Phenazine-1-carboxylic Acid Produced by Pseudomonas chlororaphis YL-1 Is Effective against Acidovorax citrulli

The bacterial pathogen Acidovorax citrulli causes the destructive fruit blotch (BFB) on cucurbit plants. Pseudomonas chlororaphis YL-1 is a bacterial strain isolated from Mississippi soil and its genome harbors some antimicrobial-related gene clusters, such as phenazine, pyrrolnitrin, and pyoverdine. Here, we evaluated the antimicrobial activity of strain YL-1 as compared with its deficient mutants of antimicrobial-related genes, which were obtained using a sacB-based site-specific mutagenesis strategy. We found that only phenazine-deficient mutants ΔphzE and ΔphzF almost lost the inhibitory effects against A. citrulli in LB plates compared with the wild-type strain YL-1, and that the main antibacterial compound produced by strain YL-1 in LB medium was phenazine-1-carboxylic acid (PCA) based on the liquid chromatography-mass spectrometry (LC-MS) analysis. Gene expression analyses revealed that PCA enhanced the accumulation of reactive oxygen species (ROS) and increased the activity of catalase (CAT) in A. citrulli. The inhibition effect of PCA against A. citrulli was lowered by adding exogenous CAT. PCA significantly upregulated the transcript level of katB from 6 to 10 h, which encodes CAT that helps to protect the bacteria against oxidative stress. Collectively, the findings of this research suggest PCA is one of the key antimicrobial metabolites of bacterial strain YL-1, a promising biocontrol agent for disease management of BFB of cucurbit plants.


Introduction
Bacterial fruit blotch (BFB), caused by Acidovorax citrulli, is a destructive disease that infects cucurbit plants [1]. The disease was first found in Florida in the USA in 1989 [2]. In recent years, the disease has been involuntarily spread to many parts of the world by contaminated commercial seeds [1]. It was reported that seeds even with a low level of A. citrulli containment can result in severe BFB epidemics under a favorable environment [3]. Besides fruit blotch, A. citrulli also causes leaf blight, seedling blight, and/or blossom rot of cucurbitaceous plants. Annual yield loss of watermelon caused by BFB in the USA reached 5-50%, depending on the stage of infection and environmental conditions [4,5].
Since BFB is a seed-borne disease [6], control of BFB mainly depends on use of A. citrulli-free seeds and on treatment of A. citrulli-contaminated seeds either with physical measures such as dry heat or with antibacterial chemicals such as thiamphenicol [7,8]. However, the use of preventive control measures as described above has led to some negative impacts on seed quality, including reduction of seed germination and seedling growth [9]. Therefore, alternative safe measures for control of BFB are needed. It was reported that the decontamination method to inactivate A. citrulli on Cucurbitaceae seeds Sangon (Shanghai, China); unless otherwise stated, the solvents used for extraction were of analytical grade and those for LC-MS of HPLC grade.
All primers used for PCR and cloning are listed in Table S1. The B. subtilis sacB gene, which encodes levansucrase that is toxic to Gram-negative bacteria in the presence of sucrose, was considered as a counter-selection marker in the system [38]. We selected phzE as a representative example to generate an in-frame deletion mutant ( Figure S1). First, the upstream and downstream homolog fragment of phzE in wild-type strain YL-1 was cloned by using two pairs of primers (phzE-F1 and phzE-R1, phzE-F2 and phzE-R2). Then, a sacB-containing suicide vector, pEX18Gm [39] was used to create a recombinant plasmid, pEX18-phzE. The recombinant plasmid was first transformed into E. coli S17-1 [40]; then, it was transformed into wild-type strain YL-1 via an optimized bacterial conjugal approach. Via a double-crossover homologous recombinant approach, an in-frame deletion mutant of phzE, named as ∆phzE, was generated and validated [41]. The mutants ∆phzF, ∆prnB, ∆hcnB, ∆acsA, and ∆acsD were generated as described above.
Gene replacements and insertions were verified by PCR amplifications using specific primers and DNA sequencing. For example, using genomic DNA samples of the mutant and wild-type strain as PCR templates, respectively, specific primers (Ex. phzE-F1and phzE-R2, Table S1) were used to amplify the target region. The resulting PCR product with different sizes was observed using agarose gel electrophoresis. Then, the two PCR products were sequenced, and sequence analysis was conducted to verify the authenticity of mutants.

Growth, Hydrogen Cyanide and Pyoverdine Measurements
P. chlororaphis YL-1 and its seven mutants (Table S1) were individually incubated in a 250 mL flask containing 100 mL liquid LB medium or low-iron succinic acid medium (SM) [42] at 28 • C with shaking (200 rpm for 24 h). All liquid cultures were observed under visible light and ultraviolet light simultaneously. Colony-forming units per mL (CFU/mL) were determined by the standard serial dilution. The experiment was repeated three times independently.
Production of hydrogen cyanide was observed according to the method of Lorck [43]. Freshly grown cells were spread on King's B medium containing glycine (4.5 g/L). A sterilized filter paper saturated with 1% solution of picric acid and 2% sodium carbonate was placed in the upper lid of a petri dish. The petri dish was then sealed with parafilm and incubated at 30 • C for 4 days. A change in color of the filter paper from yellow to reddish brown as an index of cyanogenic activity was recorded.

Antibacterial Activity of Strain YL-1 and Mutants
Strain YL-1 and three phytopathogenic bacteria were incubated in LB media at 28 • C with shaking (200 rpm for 24 h). Thereafter, the cultures were resuspended in LB broth to the desired cell density of 10 8 CFU/mL. Cells of three phytopathogenic bacteria, A. citrulli strain XJX12, Xanthomonas oryzae pv. oryzae (Xoo) strain PXO99, and X. oryzae pv. oryzicola (Xooc) strain RS11, were evenly sprayed onto the surface of LB plates for 1 s using sterile laryngeal spray (Taizhou Medical Instrument Co. Ltd., Taizhou, China). Then, 5 µL cells of strain YL-1 (10 8 CFU/mL) were added on the LB plates, 1 cm from the edge of the plate in a cross shape. Distilled LB broth was used as a control. Each treatment was replicated three times and each experiment was repeated thrice independently. The inhibitory zone diameters of strain YL-1 were measured when phytopathogenic bacteria covered the control plates entirely.
The differences in antibacterial activities between YL-1 and its seven mutants were evaluated as described above [25]. After phytopathogenic bacteria (A. citrulli or Xoo) were evenly sprayed onto the surface of LB plates, 5 µL cells of YL-1 or mutants (10 8 CFU/mL) were injected into the central hole (5 mm) of LB plates and incubated at 28 • C for 48 h. Distilled LB broth was used as a control. The inhibitory zone diameters of strain YL-1 and mutants were measured when A. citrulli or Xoo covered the control plates entirely.

Extraction and Determination of PCA
To extract the maximum PCA, strain YL-1 and mutants ∆phzE and ∆phzF were cultured in 100 mL liquid King's B medium at 28 • C for 48 h, then culture supernatants were mixed with 15 mL of 6 M HCl and extracted with 100 mL chloroform as previously described [44]. The soluble organic fraction was concentrated by a rotary evaporator at a temperature below 35 • C, 70 mg of crude compound was obtained and dissolved in 7 mL acetone, and then the concentration was 10 mg/mL. A three-microliter aliquot of extracted PCA from the fermentation broth of strain YL-1 and STD sample was then taken for HPLC analysis (Agilent Technologies 1260 Infinity) under the following conditions: XDB-C18 reversed-phase column (4.6 mm × 150 mm, 5 µm, Agilent) eluted with acetonitrile (ACN) −5 mM ammonium acetate (60:40, v/v) at a flow rate of 0.7 mL/min. Thereafter, the fraction from retention time 1.95 to 2.10 min was first collected and concentrated by a rotary evaporator, then dissolved in acetone and determined with ultraperformance liquid chromatography coupled with mass spectrometry (Agilent UPLC 1290-MS 6230, LC/MS) under the following conditions: Zorbax XDB C18 reverse-phase (4.6 mm × 150 mm, 5 µm, Agilent) eluted with gradient H 2 O with 0.5% acetic acid (A) and ACN with 0.5% acetic acid (B). Mobile phase flow rate was set at 0.4 mL/min using a linear gradient of A and B starting at 0% B, increasing to 50% B in 5 min, increasing to 100% B in 8 min, maintained at 100% for 10 min, and returned to 0% B in 2 min. Total run time was 28 min including 3 min equilibration time at the end of the gradient. The MS analysis was performed under positive mode with a scanning range of m/z = 100-1700.

Inhibition of PCA against A. citrulli
Inhibition of PCA against A. citrulli was determined as previously described [45]. Strain XJX12 was cultured to late-log phase (OD 600 of approximately 1.0) in LB medium at 28 • C with shaking (200 rpm for 24 h). The bacterial suspension was diluted 1, 10, and 50 times using sterile liquid LB broth. A 2 µL volume of diluted bacterial suspension was spotted onto LB plates containing an acetone solution without or with PCA to achieve final PCA concentrations of 0, 2, 4, 8, 16, or 32 µg/mL and a final acetone concentration of 0.4% (v/v). The growth of A. citrulli cells was examined after 3 d at 28 • C. Each treatment was represented by three replicates, and the experiment was repeated three times.
Meanwhile, a 1 mL volume of bacterial suspension (OD 600 of approximately 1.0) was then added to 100 mL of fresh liquid LB medium containing PCA at 0, 2, 4, 8, 16, or 32 µg/mL (final acetone concentration 0.4% (v/v)). Strain XJX12 was grown at 28 • C with shaking (200 rpm for 12 h) before turbidity values were determined (2100-N Scattering Turbidimeter, Hach). The average turbidity values were used to calculate the PCA concentration that resulted in 50% inhibition of bacterial cell growth (EC 50 ). The EC 50 values were calculated with the Data Processing System computer program (Hangzhou Reifeng Information Technology Ltd.). Each treatment was replicated three times and each experiment was repeated thrice independently.

Transcript Levels of Transcriptional Regulator Genes as Affected by PCA
Strain XJX12 was grown in liquid LB medium, containing acetone solution without or with PCA. The concentration of PCA was 16 or 32 µg/mL. The final acetone concentration was 0.4% (v/v). RNA isolation and quantitative reverse-transcription polymerase chain reaction (qRT-PCR) were performed as previously described [46]. Strain XJX12 was grown to OD 600 values of 0.1 (with culture time 6 h), 0.2 (with culture time 8 h), 0.35 (with culture time 10 h), 0.5 (with culture time 12 h), 0.6 (with culture time 14 h), and 1.0 (with culture time 24 h), respectively, and a 2 mL volume of bacterial suspension was harvested by centrifugation (8000× g for 2 min at 4 • C). Total RNA was extracted using a total RNA Miniprep Purification Kit (Sangon Biotech, Shanghai). Total RNA was reverse-transcribed using PrimeScript TM RT Master Mix (TaKaRa, Japan). qRT-PCR was performed on a Roche Lightcycler 2.0 System (Switzerland) using Power TB Green PCR Master Mix (TaKaRa, Japan). The conserved XJ-RT gene of A. citrulli (accession number: CP000512.1), encoding amino acid adenylation domain protein, was used as an internal control in the qRT-PCR assay [47]. Each 20 µL RT-PCR reaction volume contained 1 ng cDNA, 5× TB Green Mix, and 4 µM of forward and reverse primers (Table S1). The C T values of oxyR or soxR used to determine transcript levels were normalized to those of the XJ-RT gene. Three replicates of RT-PCR analysis for each reaction were performed and repeated independently.
2.8. ROS Accumulation, Activities of Superoxide Dismutase (SOD) and CAT in A. citrulli Induced by PCA ROS accumulation was measured with a ROS assay kit (Beyotime, China) as previously described [46]. In brief, bacterial cultures of OD 600 values 1.0 were collected by centrifugation (6000× g, 10 min) and resuspended in 2 ,7 -dichlorodihydrofluorescein diacetate (DCFH-DA) solution, then incubated at 28 • C for 20 min. After being washed three times, the cells were resuspended in XOM medium (a compound mixture, provided by the assay kit) and added to microplates. The wells were treated with an acetone solution without or with PCA to achieve final PCA concentrations of 0, 16, or 32 µg/mL and a final acetone concentration of 0.4% (v/v). The microplates were kept at 28 • C, and an Infinite ® M1000 Pro fluorescence microplate reader (Tecan, Switzerland) was used to record the fluorescence for 4 h with an excitation wavelength of 488 nm and an emission wavelength of 525 nm. Each treatment was represented by three replicates, and the experiment was conducted three times.
SOD and CAT activities were assessed as previously described [46]. Strain XJX12 was grown in liquid LB medium, containing acetone solution without or with PCA. The concentration of PCA was 16 or 32 µg/mL. The final acetone concentration was 0.4% (v/v). At different time points, bacterial cells of strain XJX12 were treated with 200 µL of Cell Lysis Liquid (Beyotime, China). The protein concentration in the supernatant was determined with a BCA Protein Determination Kit (Beyotime, China), and total CAT and SOD activities were determined with the CAT Test Kit and SOD Test Kit (Beyotime, China).
The absorption values were recorded with visible spectrophotometer Epoch2 (BioTek). Each treatment was represented by three replicates, and the experiment was conducted three times.

Sensitivity of A. citrulli as Affected by Exogenous H 2 O 2
Strain XJX12 was cultured in LB medium at 28 • C with shaking (200 rpm for 24 h, OD 600 of approximately 1.0). The bacterial suspension was diluted 1, 10, and 50 times as described above. A 2 µL volume of diluted bacterial suspension was spotted onto LB plates containing 0, 0.25, 0.5, 0.75, 1.0, or 1.25 mM H 2 O 2 . The growth of A. citrulli cells was examined after 3 d at 28 • C. Each treatment was represented by three replicates, and the experiment was repeated three times.

Inhibition of PCA against A. citrulli as Affected by Exogenous CAT
A 1 mL volume of bacterial suspension (OD 600 of approximately 1.0) was added to 100 mL of fresh LB medium, which was then supplemented with acetone or acetone containing PCA to achieve a final PCA concentration of 16 or 32 µg/mL in strain XJX12 culture and a final acetone concentration 0.4% (v/v). After that, 0, 25, 50, 75, 100, 150, or 200 U of CAT were added to the medium. The cultures were kept at 28 • C with shaking at 200 rpm, and turbidity values were determined after 12 h. The effect of CAT on PCA inhibition against A. citrulli was expressed as a percentage of growth inhibition. Each treatment was represented by three replicates, and the experiment was repeated three times.

Statistical Analysis
Data were processed by analysis of variance (ANOVA) using the SAS GLM procedure (SAS Institute, Inc., Cary, NC, USA). p < 0.05 was considered statistically significant. Significant means were further compared using Fisher's protected least significant difference (PLSD).

Growth, Hydrogen Cyanide and Pyoverdine Production of P. chlororaphis YL-1 and Mutants
As expected, no difference was observed in the fermentation broth color or the growth capacity between P. chlororaphis YL-1 and its seven mutants in LB medium. Likewise, strain YL-1 and its mutants in LB medium did not exhibit any fluorescence under UV light. Fermentation broths of strain YL-1, mutants ∆phzE, ∆phzF, ∆prnB, ∆hcnB, ∆acsA, and ∆acsD in SM medium were yellow green in visible light and exhibited fluorescence under UV light, which were the special characteristics of pyoverdine produced by Pseudomonas spp. ( Figure S2); however, fermentation broth of the mutant ∆pvdL was white and did not exhibit fluorescence under UV light ( Figure S2). The growth capacity of all the mutants in SM medium was comparable to that of the wild-type strain.
A remarkable change in color from yellow to reddish-brown of strain YL-1 and mutants ∆phzE, ∆phzF, ∆prnB, ∆pvdL, ∆acsA, and ∆acsD compared with the mutant ∆hcnB indicates the production of hydrogen cyanide ( Figure S3). These results showed that the sacB-mediated genetic manipulation system used for P. chlororaphis YL-1 was stable and efficient.

Antibacterial Activity of P. chlororaphis YL-1 and Mutants
Antibacterial activities of strain YL-1 and mutants precultured in liquid LB medium were determined on LB plates. Strain YL-1 produced clear inhibitory zones on anti-A. citrulli and anti-Xoo activities ( Figure S4). The inhibitory zone against A. citrulli was 32.5 ± 1.8 mm while that against Xoo was 42.3 ± 1.7 mm in diameter. There was a slight inhibitory effect on Xooc by strain YL-1 on LB plates with inhibitory zone 9.7 ± 2.2 mm.
Strain YL-1 and five mutants, ∆prnB, ∆hcnB, ∆pvdL, ∆acsA, and ∆acsD, produced clear inhibitory zones with no observable difference in anti-A. citrulli or anti-Xoo activities. Inhibitory zones against A. citrulli ranged from 55.8 ± 2.3 mm to 66.1 ± 1.2 mm (Figure 1), and those against Xoo ranged from 70.2 ± 1.5 mm to 75.3 ± 2.1 mm in diameter ( Figure S5). However, there was little inhibitory effect on A. citrulli or Xoo treated by mutants ∆phzE and ∆phzF (Figure 1, Figure S5). These results suggest that phenazine is the key metabolite in antibacterial activity against A. citrulli. Strain YL-1 and five mutants, ΔprnB, ΔhcnB, ΔpvdL, ΔacsA, and ΔacsD, produced clear inhibitory zones with no observable difference in anti-A. citrulli or anti-Xoo activities. Inhibitory zones against A. citrulli ranged from 55.8 ± 2.3 mm to 66.1 ± 1.2 mm (Figure 1), and those against Xoo ranged from 70.2 ± 1.5 mm to 75.3 ± 2.1 mm in diameter ( Figure S5). However, there was little inhibitory effect on A. citrulli or Xoo treated by mutants ΔphzE and ΔphzF (Figure 1, Figure S5). These results suggest that phenazine is the key metabolite in antibacterial activity against A. citrulli. Figure 1. Antibacterial activity of P. chlororaphis YL-1 and its seven mutants on LB plates. All strains precultured in liquid LB medium at 28 °C and 200 rpm for 24 h were resuspended in distilled liquid LB medium to the desired cell density of 10 8 CFU/mL. After spraying the suspension of A. citrulli (10 8 CFU/mL) onto the surface of LB plates for 1 s, 5 μL cells of strain YL-1 or mutants were injected into the central hole (5 mm) of LB plates and incubated at 28 °C for 48 h. Distilled LB medium was used as a control. The inhibitory zone diameters of strain YL-1 and mutants were measured when A. citrulli covered the control plates entirely. Each treatment was replicated three times, and each experiment was repeated thrice independently.

Isolation and Purification of Phenazine
In this study, it was found that one comparable peak was produced by STD sample (blue line) and the wild-type strain YL-1 (pink line) under the absorbance wavelength of 254 nm, with the retention time 2.0 min (Figure 2a). No obvious peak was detected from the samples of the mutants ΔphzE (green line) and ΔphzF (black line) (Figure 2b). A comparable peak was also detected at a higher level in STD sample, compared to that extracted from strain YL-1 (Figure 2a). The fraction collected from 1.95 to 2.10 min was further purified and analyzed by LC-MS. As shown in Figure 3, the characteristic ions of the mass spectrum in strain YL-1 are also consistent with those in STD sample, with m/z of 225.07, 226.07, and 227.07 (three isotopic peaks). The main antibacterial compound produced by strain YL-1 in LB medium was identified as PCA. Figure 1. Antibacterial activity of P. chlororaphis YL-1 and its seven mutants on LB plates. All strains precultured in liquid LB medium at 28 • C and 200 rpm for 24 h were resuspended in distilled liquid LB medium to the desired cell density of 10 8 CFU/mL. After spraying the suspension of A. citrulli (10 8 CFU/mL) onto the surface of LB plates for 1 s, 5 µL cells of strain YL-1 or mutants were injected into the central hole (5 mm) of LB plates and incubated at 28 • C for 48 h. Distilled LB medium was used as a control. The inhibitory zone diameters of strain YL-1 and mutants were measured when A. citrulli covered the control plates entirely. Each treatment was replicated three times, and each experiment was repeated thrice independently.

Isolation and Purification of Phenazine
In this study, it was found that one comparable peak was produced by STD sample (blue line) and the wild-type strain YL-1 (pink line) under the absorbance wavelength of 254 nm, with the retention time 2.0 min (Figure 2a). No obvious peak was detected from the samples of the mutants ∆phzE (green line) and ∆phzF (black line) (Figure 2b). A comparable peak was also detected at a higher level in STD sample, compared to that extracted from strain YL-1 (Figure 2a). The fraction collected from 1.95 to 2.10 min was further purified and analyzed by LC-MS. As shown in Figure 3, the characteristic ions of the mass spectrum in strain YL-1 are also consistent with those in STD sample, with m/z of 225.07, 226.07, and 227.07 (three isotopic peaks). The main antibacterial compound produced by strain YL-1 in LB medium was identified as PCA.    Reversed phase (RP)-HPLC chromatograms. An overlay of the chromatograms at 254 nm of STD sample (blue) and the final extraction step of the wild-type YL-1 (pink) (a), mutants ΔphzE (green), and ΔphzF (black) (b), using an XDB-C18 reversed-phase column (4.6 mm  150 mm, 5 μm, Agilent), is shown. Here, STD sample (blue) was used as a positive control. The samples were eluted with acetonitrile (ACN)-5 mM ammonium acetate (60:40, v/v) at a flow rate of 0.7 mL/min. The experiment was repeated three times.

The Inhibition of PCA against A. citrulli
As shown in Figure 4, colony growth of A. citrulli was inhibited when PCA concentration in LB plates reached 8 µg/mL. A. citrulli can grow well on LB plates containing 4 µg/mL PCA even if diluted 50 times. However, it can hardly grow on LB plates containing 32 µg/mL PCA.

The Inhibition of PCA against A. citrulli
As shown in Figure 4, colony growth of A. citrulli was inhibited when PCA concentration in LB plates reached 8 μg/mL. A. citrulli can grow well on LB plates containing 4 μg/mL PCA even if diluted 50 times. However, it can hardly grow on LB plates containing 32 μg/mL PCA. The inhibition of PCA against A. citrulli was more sensitive in liquid culture than that on LB plates ( Figure 5). The inhibition rates were 10  The inhibition of PCA against A. citrulli was more sensitive in liquid culture than that on LB plates ( Figure 5). The inhibition rates were 10  The inhibition of PCA against A. citrulli was more sensitive in liquid culture than that on LB plates ( Figure 5). The inhibition rates were 10.96%, 19.77%, 45.18%, or 66.11% when PCA concentration in liquid LB medium reached 4, 8, 16, or 32 μg/mL, respectively. The linear regression equation was y = 2.7545 + 1.7315x and the EC50 value of PCA was 19.81 μg/mL.

Expression of oxyR, soxR, and the Genes Involved in Antioxidant Systems as Affected by PCA
To determine whether transcriptional regulators SoxR and OxyR were activated by PCA, and to determine whether the genes encoding antioxidant enzymes were also induced by PCA in A. citrulli, expression of the genes of interest was performed using qRT-PCR. When the concentrations of PCA were 16 or 32 μg/mL, the expression levels of oxyR were 1.33 or 2.13 times higher at 6 h (Figure 6a), and 1.84 or 3.31 times higher at 8 h ( Figure  6b) than those of the acetone control, respectively. Then, the expression level of oxyR decreased to control level at 10 h (Figure 6c). The expression levels of katB, ahpC, and ahpF

Expression of oxyR, soxR, and the Genes Involved in Antioxidant Systems as Affected by PCA
To determine whether transcriptional regulators SoxR and OxyR were activated by PCA, and to determine whether the genes encoding antioxidant enzymes were also induced by PCA in A. citrulli, expression of the genes of interest was performed using qRT-PCR. When the concentrations of PCA were 16 or 32 µg/mL, the expression levels of oxyR were 1.33 or 2.13 times higher at 6 h (Figure 6a), and 1.84 or 3.31 times higher at 8 h (Figure 6b) than those of the acetone control, respectively. Then, the expression level of oxyR decreased to control level at 10 h (Figure 6c). The expression levels of katB, ahpC, and ahpF regulated by OxyR began to increase at 6 h after PCA treatment. At 8 h, treated by 16 or 32 µg/mL PCA, the expression levels of katB, ahpC, and ahpF were 8.00 or 20.46 times higher, 1.49 or 3.26 times higher, and 1.49 or 2.14 times higher than those of the acetone control, respectively (Figure 6b). The expression level of katG was 1.65 or 2.67 times higher than that of the control at 6 h treated by 16 or 32 µg/mL PCA, then it began to decrease sharply. The expression of other genes (katE and trxA) had no significant change from 6-10 h after being treated by PCA. The expression level of soxR, encoding another transcription regulator SoxR, increased at 6 h after being treated by 16 or 32 µg/mL PCA, being 1.46 or 1.64 times higher than that of the control (Figure 6a), then decreasing to control level at 8 h. However, the expression of sodA, sodB, and sodC remained unchanged from 6-10 h after being treated by PCA. These results indicate that katB in A. citrulli plays an important role in the process of protecting the bacteria against PCA-induced oxidative stress.
sharply. The expression of other genes (katE and trxA) had no significant change from 6-10 h after being treated by PCA. The expression level of soxR, encoding another transcription regulator SoxR, increased at 6 h after being treated by 16 or 32 μg/mL PCA, being 1.46 or 1.64 times higher than that of the control (Figure 6a), then decreasing to control level at 8 h. However, the expression of sodA, sodB, and sodC remained unchanged from 6-10 h after being treated by PCA. These results indicate that katB in A. citrulli plays an important role in the process of protecting the bacteria against PCA-induced oxidative stress.

ROS Accumulation, CAT and SOD Activities Induced by PCA
ROS levels in A. citrulli increased significantly when treated with PCA or Rosup reagent (the positive control) (Figure 7a). ROS accumulation was greater with 16 or 32 µg/mL PCA than with Rosup reagent. As expected, ROS accumulation was not detected in the negative control or acetone treatment. assay. The experiment was repeated three times.

ROS Accumulation, CAT and SOD Activities Induced by PCA
ROS levels in A. citrulli increased significantly when treated with PCA or Rosup reagent (the positive control) (Figure 7a). ROS accumulation was greater with 16 or 32 μg/mL PCA than with Rosup reagent. As expected, ROS accumulation was not detected in the negative control or acetone treatment. Figure 7. ROS accumulation (a) and CAT (b) activity of A. citrulli strain XJX12 induced by PCA. The treatments included PCA in acetone and acetone alone. ROS accumulation was measured based on fluorescence in microplates, the wells of which contained a bacterial suspension and no additional agent (the negative control) or one of the following: PCA at 0, 16, or 32 μg/mL, with 0.4% (v/v) acetone (as the solvent) in each case; acetone alone (0.4%, v/v); or Rosup reagent (a positive control provided with the assay kit). The experiment was repeated three times.
As shown in Figure 7b, CAT activity in A. citrulli was found to increase from 6 h to 12 h after being treated by acetone, 16 or 32 μg/mL PCA, with the maximum value 12.53, 13.09, or 15.02 U at 8 h, respectively. Then, CAT activity was decreased to normal level at 14 h. No significant difference was found for SOD activity treated by acetone, 16 or 32 Figure 7. ROS accumulation (a) and CAT (b) activity of A. citrulli strain XJX12 induced by PCA. The treatments included PCA in acetone and acetone alone. ROS accumulation was measured based on fluorescence in microplates, the wells of which contained a bacterial suspension and no additional agent (the negative control) or one of the following: PCA at 0, 16, or 32 µg/mL, with 0.4% (v/v) acetone (as the solvent) in each case; acetone alone (0.4%, v/v); or Rosup reagent (a positive control provided with the assay kit). The experiment was repeated three times.
As shown in Figure 7b, CAT activity in A. citrulli was found to increase from 6 h to 12 h after being treated by acetone, 16 or 32 µg/mL PCA, with the maximum value 12.53, 13.09, or 15.02 U at 8 h, respectively. Then, CAT activity was decreased to normal level at 14 h. No significant difference was found for SOD activity treated by acetone, 16 or 32 µg/mL PCA. These data suggest the A citrulli cells enhanced CAT activity to respond to the PCA treatment.
It should be noted that acetone (the solvent of PCA) reaching a certain concentration would have an inhibitory effect on A. citrulli; therefore, the concentration of acetone in the research study was set as 0.4% (v/v). A. citrulli was not inhibited and can grow well on LB plates ( Figure 4) and in LB liquid medium ( Figure 5) containing 0.4% acetone (v/v). Although 0.4% acetone has a slight influence on CAT activity in A. citrulli (Figure 7), significant difference was found as compared with 32 µg/mL PCA from 6 h to 10 h.

Sensitivity of A. citrulli as Affected by Exogenous H 2 O 2
Strain XJX12 was sensitive to H 2 O 2 and did not grow on LB plates containing 1.5 mM H 2 O 2 . It grew well at 0.25 mM even if diluted 50 times. The growth of strain XJX12 was inhibited on LB plates containing 0.5 mM H 2 O 2 when diluted 50 times, and those containing 1.0 mM H 2 O 2 when diluted 10 times ( Figure S6). These data demonstrate that bacterial cells are sensitive to the accumulation of active oxygen.

Inhibition of PCA against A. citrulli as Affected by Exogenous CAT
The inhibition effect of PCA against A. citrulli was lowered by adding exogenous CAT. As shown in Figure 8, when treated with the same concentration of PCA (32 µg/mL), decreases in inhibition effect correlated to increases in the exogenous CAT content of the medium, with inhibition rates of 66.10 ± 7.93, 58.64 ± 7.03, 50.24 ± 6.02, 44.40 ± 5.32, 34.96 ± 4.19, 20.07 ± 2.41, and 9.42 ± 1.13 when treated with 0, 25, 50, 75, 100, 150, and 200 U CAT in liquid LB medium, respectively. As shown in Figure 8, the inhibition effect upon treatment with 16 µg/mL of PCA was substantially weaker than that upon treatment with 32 µg/mL at any concentration of exogenous CAT.
acetone (v/v). Although 0.4% acetone has a slight influence on CAT activity in A. citrulli (Figure 7), significant difference was found as compared with 32 μg/mL PCA from 6 h to 10 h.

Sensitivity of A. citrulli as Affected by Exogenous H2O2
Strain XJX12 was sensitive to H2O2 and did not grow on LB plates containing 1.5 mM H2O2. It grew well at 0.25 mM even if diluted 50 times. The growth of strain XJX12 was inhibited on LB plates containing 0.5 mM H2O2 when diluted 50 times, and those containing 1.0 mM H2O2 when diluted 10 times ( Figure S6). These data demonstrate that bacterial cells are sensitive to the accumulation of active oxygen.

Inhibition of PCA against A. citrulli as Affected by Exogenous CAT
The inhibition effect of PCA against A. citrulli was lowered by adding exogenous CAT. As shown in Figure 8, when treated with the same concentration of PCA (32 μg/mL), decreases in inhibition effect correlated to increases in the exogenous CAT content of the medium, with inhibition rates of 66.10 ± 7.93, 58.64 ± 7.03, 50.24 ± 6.02, 44.40 ± 5.32, 34.96 ± 4.19, 20.07 ± 2.41, and 9.42 ± 1.13 when treated with 0, 25, 50, 75, 100, 150, and 200 U CAT in liquid LB medium, respectively. As shown in Figure 8, the inhibition effect upon treatment with 16 μg/mL of PCA was substantially weaker than that upon treatment with 32 μg/mL at any concentration of exogenous CAT.

Discussion
Genes responsible for phenazine biosynthesis are organized in a seven-gene operon phz-ABCDEFG, which is conserved in all phenazine-producing Pseudomonas spp. [29]. The first enzyme involved in phenazine biosynthesis is a type II 3-deoxy-D-arabinoheptulosonate-7phosphate (DAHP) synthase, PhzC. Five enzymes-PhzE, PhzD, PhzF, PhzB, and PhzGare conserved in all phenazine-producing bacteria, leading to PCA production in Pseudomonas spp., but the same enzymes can also produce phenazine-1,6-dicarboyxlic acid (PDC) in other phenazine-producing bacteria [30]. The gene phzA, present in Pseudomonas spp., has a high homology to phzB (≈70% identity) and is important in PCA production [30]. Some phenazine-producing Pseudomonas spp. harbor accessory phenazine biosynthetic genes, such as phzH and phzO, which are located immediately after phzG. These genes encode phenazine-modifying enzymes that are responsible for the production of strainspecific phenazine derivatives [30]. Bioinformatics analysis of the genome sequence (accession number: NZ_AWWJ00000000.1) revealed the presence of phzABCDEFGO gene clusters in P. chlororaphis YL-1. We selected two conserved genes, phzE and phzF, to generate in-frame deletion mutants; the mutants ∆phzE and ∆phzF did not produce PCA and lost their antibacterial activity compared with wild-type strain YL-1. Surprisingly, complementation of the mutants, using the functional phzE and phzF gene fragments harbored in the expression vector pUCP26 [48], failed, although considerable efforts were made. In fact, mutations of multiple phz genes could not be complemented in numerous studies [49][50][51]. The Pseudomonas expression vector pUCP26 has been successfully used for genetic complementation of mutations in Pseudomonas [25,52]. It remains to be investigated if there are any effects of copy number on gene expression in this case. As expected, gene-specific deletions in the mutant ∆phzE and ∆phzF, which were generated by the sac B-based counter-selection marker mutagenesis approach [38], were confirmed by sequencing analysis ( Figure S1). Nevertheless, the results suggest that PCA is the main antibacterial phenazine compound produced by stain YL-1.
Secondary metabolites produced by a microorganism and its antimicrobial activities can vary depending on the culture media used for growth. Previous results showed that the main antibacterial secondary metabolite produced by strain YL-1 is pyoverdine in low-iron condition. Compared with wild-type strain YL-1, pyoverdine-deficient mutants cannot produce pyoverdine and their antibacterial activity reduced significantly when grown on low-iron medium [25]. However, the antibacterial activity of PVD-deficient mutants did not change in iron-sufficient conditions (such as LB medium) compared with that of wild-type strain YL-1 [25]. These results suggest that strain YL-1 and PVD-deficient mutants secrete different antibacterial compounds when cultured on LB plates as compared with low-iron conditions. In this study, we evaluated the antimicrobial activity of strain YL-1 on LB plates as compared with its phenazine-deficient mutants ∆phzE and ∆phzF, pyrrolnitrindeficient mutant ∆prnB, hydrogen-cyanide-deficient mutant ∆hcnB, pyoverdine-deficient mutant ∆pvdL, and achromobactin-deficient mutants ∆acsA and ∆acsD. The results showed that only phenazine-deficient mutants ∆phzE and ∆phzF almost lost the inhibitory effects against A. citrulli and Xoo on LB plates compared with strain YL-1 (Figure 1 and Figure  S5), and that the main antibacterial compound produced by strain YL-1 on LB plates was PCA, based on LC-MS/MS analysis (Figures 2 and 3). Later, it was also verified that purified PCA produced by strain YL-1 can inhibit the growth of A. citrulli in LB plates and in liquid culture (Figures 4 and 5). These findings of the research further demonstrate that P. chlororaphis YL-1 can produce multiple antimicrobial compounds to target various microbial organisms under different conditions.
According to the reports of comparative genomic and core genome analyses of the various Pseudomonas groups, phenazine biosynthetic gene phzF is present in 86% (37/43) of all tested P. chlororaphis strains while as a core gene it exists in the analyzed P. aeruginosa strains (189/189). The pvdL gene was identified as a core gene in all tested P. chlororaphis strains (43/43) while it exists in 90% (171/189) of the analyzed P. aeruginosa strains and 81% (212/262) of all the other Pseudomonas strains [53]. These results indicate that both phenazine and pyoverdine biosynthesis genes are distributed widely across the Pseudomonas genus. In addition, it was reported that production of antifungal compounds phenazine and pyrrolnitrin in P. chlororaphis O6 is differentially regulated by glucose [54]. More research is needed to investigate any possible variation of PCA production under a different medium. Moreover, it is worth noting that PCA is categorized as an irritant in PubChem toxicology databases and it will be harmful to eyes, skin, and the respiratory tract after inappropriate operation.
PCA is a natural product that is isolated from Pseudomonas spp. and is used to control many rice diseases in China, such as rice sheath blight, rice blast, rice bacterial leaf blight, and rice bacterial leaf streak [55]. PCA also exhibits extensive antimicrobial activities against phytopathogens in vitro, such as Xoo, Xooc, Rhizoctonia solani, Alternaria solani, Fusarium oxysporum, Fusarium graminearum, and Pyricularia oryzac, with EC 50 values of 0. 18, 13.11, 7.44, 18.26, 41.78, 39.51, and 8.36 µg/mL, respectively [55,56]. To our knowledge, there was only one report that PCA, produced by a marine bacterium P. aeruginosa PA31x, can inhibit the growth of A. citrulli in vitro, with the inhibitory efficiency 70% when treated by 10 mg/mL of PCA [57]. However, the mechanism of PCA against A. citrulli remains unclear. In this study, A. citrulli can hardly grow on LB plates and the inhibitory efficiency in the liquid coculture test was 66.11% when treated by 32 µg/mL of PCA. The EC 50 value of PCA against A. citrulli was 19.81 µg/mL. Moreover, we also found that PCA enhanced the accumulation of ROS and increased the activity of CAT in A. citrulli. The inhibition of PCA against A. citrulli was reduced by adding exogenous CAT (Figure 8), indicating that PCA can interfere with electron transport chains, generating ROS, damaging the target bacteria, and ultimately leading to cell death in bacterial cells [58]. P. chlororaphis YL-1 can inhibit the growth of both Gram-positive and Gram-negative bacteria, such as Burkholderia glumae, Clavibacter michiganesis, and Erwinia amylovora [22]. To discern whether PCA produced by strain YL-1 is also the main antibacterial secondary metabolite against Gram-positive bacteria and whether its mechanism is the same as those in Gram-negative bacteria, further investigation is needed.
Two key transcriptional regulators, SoxR and OxyR, have been well known to respond to oxidative stress and regulate antioxidant-related enzymes in most bacteria. SoxR is activated in response to O 2 and, in turn, activates SOD. In contrast, OxyR is activated in response to H 2 O 2 and, in turn, activates CAT, Trx, and AhpCF [46]. In this study, we found that expression levels of katB, katG, and ahpCF regulated by transcription regulator OxyR in A. citrulli increased 6 h after being affected by PCA. Among them, the most significant one is katB, with expression levels 8.00 or 20.46 times higher than those of the acetone control 8 h after being treated by 16 or 32 µg/mL PCA, respectively (Figure 6b). Similarly, expression levels of catB in Xoo were 18.24 times higher than those of the acetone control when treated by 4 µg/mL PCA while they were 6.63 times higher in Xooc when treated by 32 µg/mL PCA [46]. However, the expression of katE and trxA in both Xoo and Xooc increased when affected by PCA [46], while not being changed in A. citrulli. The expression of sodA, sodB, and sodC in A. citrulli regulated by SoxR remained unchanged when affected by PCA. PCA only induced the expression of sodM1 in Xoo while PCA did not induce the expression of soxR or the genes encoding SOD in Xooc [46]. These results show that different bacteria have various CATs that are encoded and activated by different genes although the accumulation of ROS in A. citrulli affected by PCA is similar to that in Xoo [55]. Currently there is no fully clarified association network of related genes in response to oxidative stress in the above described bacteria, but there is evidence that expression intensity of these different genes determines the difference in antioxidant capacity of bacteria. For example, it was reported that Xooc has a stronger antioxidant system than Xoo and accumulates lower ROS in response to redox compounds by deleting or exchanging the antioxidant-related genes, although Xoo and Xooc are very closely related [55]. Therefore, our results provide a scientific basis for the study of antioxidant capacity, virulence, and control of A. citrulli in the future.
Moreover, the different genes encoding antioxidant enzymes as described above possess different functions at different growth phases. In a rhizobium, Azorhizobium caulinodans, KatG and OxyR are not only critical for antioxidant defense in vitro, but are also important for nodule formation and nitrogen fixation during interaction with plant hosts [59]. P. aeruginosa possesses an extensive armament of genes involved in oxidative stress defense, including katB-ankB, ahpB, and ahpC-ahpF. Expression of these genes was not observed in ∆oxyR mutant, indicating that OxyR was essential for this response. Moreover, increased katB expression and higher CAT levels were detected in ∆ahpCF, suggesting a compensatory function for KatB in the absence of AhpCF [60]. Cyclic dimeric guanosine monophosphate (c-di-GMP), as a dinucleotide second messenger, increased CAT activity and expression of the CAT encoding gene katB in Vibrio cholerae [61]. The monofunctional CAT, KatE, of X. axonopodis pv. citri is not only important for its colonization and survival in citrus plants, but also is required for full virulence in citrus plants [62]. In a variety of bacteria, including mycobacteria, KatG is a unique dual function enzyme acting as CAT and peroxidase, which are important for protection against damage from ROS [63]. SOD isozymes in Klebsiella pneumonia have unique roles beyond oxidative stress resistance, such as biofilm formation, cell morphology, metabolism, and in vivo colonization and persistence and there is a regulatory interplay among SODs [64]. The FurA gene, which codes for a ferric uptake regulator, is required for A. citrulli virulence on watermelon; Ac∆furA displayed increased sensitivity to H 2 O 2 and significant downregulation of sodB [47]. As far as we know, the transcriptional regulator and related genes involved in oxidative stress defense were seldom reported in A. citrulli. Therefore, it remains to be investigated to explore more functions of antioxidant-related enzymes in A. citrulli.
In conclusion, the findings of this research demonstrate that strain YL-1 could produce PCA, which has strong inhibitory effects on Gram-negative bacteria as demonstrated by using bacterial cells and purified compounds, when grown on LB medium. PCA can enhance the accumulation of reactive oxygen species (ROS) and the activity of catalase (CAT) in A. citrulli, and the inhibition effect of PCA against A. citrulli was lowered by adding exogenous CAT. The results suggest that strain YL-1 is a promising biocontrol bacterium and PCA could inhibit the growth of A. citrulli in vitro. However, biocontrol bacteria with in vitro antagonistic activity may not effectively control plant diseases in the field. Additional studies regarding the characteristics of PCA, such as stability, the duration of the efficacy, and product formulation, are needed to confirm the biocontrol effect of PCA under field conditions. Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/microorganisms9102012/s1, Figure S1: Two deletion mutants ∆phzE and ∆phzF of Pseudomonas chlororaphis YL-1, Figure S2: Comparison of the pyoverdine production between P. chlororaphis YL-1 and mutants cultured in liquid SM, Figure S3: Comparison of the production of hydrogen cyanide between P. chlororaphis YL-1 and mutants cultured in King's B medium containing glycine (4.5 g/L), Figure S4: Antibacterial activity of P. chlororaphis YL-1 on LB plates, Figure S5: Anti-Xoo activity of P. chlororaphis YL-1 and its seven mutants on LB plates, Figure S6: H 2 O 2 sensitivity of A. citrulli on LB plates, Table S1: Strains, plasmids and primers used in this study.