Approximately 4000 transconjugants were screened in radial diffusion plate assays to identify mutants displaying increased or decreased antifungal activity compared to the wild type. One mutant was identified, PA23-443, that exhibited no antifungal activity and was white in colour, indicating a loss of phenazine production 5 (Figures 1 and 2B). DNA flanking the Tn exhibited 89% identity at the amino acid level to a Pseudomonas fluorescens LTTR [Genbank: AAY90576]. The newly identified gene was designated ptrA. To verify that the phenotype of PA23-443 was due to ptrA inactivation, the ptrA gene was PCR amplified and cloned into pUCP22 for complementation. The presence of pUCP22-ptrA restored antifungal activity to that of the wild type (Figure 1).

Sequence analysis revealed that the site of Tn insertion lies 803 bp downstream of the PtrA translational start (data not shown), which is predicted to disrupt the co-inducer recognition/response domain 15 . Previous studies of the LTTRs NodD and NahR revealed that mutations in this region result in a co-inducer-independent phenotype which affects DNA binding and thus the activation/repression properties of the proteins 14 15 . Directly downstream of ptrA but in the opposite orientation lies a gene encoding a protein that is 99% identical at the amino acid level to a DoxX-family protein found in P. chlororaphis subsp. aurantiaca PB-St2 [Genbank accession #WP_023968058]. Based on sequence similarity, DoxX could be involved in pathways related to elemental sulfur oxidation 16 . Immediately upstream of ptrA, in the opposite orientation, lies a gene encoding a short-chain dehydrogenase (scd). Short-chain dehydrogenases are part of a superfamily of enzymes designated as the NAD(H)- or NADP(H)-dependent short-chain dehydrogenases/reductases (SDRs). The SDRs comprise a very large grouping of biologically important proteins found in virtually all forms of life 17 . At present, it is unclear whether the genes upstream and downstream of ptrA play a role in regulation.

Through blastn analysis, ptrA homologs were found within the genomes of several Pseudomonas species, with the highest degree of nucleotide identity exhibited by Pseudomonas sp. UW4 (85%), followed by Pseudomonas protegens strains Pf-5 (84.7%) and CHA0 (84.7%), Pseudomonas fluorescens strains Pf0-1 (84.5%) and F113 (82.5%), Pseudomonas brassicacearum subsp. brassicacearum NFM421 (82.4%), Pseudomonas poae RE*1-1-14 (79.3%), and Pseudomonas resinovorans NBRC 106553 (76.1%) 18 . Collectively, our findings indicate that PtrA is a newly identified regulator of PA23 biocontrol, and homologs of this regulator are present in a number of Pseudomonas species.

PtrA belongs to the LTTR family, which is the largest known family of prokaryotic DNA binding proteins 14 . LTTRs can function as either repressors or activators for single or operonic genes. Furthermore, these regulators may be divergently transcribed from their target genes or may control expression of numerous genes scattered about the chromosome 14 . In PA23, expression of antifungal metabolites is governed by a complex network of regulatory elements and substantial interaction occurs between the regulators themselves 4 11 12 13 . To understand the global impact of the ptrA mutation on PA23 physiology, iTRAQ proteomic analysis was carried out to reveal proteins that were differentially expressed between the PA23 wild type and the ptrA mutant. A total of 771 proteins were matched to proteins found within the P. chlororaphis gp72 reference genome 19 . Fifty nine of these proteins were differentially expressed between the two strains, exhibiting a vector difference (V_diff) greater than or equal to +1.65 and less than or equal to −1.65, corresponding to proteins in the upper or lower 10% of the population distribution (Table 1). The 59 proteins could be classified into 16 clusters of orthologous groups (COGs) based on their predicted function. Figure 3 summarizes the classification of the identified proteins, indicating significant up- or downregulation of protein expression. The largest COG category was the unknown function group, suggesting that many yet-to-be-identified proteins play a role in the loss of biocontrol exhibited by PA23-443.

Proteins with V_diff ≥ +1.65 or V_diff ≤ −1.65, corresponding to proteins expressed in the upper or lower 5% of the population distribution are shown.

^alog₂(tag₁₁₅/tag₁₁₇).

The secondary metabolite biosynthesis, transport and catabolism COG category represented the next largest grouping (Table 1). Initially, two of the proteins (MOK_01048, MOK_01053) were classified under the general function category and one protein (MOK_01054) was categorized under the transport and metabolism grouping. Upon further investigation, the locus tags indicated that they are part of the phenazine biosynthetic operon, leading to their reclassification into the secondary metabolite biosynthesis COG.

The phenazine operon has been well characterized in many pseudomonads, with phzABCDEFG comprising the core biosynthetic locus 20 . In this study, proteins with locus tags MOK_01048 and MOK_01049, identified as phenazine biosynthesis protein A/B, were significantly downregulated (Table 1). All phenazine-producing pseudomonads have an adjacent and nearly identical copy of the phzB gene, termed phzA 20 . PhzA catalyzes the condensation reaction of two ketone molecules in the phenazine biosynthesis pathway 20 . PhzF (identified as MOK_01053 in this study) works as an isomerase, converting trans-2,3-dihydro-3-hydroxyanthranilic acid (DHHA) into 6-amino-5-oxocyclohex-2-ene-1-carboxylic acid prior to the condensation reaction catalyzed by the PhzA/B proteins 20 . phzG encodes an FMN-dependent pyridoxamine oxidase (identified as MOK_01054 in this study), which is hypothesized to catalyze the conversion of DHHA to 5,10-Dihydro-PCA 21 . In some pseudomonads, genes downstream of the core biosynthetic operon are required for generation of phenazine derivatives 22 23 24 . In P. chlororaphis 30–84, for example, phzO lies downstream of the core operon; PhzO is an aromatic hydroxylase that catalyzes the conversion of PCA into 2-OH-PHZ 23 . More recently, in P. chlororaphis gp72, the phzO gene was shown to convert PCA into 2-OH-PHZ through a 2-OH-PCA intermediate 25 . Like other P. chlororaphis strains, PA23 produces 2-OH-PHZ and we believe the downregulated aromatic ring hydroxylase (MOK_01055) is PhzO. Therefore, in the absence of a functional ptrA gene, four of the core phenazine biosynthetic enzymes (PhzA, PhzB, PhzF, PhzG) and one aromatic ring hydroxylase (PhzO) are significantly downregulated. The fact that PtrA plays a critical role in regulating phz expression was not surprising considering the lack of orange pigment produced by the ptrA mutant (Figures 1 and 2A). Reduced phenazine expression was further substantiated by quantitative assays. As illustrated in Figure 2B, there is a 15-fold decrease in phenazine production in PA23-443 compared to the PA23 wild type. When ptrA was expressed in trans, some restoration of phenazine production was achieved.

Our iTRAQ proteomic results showed that two chitinase enzymes (MOK_03378 and MOK_05478) were significantly downregulated in the PA23-443 mutant (Table 1). These results were supported by chitinase assays, which clearly indicated no detectable enzyme activity in the ptrA mutant (Table 2). Addition of plasmid-borne ptrA elevated chitinase activity close to that of the wild type (Table 2). Collectively our findings indicate that ptrA is necessary for chitinase production. The LTTR, ChiR, has been previously shown to indirectly regulate all chitinases produced in Serratia marcescens 2170 26 . Proteomic analysis of a P. aeruginosa gacA mutant revealed that chitinase (ChiC) and a chitin-binding protein (CbpD) were decreased 8-fold and 2.2-fold respectively, as compared to the wild type 27 .

^aMean (standard deviation) of enzyme activity of three replicates.

^bSignificantly different from wild type (P < 0.005).

^cSignificantly different from wild type (P < 0.0001).

^dNot significantly different from wild type.

^eSignificantly different from wild type (P < 0.05).

In the ptrA mutant, a lipoprotein involved in iron transport (MOK_05447) was found to be significantly upregulated (Table 3). This finding prompted us to explore whether the mutant exhibited elevated siderophore expression. Siderophores are thought to contribute to biocontrol by sequestering iron, thereby restricting pathogen growth. Following 24 hours growth on CAS agar plates, mutant PA23-443 showed a 3-fold increase in the size of the orange halo surrounding the colony, indicating increased siderophore production compared to the wild type (Table 3). As expected, overexpression of ptrA restored the wild-type phenotype. Since the ptrA mutant expresses significantly increased levels of siderophore but exhibits a complete loss of antifungal activity, it is clear that elevated siderophore expression alone is not sufficient for S. sclerotiorum control.

^aMean (standard deviation) of orange haloes (mm) surrounding colonies on CAS agar. Five replicates were examined.

^bSignificantly different from the wild type (p < 0.0001).

^cNot significantly different from the wild type.

We observed significant upregulation of proteins involved in translation, ribosomal structure and biogenesis in the ptrA mutant (Table 1). These proteins include a translation elongation factor (MOK_00565), a tRNA amidotransferase (MOK_02337) and ribosomal proteins L32 and S19 (MOK_01324 and MOK_04471, respectively) which make up structural components of both the large and small ribosomal subunits of the 70S ribonucleoprotein complex 28 (Table 1). To determine whether PA23-443 exhibited an altered pattern of growth compared to the wild type, growth rate analysis was undertaken. As depicted in Figure 4, the mutant enters the logarithmic (log) growth phase around hour 8, which starts to plateau by hour 13. Conversely, the PA23 wild type does not enter log phase until hour 11, ending with entrance into early stationary phase at 19 hours of growth. Another interesting difference observed was the maximum population density achieved. The PA23 wild type consistently reached a higher OD₆₀₀ in stationary phase compared to PA23-443 (Figure 4). A similar altered pattern of growth has been observed for gacS mutants of PA23 and 30–84 which exhibit a shorter lag phase and earlier entry into logarithmic growth phase 4 29 . LTTRs have previously been implicated in the regulation of cellular growth factors. For example, the well-studied LTTR OxyR is involved in regulating the expression of various metabolic genes such as tRNA nucleotidyl transferases and synthetases, ribosomal proteins and QS-regulated targets 30 .

Our iTRAQ proteomic data indicated upregulation of the flagellin and related hook-associated protein (MOK_01499) in PA23-443. Further inspection of the locus tags upstream of MOK_01499 also indicated upregulation of proteins FliG (MOK_01489; V_diff = +0.72) and FliS (MOK_01496; V_diff = +0.66), although this upregulation was not considered significant. The upregulated flagellin and related hook-associated protein, therefore, is likely part of the Fli operon based on its proximity to upstream genes. To verify the results of the proteomic analysis, motility assays were conducted. As outlined in Table 4, swimming (flagellar) motility was almost 3-fold greater in PA23-443 compared to the wild type, indicating that PtrA is having a repressive effect on this phenotype. In a similar fashion, proteomic analysis of a P. aeruginosa gacA mutant revealed a 7.5-fold and 8.8-fold increase in expression of a flagellin (FliC) and flagellar-capping protein (FliD), respectively 27 . Introduction of ptrA in trans caused a modest reduction in motility, but did not fully restore the wild-type phenotype. It is important to bear in mind that for our complementation studies, multiple copies of the ptrA gene were provided rather than a single chromosomal copy. Because LTTRs bind both activation binding sites and regulatory binding sites upstream of target genes 14 , the number of copies of the regulator may be of critical importance for proper binding and subsequent regulation of target genes. This observation was noted with complementation studies involving the LTTR OxyR in restoration of rhamnolipid and pyocyanin production in P. aeruginosa 31 . When multiple copies of oxyR were present in the cell, the wild-type phenotype was not restored; whereas insertion of single chromosomal copy of the LTTR gene resulted in full complementation 31 .

^aMean (standard deviation) of swim zones from four replicates.

^bSignificantly different from the wild type (p < 0.0001).

Based on iTRAQ analysis, a tryptophan halogenase (MOK_04031) was identified under the amino acid transport and metabolism COG category, but was not significantly differentially expressed in the ptrA mutant (V_diff = −0.24). At locus tag MOK_04033, another chlorinating halogenase was identified in the P. chlororaphis gp72 genome, but was not differentially expressed in the ptrA mutant. These enzymes are likely prnA and prnC, forming part of the prnABCD pyrrolnitrin biosynthetic operon 32 . Subsequent pyrrolnitrin quantification via HPLC analysis revealed that wild type PA23 produced an average of 3.48 (±0.45) μg of pyrrolnitrin, whereas in the ptrA mutant, no pyrrolnitrin was detected. However, when ptrA was expressed in trans in PA23-443, pyrrolnitrin production was restored to wild-type levels (3.90 ± 0.20 μg). Significant downregulation of pyrrolnitrin expression may not have been identified through iTRAQ analysis as cell samples were taken at the onset of stationary phase. To obtain enough pyrrolnitrin for quantification, cell culture extracts are routinely performed after five days of growth 5 . Thus, there may have been differences in protein expression in late stationary phase that were not detected in our iTRAQ analysis. As pyrrolnitrin has previously been reported as essential for PA23 biocontrol 5 , the lack of pyrrolnitrin production by the ptrA mutant is likely a major contributor to the loss of antifungal activity.

The bacterial strains and plasmids used in this study are listed in Table 5. Escherichia coli strains were cultured at 37°C on Lennox Luria Bertani (LB) agar (Difco Laboratories, Detroit, Michigan). P. chlororaphis PA23 and its derivatives were cultured at 28°C on LB agar or M9 minimal media supplemented with 1 mM MgSO₄ and 0.2% glucose. For antifungal assays, bacteria were grown on potato dextrose agar (PDA; Difco). As required, media were supplemented with the following antibiotics: tetracycline (Tc; 15 μg/mL), gentamicin (Gm; 15 μg/mL), ampicillin (Amp; 100 μg/mL) for E. coli, and rifampicin (Rif; 25 μg/mL), Tc (15 or 100 μg/mL), Gm (25 μg/mL), piperacillin (30 or 500 μg/mL) for P. chlororaphis. All antibiotics were obtained from Research Products International Corp. (Mt. Prospect, Illinois).

Polymerase Chain Reaction (PCR) was performed under standard conditions as suggested by Invitrogen Life Technologies data sheets supplied with their Taq polymerase.

Cloning, purification, electrophoresis, and other manipulations of nucleic acid fragments and constructs were performed using standard techniques 36 . To clone the PA23 ptrA gene, oligonucleotide primers ptrA-F and ptrA-R were used to amplify a 2.2-kb product which was cloned into vector pCR2.1-TOPO following manufacturer’s instructions. The 2.2-kb ptrA insert was then excised with XbaI and BamHI and cloned into the same sites of pUCP22, generating pUCP22-ptrA.

Bacterial conjugations were performed to introduce Tn5-OT182 into P. chlororaphis PA23 by biparental mating following the method of Lewenza et al., 37 . For each mating, 5–10 Tc^R colonies were screened by PCR to ensure that transconjugants contained a Tn5 insertion using TNP5-FORWARD and TNP5-REVERSE primers. To determine the site of Tn5-OT182 insertion, rescue cloning was performed following previously described methods 37 .

Plasmids isolated from Tc^R XhoI clones were sent for sequencing using oligonucleotide primer Tn5-ON82, which anneals to the 5′ end of Tn5-OT182. BamHI or ClaI rescue plasmids were sequenced using primer Tn5-OT182 right, which anneals to the 3′ end of the transposon. All sequencing was performed at the University of Calgary Core DNA Services facility. Sequences were analyzed using BLASTn and BLASTx databases (http://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=BlastHome). The GenBank accession number for the P. chlororaphis PA23 ptrA gene sequence is EF054873.

Radial diffusion assays to assess fungal inhibition against S. sclerotiorum in vitro were performed with wild-type PA23, mutant PA23-443 and PA23-443 harboring the ptrA gene in trans according to previously described methods 4 . Five replicates were analyzed for each strain and assays were repeated three times.

Wild-type PA23 and mutant PA23-443 cells were grown as duplicate samples. At the point when cultures were just entering stationary phase (OD₆₀₀ = 1.2), they were centrifuged at 10,000 × g for 10 minutes at 4°C, and pellets were washed three times in PBS buffer and frozen at −80°C. Further sample preparation and iTRAQ labelling was carried out at the Manitoba Centre for Proteomics and Systems Biology. Briefly, 100 μg protein samples were mixed with 100 mM ammonium bicarbonate, reduced with 10 mM dithiothreitol (DTT) and incubated at 56°C for 40 min. Samples were then alkylated with 50 mM iodoacetamide (IAA) for 30 min at room temperature in the dark. Addition of 17 mM DTT was used to quench excess IAA, and proteins were digested with sequencing-grade trypsin (Promega, Madison, WI, USA) overnight. Dried samples were then desalted with 0.1% trifluoroacetic acid and subjected to two-dimensional high-performance liquid chromatography (2D-HPLC)-mass spectrometry (MS) according to previously described methods 38 .

2D-HPLC-MS/MS spectra data from three independent runs were analyzed using ProteinPilot (v2.0.1, Applied Biosystems/MDS Sciex, Concord, ON, Canada) which employs the Paragon™ algorithm. Searches were performed against the P. chlororaphis strain gp72 reference genome. Reporter ion iTRAQ tags were labelled as follows: tags 114 and 115 to replicates of wild-type PA23 grown to early stationary phase, and tags 116 and 117 to replicates of mutant PA23-443 grown to early stationary phase. Results were reported as Z-scores, the log2 of the ratio among replicates (Z0 = tag₁₁₆/tag₁₁₄; Z1 = tag₁₁₇/tag₁₁₅; Z2 = tag₁₁₅/tag₁₁₄; Z3 = tag₁₁₇/tag₁₁₆). Peptide Z-scores values were histogrammed (Z0, Z1) to determine the overall population distribution. Further statistical analysis was performed according to the methods outlined in Rydzak et al., 38 . Briefly, V_diff scores were assigned to allow the determination of statistical significance of protein expression ratios between both the wild-type and mutant samples while also taking into account the variation between biological replicates. Plotted Z-scores were transformed into vector values, allowing comparison between points (Z0,Z1) and (Z2,Z3). Differences between magnitudes of the vector values from the origin to points (Z0,Z1) and (Z2,Z3) were adjusted to the widths of the peptide population distributions. Direction of the vector values (+or -) were assigned based on the angle subtended by the vector value from the origin to point (Z0,Z1). A V_diff value greater than or equal to +1.65 and less than or equal to −1.65 corresponds to proteins expressed in the upper or lower 10% of the population distribution 38 . Functional classification of proteins was carried out using the Integrated Microbial Genomes (IMG) database (http://img.jgi.doe.gov/cgi-bin/w/main.cgi) against the P. chlororaphis strain gp72 genome.

Cultures of wild-type PA23 and mutant PA23-443 were inoculated at a starting optical density (OD) ₆₀₀ of 0.01 and grown in M9 minimal media (1 mM MgSO₄; 0.2% glucose). OD₆₀₀ readings were taken at 1 hour, 5 hours and 9 hours, followed by readings every 2 hours until 27 hours of growth. Triplicate samples were analyzed.

PA23 and derivative strains were assayed for chitinase production during early stationary and late stationary phases following the methods outlined by Wirth and Wolf 39 . Briefly, cultures were grown to the desired growth phase in M9 minimal media (1 mM MgSO₄; 0.2% glucose) and 250 μL aliquots of each of cell-free supernatant, 0.1 M NaOAc, pH 5.2 and carboxymethyl-chitin-Remazol brilliant violet aqueous solution (Loewe Biochemica, Germany) were incubated for 1 hour at 37°C. The reaction was stopped by the addition of 250 μL 1 M HCL. Reaction mixtures were cooled on ice for 10 min and spun at 20,000 × g for 10 min, and the absorbances at 550 nm were recorded. Each experiment was performed in triplicate.

Flagellar (swimming) was monitored according to Poritsanos et al., 4 . Strains were grown overnight in M9 minimal media (1 mM MgSO₄; 0.2% glucose) and 5 μL was inoculated into the center of 0.3% M9 agar plates. Four replicates were analyzed and the experiment repeated three times.

Overnight cultures in M9 minimal media (1 mM MgSO₄; 0.2% glucose) were subjected to phenazine extraction and quantification by UV-visible light spectroscopy at 367 nm and 490 nm for PCA and 2-OH-PHZ, respectively 5 . Phenazine analysis was performed in triplicate.

Overnight cultures grown in M9 minimal media (1 mM MgSO₄; 0.2% glucose) were spotted onto CAS media according to the methods outlined in Schwyn and Neilands 40 to analyze siderophore production.

Production of the antibiotic PRN was quanitified according to the methods outlined in 5 . Briefly, 20 mL cultures of PA23 and its derivatives were grown for 5 days in M9 minimal media and PRN was extracted with an equal volume of ethyl acetate. Before extraction, toluene (5 mL) was added to each sample as an internal control. Toluene and PRN UV absorption maxima were recorded at 225 nm with a Varian 335 diode array detector. PRN peaks were detected at 4.7 mins. Samples were analyzed in duplicate.

All statistical analysis was performed using unpaired Students’s t test.

The data sets supporting the results of this article are included within the article.