The bacterial strains and plasmids used in this study are listed in Table 1, and the primers are listed in Table 2. Escherichia coli and P. aeruginosa strains were grown in Luria-Bertani (LB) medium (for broth culture) or 1.5% (wt/vol) agar LB plates at 37°C. Antibiotics were used at the following concentrations (μg/mL): ampicillin (Amp; 100) and gentamicin (Gm; 20) for E. coli and P. aeruginosa.

For both microarray analysis and RT-qPCR, total RNA was isolated from lag phase (2 h), logarithmic phase (7 h) or stationary phase (15 h) cultures of PA3/ctrl and PA3/orf70.1 by SV Total RNA Isolation System (Promega, USA). In RT-qPCR analysis, the synthesis of cDNA was prepared with a PrimeScript RT reagent kit (TaKaRa Bio; Dalian, China) according to the manufacturer’s recommendations. Quantitative real-time PCR was performed using SYBR Premix Ex Taq II (TaKaRa Bio). Primers used in this study are listed in Table 2. 16S rRNA was selected as the reference gene for normalization. The raw data of the microarray experiments was deposited in the GEO database¹ with accession number GSE77297.

A transmission electron microscope was employed for the bacterial morphology with 2% p – tungstic acid. Laser Scanning Microscope (Zeiss, model LSM 510; Jena, Thuringia, Germany) was used to observe the proliferation of bacteria. The twitching/swimming-motility assay was performed by stabbing a colony of bacteria into the bottom of a petri plate containing 10 ml of agar medium (1% agar content for twitching and 0.3% for swimming). Following incubation at 37°C for 17 h, the zone of motility was measured (Alm and Mattick, 1995). Sensitivity to hydrogen peroxide (H₂O₂) was measured as previously described (Hassett et al., 1995). For sensitivity to PaP3, bacterial cultures in mid-log (OD600 = 0.3) were placed on LB plates, and 1 μl PaP3 (10⁹ pfu/ml) was dropped on the bacteria-seeded plates and incubated at 37°C for 18h, and the zones of plaque were measured. For antibiotic sensitivity tests, P. aeruginosa PA3/ctrl or PA3/orf70.1 were tested against 11 antimicrobial agents: piperacillin, netilmicin, ceftazidime, aztreonam, meropenem, polymyxin, amikacin, ciprofloxacin, tobramycin, cefepime, and imipenem. Antimicrobial sensitivity was monitored by Kirby-Bauer disk diffusion assay with commercially available disks (Bio-Rad) (Bauer et al., 1966).

To identify proteins from P. aeruginosa that could interact with gp70.1, we used the bacterial two-hybrid system (Dove et al., 1997). Genomic DNA was isolated from P. aeruginosa strain PA3 and partially digested with Sau3AI. Sau3AI fragments (∼ 500–1,500 bp) were gel purified and ligated into BamHI-digested pRAC. This plasmid library was transformed into E. coli strain KS1 containing plasmid pRBR-orf70.1 or pRBR (negative control). As a positive control, we used plasmids pBRL28 and pACλCI-β-flap (Rao et al., 2009). Co-transformants were then placed on LB plates supplemented with 100 μg/mL of Amp, 50 μg/mL of kanamycin, 34 μg/mL of chloromycetin (Cm), 0.1 mM of IPTG, and 50 μg/mL of X-gal. Plates were incubated at 30°C for 24 h, and then β-galactosidase assays were performed as previously described (Thibodeau et al., 2004).

His-tagged gp70.1, GST-tagged FlgM, GST-tagged RpoS and GST alone were purified with Ni-agarose or glutathione-agarose. GST pull-down assays were performed by Pierce GST Protein Interaction Pull-Down kit (Thermo Scientific, USA) according to the manufacturer’s instructions. Samples were separated by SDS-PAGE, stained with Coomassie blue, and the captured His-gp70.1 was detected by a monoclonal antibody against the His tag.

Bacteria were cultured for 7 h and 2 mL media was placed in the ice for 20 min, followed by centrifugation at 15, 000 × g for 10 min at 4°C. After filtering the bacteria onto the membrane (0.22 μm), 1 mL of supernatant per sample was stored at -80°C until NMR analysis. For NMR analysis, 450 μl supernatant per sample was added with 50 μl Anachro Certified DSS Standard Solution (ACDSS, 4.088 mM) and centrifuged at 13, 000 × g at 4°C for 2 min before transferred into the NMR tubes. All the NMR experiments were performed on an Agilent DD2 600 MHz spectrometer (Agilent Technologies, Santa Clara, CA, USA) equipped with a triple-resonance cryogenic probe at 298.15 K. Metabolites were identified and quantified using Chenomx NMR Suit (version 8.0, Chenomx, Edmonton, AB, Canada). Prior to Fourier transformation, phasing and baseline correction, a line broadening of 0.5 Hz was applied to the free induction decay. The full width at half maximum (FWHM) of the DSS-d6 peak at 0.00 ppm was assessed for each spectrum. A total of 49 metabolites were obtained from the Chenomx database containing more than 250 compounds. The concentrations of the metabolites in each spectrum were normalized by the absolute spectral intensity.

The animal experiment was approved by the Animal Research Ethics Committee of Third Military Medical University, China. P. aeruginosa were grown in LB medium at 37°C until the early stationary phase and were collected by centrifugation at 10,000 × g for 1 min. The cell pellet was washed twice with saline buffer and resuspended in the same buffer at a final concentration of 3 × 10⁷ cfu/ml. Each of two groups of 10 mice (6–8-week-old BALB/c female mice) was injected intraperitoneally with 1ml of PA3/orf70.1 or PA3/ctrl, followed by 24 h of observation. Three independent experiments were performed.

For microarray dataset, PseudoCAP (Pseudomonas aeruginosa Genome Database) function analysis was used to assess the functions of the DEGs (differentially expressed genes) (Winsor et al., 2005). Package “OmicCircos” in R was used for the generation of circular plots (Hu et al., 2014). The software STEM (Short Time-series Expression Miner) was applied for the comparison and visualization of the short time series (2, 7, and 15 h) gene expression datasets between PA3/orf70.1 and PA3/ctrl (Ernst and Bar-Joseph, 2006). STEM analysis uses a correlation-based clustering algorithm. Briefly, 49 distinct and representative model profiles of temporal expression were selected independent of the data. Each gene of microarray dataset was assigned to the model profile that most closely matched the gene’s expression profile by the correlation coefficient (r). The algorithm then determined which profiles have a statistically significant higher number of genes assigned using a permutation test (p).

In NMR metabolomics, software MestReNova 9.0 (MestreLab, Santiago de Compostela, Spain) was used to obtain peak areas from the raw spectrum. Principal Components Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA) were performed by PCA Methods Bioconductor and pls package in R. For each metabolite, the Variable Importance in the Projection (VIP) score was calculated to determine the most relevant metabolites in phage infection (Chong and Jun, 2005). The R’s ggplot2 was used for data visualization.

Aimed to screen inhibitors from early gene products of PaP3, early genes were cloned into expression vector pUCP24. P. aeruginosa PA3 were transformed with early gene expression vectors (PA3/orf) or the empty vector (PA3/ctrl) and the changes of colony phenotype were first observed. Then, a dwarf colony was observed in PA3 containing orf70.1 (PA3/orf70.1), which showed a needle-like colony and a little secretion around the colony compared to the control (Figure 1A). To evaluate the influence of the gene products (gp70.1) of orf70.1 on PA3, the growth efficiency of bacteria was further examined by growth curves based on optical cell densities at 600 nm (OD600) measurement (Figure 1C) and CFU (colony-forming unit) calculation (Figure 1D). Gp70.1 had a severe inhibitory effect on the concentration of bacterial culture. Intriguingly, the growth curve based on CFU showed that the growth of PA3 was delayed by gp70.1 and the number of living bacteria was similar with the control in the stationary phase. At the same time, macroscopic cell aggregates of PA3/orf70.1 appeared in the culture, which were also observed in Gram staining (Figures 1E,F). This might be a reason for the low OD values caused in the absence of gp70.1. Moreover, the growth-inhibitory effect of gp70.1 was non-specific so it also worked in E. coli (Figure 1B).

The gene orf70.1 was first recognized as a probable gene by GeneMark and showed a high expression level at the early stage of phage infection in the RNA-seq analysis (data to be published separately). This gene was located on the positive strand, beginning at position 44,741 and ending at position 45,133 (393 bp). According to the one-step growth curve of phage PaP3, RT-qPCR analysis was used to examine the expression of orf70.1 at different infection periods. The results also showed a high expression level of orf70.1 at 5 min and reached a peak value at 10 min (Figure 2), which confirmed that the orf70.1 was an early gene.

The gene product of orf70.1 (gp70.1) was predicted to consist of 131 amino acids. A similarity search was run by BlastP against non-redundant protein sequences and no putative conserved domains were detected in gp70.1. In the BlastP analysis, 8 similar sequences (hits) were found. The top two hits were metallophosphoesterase (e-value = 0.57, ident = 38%) and anti-anti-sigma factor (e-value = 1.4, ident = 38%). The purified gp70.1 was obtained from BL21 (DE3)/pET22b cells. The molecular weight of gp70.1 was approximately 13 kd and consistent with the prediction according to SDS-PAGE. The record of the Pseudomonas phage PaP3 RefSeq Genome (GenBank: AY078382.2) was updated based on this study.

The early genes of phage usually encoded regulatory proteins to control the expression of host genes. Based on the growth curve of PA3/orf70.1 and PA3/ctrl, three time points containing lag (2 h), logarithm (7 h), and stationary phase (15 h) were chosen to compare the whole gene expression profile of PA3 with or without gp70.1 at each time point. Compared to the control (PA3/ctrl), a total of 178 genes of PA3 were differentially expressed (ratio > 2, p < 0.05) in the presence of gp70.1, of which, 40 DEGs(differentially expressed genes) were detected at 2 h, 79 DEGs at 7 h and 75 DEGs at 15 h. As shown in Figure 3A, 14 PseudoCAP (Pseudomonas aeruginosa Genome Database) function terms were attributed to the 178 DEGs, which could be divided into two main categories: extracellular functions and metabolism functions (Winsor et al., 2005). About one third of the DEGs were involved in the extracellular function including cell wall/LPS/capsule, motility and attachment, membrane protein, transport of small molecules and protein secretion/export apparatus. In addition, subcellular localization of the DEGs in Figure 3B suggested that “periplasmic” and “extracellular” locations contributed larger proportions of DEGs. The metabolism functions were related to energy, amino acid and carbon compound metabolism. Moreover, a large number of genes related to translation were up-regulated at 7 h and 54% of them were genes encoding ribosomal proteins.

Since the growth of PA3 were delayed in the presence of gp70.1, the comparison at each time point did not eliminate the influence caused by the different growth state between PA3/ctrl and PA3/orf70.1. In order to stringently examine the effects of gp70.1 on the gene expression of PA3, the Short Time-series Expression Miner (STEM) was used to complementarily analyze the microarray data (Ernst et al., 2005). In STEM analysis, 49 temporal expression profiles were selected as model profiles, which were ordered from 1 to 49 (Figure 4A, top left). Each gene of PA3 was assigned to different model profiles based on the time series expression values (log ratios) from different experimental conditions (PA3/ctrl or PA3/orf70.1). A total of 2,987 genes from PA3/ctrl were assigned to profile 49, while these genes exhibited three patterns of expression: profile 49 (1,534 genes, p = 0), 48 (1,443 genes, p = 0) and 46 (10 genes, p = 2e - 9) when gp70.1 was overexpressed in PA3. The minimum of the correlation (r) of 2 profiles was 0.7 in STEM analysis, so that the expression patterns of 10 genes in profile 46 (r = 0.62) were significantly changed by gp70.1. Compared to the model profile 49, the significant difference of these genes in profile 46 was mainly reflected in the reduction at 7 h (Figure 4B). PseudoCAP functions and subcellular localizations of the 10 genes in Figure 4C showed that three genes (PA0041, lipH, and lldP) were related to the extracellular function and four genes (PA0041, lipH, lldP, and ftsW) were located in extracellular locations (Winsor et al., 2005). Notably, RNA polymerase sigma factor RpoN that positively regulating bacterial nitrogen metabolism and motility also appeared to have a changed expression pattern in the presence of gp70.1 (Totten et al., 1990). Thus it can be seen, orf70.1 might encode a regulatory product, which caused differentially expressed genes mainly related to extracellular function and metabolism of P. aeruginosa.

Although gp70.1 caused inhibited growth of PA3, there were unexpectedly more activated genes than inhibited genes at each time point in our microarray analysis. Seven DEGs selected from microarray data were validated by real-time qPCR (Figure 5), as being related to bacterial motility (pilI and fliC), secretion and transport (ffh and cypS), anima acid metabolism (arcC and sucA) and ribosome protein (rplR). The expression levels of orf70.1 in P. aeruginosa were also examined. The results showed that orf70.1 was stably expressed at each growth stage with the Ct values ranging from 16 to 17 (Figure 5A). The RT-qPCR results showed a consistent directional change compared to microarray assay. All the 7 detected DEGs were shown with varying degrees of upregulation at different time points. The expression of the twitching motility protein PilI gene (pilI) and flagellin type B gene (fliC) were reduced at 2 h and both up-regulated at 15 h (Figure 5B). Meanwhile, ffh and cysP, related to protein export and sulfate transport were activated at the lag phase. Remarkably, the RT-qPCR results confirmed the differential expression of the genes involved in amino acid metabolism, such as arcC (carbamate kinase) and sucA (2-oxoglutarate dehydrogenase), which were related to arginine deiminase pathway and lysine degradation, respectively. The serious inhibition of arcC at the logarithmic phase might hinder arginine utilization of bacteria, while the over-expression of sucA would increase lysine consumption. In addition, the significant upregulation of rplR at 7 h was consistent with the microarray results.

Based on the initial observation that the OD values of the bacterial culture were reduced and the differentially expressed genes were related to bacterial metabolism in the presence of gp70.1, the metabolites in bacterial supernatant at the logarithmic phase were detected by NMR spectra. There were a total of eight samples – four biological replicates for PA3 with overexpressed gp70.1 (PA3/ctrl) and 4 biological replicates for the control group (PA3/orf70.1). The eight samples in principle component analysis (PCA) diagram were distinctly separated into two groups, indicating the difference of supernatant components between PA3/ctrl and PA3/orf70.1 (Figure 6A). Finally, 15 significantly differential metabolites, including 12 amino acids, two organic acids, and one sugar, were identified by Variable Importance in Projection (VIP) analysis (Figure 6B). The contribution of metabolites in distinguishing the sample classification was measured by VIP score. The top five metabolites (alanine, arginine, pyroglutamate, glutamate, and ornithine) with VIP scores of more than two all belonged to amino acids. Most (9/12) of the differential amino acids were shown with higher concentrations in PA3/orf70.1 than the control. It meant that the main effects of gp70.1 on the metabolism of PA3 were the reduction of amino acid consumption. Figure 6C presents three significantly differential metabolites: alanine and pyroglutamate with the higher concentrations and ornithine with a lower concentration. Furthermore, the NMR results confirmed RT-qPCR analysis: the down-regulated arcC did reduce arginine uptake, while the activation of sucA increased lysine consumption.

To further demonstrate the effects of gp70.1, a series of phenotype experiments were performed on the basis of the above results. First, the transmission electron microscope image showed no difference of morphology between PA3/ctrl and PA3/orf70.1 (Figure 7A). However, it was obvious that PA3/orf70.1 showed fewer secretions around the cell compared to PA3/ctrl. In addition, we further found that gp70.1 severely inhibited the measured extracellular metabolites of P. aeruginosa including protease, polysaccharide, extracellular cellulase, and pyocyanin (Figures 7B–D). The genes (pilI and fliC) related to motility were down-regulated in the microarray assay and RT-qPCR analysis. Thus, bacterial motility was examined on LB agar plates (1% agar for twitching and 0.3% agar for swimming). The results showed a strong inhibitory effect of gp70.1 on the motility of P. aeruginosa (Figures 7E,F). At this point, it was clear that gp70.1 exerted broad inhibitory effects on extracellular functions and metabolism, which caused multiple repressed phenotypes of P. aeruginosa.

We hypothesized that gp70.1 exerted the inhibitory effects on P. aeruginosa by the interactions with its cellular targets. To track this, a bacterial two-hybrid (B2H) assay was used to screen the targets of gp70.1 in the PA3 genomic library. Two potential target proteins, FlgM and RpoS, of gp70.1 were finally identified (Figure 8A). To validate the B2H results, pull-down experiments were performed using His-tagged gp70.1, GST-tagged FlgM and GST-tagged RpoS. The results indicated that the RpoS could interact with gp70.1 directly in vitro, while no interaction was detected between gp70.1 and FlgM (Figure 8B). As a global regulator, RpoS is involved in the regulation of broad functions in P. aeruginosa including: biofilm, motility, virulence, antibiotic resistance and, most primarily, the stress response. We then tested these RpoS-regulated functions of PA3/ctrl and PA3/orf70.1. Antibiotic sensitivity tests were performed with 11 antibiotics (Figure 8C), the results indicated that gp70.1 increased the sensitivity of PA3 to 6 antibiotics: ceftazidime, aztreonam, polymyxin B, amikacin, tobramycin, and cefepime, while it decreased the sensitivity to netilmicin. In the phage and H₂O₂ sensitivity experiments, PA3 containing gp70.1 showed a higher sensitivity than the control group (Figures 8D,E). Furthermore, biofilm formation of PA3 was significantly inhibited by gp70.1 (Figure 8F). Lastly, and most importantly, gp70.1 greatly reduced the virulence of PA3 so that the survival rate of mice vaccinated with PA3/orf70.1 was up to 80% within 24 h of observations, but 0% in the case of PA3/ctrl (Figure 8G). These results suggested that the interactions of gp70.1 and RpoS had an inhibitory effect on the functions regulated by RpoS.