoxicity on A549 cells was assayed using a colorimetric assay that detects the number of metabolically active eukaryotic cells able to cleave the MTT tetrazolium salt (Sigma-Aldrich) to the insoluble formazan dye. The A549 cell line (ATCC® CCL-185™) was maintained in Dulbecco’s Modified Eagle Medium (DMEM) medium supplemented with 10% (v/v) fetal bovine serum (FBS) (Gibco) in a 5% CO2 incubator at 37 °C. One day prior to infection, the A549 cells were placed in 96-well plates at a concentration of 4 ×104 cells/well and cultured in phosphate-free DMEM medium (Gibco) with 5% (v/v) FBS. In this condition, cell mitosis does almost not occur. Late exponentially grown P. aeruginosa strains in low or high phosphate conditions were then inoculated at a multiplicity of infection (MOI) of 20. At 3 hpi, 30 μl of a 5 mg/ml MTT solution in PBS was added to the wells and the plates were incubated for 2 h. The culture medium was then removed and 100 μl of dimethyl sulfoxide (DMSO) was added to solubilize the formazan. Production of formazan, which directly correlates to the number of viable cells, was quantified using a scanning multi-well spectrophotometer (Infinite® 200 PRO Tecan) at 620 nm.

Confocal fluorescence imaging of A549 cells

For confocal imaging, 2 ×105 A549 cells were seeded in 24-well plates containing 11 mm round glass coverslips and phosphate-free DMEM medium with 5% (v/v) FBS one day prior to infection. Infections with green fluorescence-labelled P. aeruginosa cells containing a σVreI-dependent red fluorescence transcriptional fusion were performed at a MOI of 10. At 0 and 10 hpi, the cells were fixated with 4% (v/v) paraformaldehyde in PBS. Samples were washed with PBS and coverslips were mounted on glass slides containing Fluoroshield mounting medium with DAPI (Sigma-Aldrich) to retain fluorescence and stain the A549 cells DNA. Confocal images were generated with a Nikon A1R confocal scanning laser microscope. NIS-Elements and ImageJ software were used to process the confocal images. Total corrected cellular fluorescence (TCCF) was calculated using the following equation: TCCF = integrated density − (area of selected cell × mean fluorescence of background readings), as described before56,57.

Internalization assay

To enumerate bacteria internalized into A549 cells, a polymyxin B protection assay was performed with P. aeruginosa strains grown at late exponential phase in low phosphate conditions. A549 cells were cultured in 24-well plates at a concentration of 2 ×105 cells/well in phosphate-free DMEM medium supplemented with 5% (v/v) FBS. Bacterial infections were performed at a MOI of 10. At 6 hpi, the culture supernatants were collected and serial dilutions plated on LB for bacterial counting. Fresh DMEM medium containing 20 μg/ml polymyxin B was added to the infection wells and incubated during 45 min to kill extracellular bacteria. Afterward, the antibiotic-containing medium was removed and the cells were lysed with 1% (v/v) Triton X-100 (Sigma-Aldrich) in PBS during 10 min. Serial dilution of the lysed cells were plated on LB for bacterial counting. Internalization is reported as the ratio between the CFU after the lysis of the A549 cells and CFU in the culture supernatant.

RNA isolation

Total RNA was extracted by the hot phenol method using the TRI® Reagent protocol (Ambion) as described elsewhere58 and subjected to two DNase I treatments with the Turbo DNA-free kit (Ambion) and RNaseOUT (Invitrogen). RNA quality, including purity, integrity and yield, was assessed by electrophoresis of 1 µg of total RNA and by UV absorption at 260 nm in a ND-1000 spectrophotometer (NanoDrop Technologies, USA).

Quantitative RT-PCR analyses

A549 cells cultured in 6-well plates in phosphate-free DMEM medium with 5% (v/v) FBS were infected with exponentially grown P. aeruginosa cells in high phosphate conditions at a MOI of 10. At 3 hpi, the culture supernatants were spun down at 1600 × g and 4 °C and total RNA was isolated. cDNA from eight biologically independent replicates was produced in triplicate by reverse transcription reactions of 0.5–1 µg RNA using SuperScript II reverse transcriptase (Invitrogen) and random hexamers as primers according to the protocol supplied. Real-time PCR amplifications were carried out on a MyiQ2 system (Bio-Rad) associated with iQ5 optical system software (version 2.1.97.1001) in 11.5 µl reaction mixture containing 6.25 µl of iQ SYBR green Supermix (Bio-Rad), 0.4 µM of each primer (Table S2) and 1 µl of the template cDNA (diluted 1000-fold when measuring the 16S rRNA reference gene). Thermal cycling conditions were the following: one cycle at 95 °C for 10 min and then 40 cycles at 95 °C for 15 s, 65 °C for 30 s, and 72 °C for 20 s, with a single fluorescence measurement per cycle according to the manufacturers’ recommendations. Melting curve analyses were performed by gradually heating the PCR mixture from 55 to 95 °C at a rate of 0.5 °C per 10 s for 80 cycles. The relative expression of the genes was normalized to that of 16 S rRNA and the results were analyzed by means of the comparative cycle threshold (∆∆ct) method59.

RNA-seq analysis

P. aeruginosa PAO1 and its isogenic ΔvreR mutant were grown until late exponential phase in low Pi medium and total RNA was isolated. A mixture of the three isolations was used for RNA-seq analyses, which were performed at Era7 Bioinformatics (Granada, Spain). First, the rRNA from 4 µg of total RNA was removed using the Ribo-Zero rRNA removal kit (Illumina) following the manufactures’ recommendations. Sequencing libraries were prepared with the NEBNext Ultra directional RNA library Prep kit (New England Biolabs). RNA was fragmented at 94 °C for 7.5 min and first strand cDNA synthesis was performed at 42 °C during 50 min using the adaptor and primers recommended by the manufacturer (NEBNext Multiplex Oligos, Illumina). Samples were then processed on the Illumina NextSeq. 500 Sequencer in one run with a read length of 2 × 75 bp. The pipeline recommended in the tool Cufflinks 2.2.160 was used to analyze the data and obtain the set of differentially expressed genes following these steps: 1) Raw reads quality analysis with the tool FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc); 2) Reads mapping to the P. aeruginosa PAO1 reference genome (NCBI reference sequence NC_002516.2) with the tool Bowtie integrated in the Tophat suite61; and 3) analysis of differences in gene expression with Cufflinks 2.2.160. All the samples passed the quality analysis. Tests related to overrepresented sequences failed in some of the samples (corresponding to overexpressed genes in this case, e.g. pdtA, Table 1). Significant differentially expressed genes depending on whether P is greater than the false discovery rate (FDR) after Benjamini-Hochberg correction for multiple-testing (as indicated in the Cufflinks 2.2.1 tool) are shown in Table 1. P-value is below 0.0001 in all cases. RNA-seq data set have been deposited in NCBI’s Gene Expression Omnibus (GEO) database under accession number GSE122253 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE122253). RNA-seq results were confirmed by qRT-PCR using RNA from three biological replicates obtained from P. aeruginosa cells grown in the same conditions that for the RNA-seq assay and primers listed in Table S2.

Other bioinformatics analyses

Statistical analyses are based on t-test in which two conditions are compared independently. P-values from raw data were calculated by independent two-tailed t-test and from ratio data to the control by one-sample t-test using GRAPHPAD PRISM version 5.01 for Windows and are represented in the graphs by ns, non-significant; *P < 0.05; **P < 0.01; and ***P < 0.001.