Characterization of the Biocontrol Activity of Pseudomonas fluorescens Strain X Reveals Novel Genes Regulated by Glucose

Pseudomonas fluorescens strain X, a bacterial isolate from the rhizosphere of bean seedlings, has the ability to suppress damping-off caused by the oomycete Pythium ultimum. To determine the genes controlling the biocontrol activity of strain X, transposon mutagenesis, sequencing and complementation was performed. Results indicate that, biocontrol ability of this isolate is attributed to gcd gene encoding glucose dehydrogenase, genes encoding its co-enzyme pyrroloquinoline quinone (PQQ), and two genes (sup5 and sup6) which seem to be organized in a putative operon. This operon (named supX) consists of five genes, one of which encodes a non-ribosomal peptide synthase. A unique binding site for a GntR-type transcriptional factor is localized upstream of the supX putative operon. Synteny comparison of the genes in supX revealed that they are common in the genus Pseudomonas, but with a low degree of similarity. supX shows high similarity only to the mangotoxin operon of Ps. syringae pv. syringae UMAF0158. Quantitative real-time PCR analysis indicated that transcription of supX is strongly reduced in the gcd and PQQ-minus mutants of Ps. fluorescens strain X. On the contrary, transcription of supX in the wild type is enhanced by glucose and transcription levels that appear to be higher during the stationary phase. Gcd, which uses PQQ as a cofactor, catalyses the oxidation of glucose to gluconic acid, which controls the activity of the GntR family of transcriptional factors. The genes in the supX putative operon have not been implicated before in the biocontrol of plant pathogens by pseudomonads. They are involved in the biosynthesis of an antimicrobial compound by Ps. fluorescens strain X and their transcription is controlled by glucose, possibly through the activity of a GntR-type transcriptional factor binding upstream of this putative operon.


Introduction
Many bacterial strains from the genus Pseudomonas are capable of suppressing a range of plant diseases caused by soil-borne plant pathogenic fungi, due to their ability to biosynthesize antimicrobial metabolites. Antibiotics, cyclic lipopeptides (CLPs) with antimicrobial activity, siderophores and hydrogen cyanide are the main secondary metabolites to which the biological control is attributed [1]. Regulation of the biosynthesis of these antimicrobial metabolites has been extensively studied. A wide range of environmental as well as endogenous factors control the transcription of several genes involved in the biosynthesis of antimicrobial metabolites [2,3].
Glucose is one of the environmental factors which affect the biosynthesis of secondary metabolites such as oomycin A [4], kanosamine [5], DAPG [6], pyoluteorin and pyochelin [7], prodigiosin [8], pyrrolnitrin and phenazine [9]. Recently, it has been proposed that it is gluconic acid, not glucose, that regulates the production of antimicrobial metabolites [10]. Moreover, gluconic acid has been suggested as having a direct inhibitory effect on phytopathogenic fungi sensitive to lower concentrations of the acid [11].
Gluconic acid derives from glucose by an oxidative reaction in the periplasmic space, thus affecting the pH and the availability of soluble phosphates in all glucose-containing media [12]. The oxidation of glucose to gluconic acid is catalysed by membranebound quinoprotein glucose dehydrogenases (Gcd) that are involved either in biocontrol of plant pathogens [10,13] or in pathogenicity of bacteria in mammals [14]. Among various quinoprotein dehydrogenase enzymes in bacteria, Gcd uses pyrroloquinoline quinone (PQQ) as an essential cofactor [15]. Genes involved in the biosynthesis of PQQ are organised in a putative gene cluster that is expressed as an operon and its transcription is regulated by various carbon sources [16]. Insertional inactivation of the PQQ biosynthetic genes has proven their significance in biocontrol [17,18,19,13] and in plant growth promotion [20].
A rise of interest in CLP metabolites has been recently noted, due to their biosurfactant, antimicrobial and phytototoxic activity [21]. The biosynthetic model of various CLPs has been fully elucidated, yet regulation for most of them is still under study [22]. Environmental factors such as pH, temperature, carbon and nitrogen sources [23,24], plant signal molecules [25] and protozoan predators [26] affect the production of CLP metabo-lites. Also, endogenous factors like the two component regulatory system GacA/GacS [27,28,29], quorum sensing [30,31,32,33], sigma factors and Hsp [34] control the expression of several CLP biosynthetic genes.
Pseudomonas fluorescens strain X is a bacterial biocontrol agent able to suppress cucumber and sugar beet damping-off caused by Pythium ultimum. Suppression of damping-off by Ps. fluorescens strain X has been proved to be more effective over other Pseudomonas and Bacillus strains [35]. Nevertheless, its biocontrol ability has not yet, been linked to any known antimicrobial metabolites.
The aim of the present study was to elucidate the mechanism by which Ps. fluorescens strain X suppresses damping-off. In order to achieve this, transposon mutants of strain X were created (designated sup 2 ) which were impaired in their ability to suppress the in vitro radial growth of P. ultimum. Sup 2 mutants were found to carry the transposon integration in six different genes. Results indicate that genes sup5 and sup6 have key role in the biosynthesis of a antimicrobial metabolite, while genes sup1, sup2, sup3 and sup4 play a secondary role, indirectly controlling the transcription of the first two.

Biochemical Characterization of Ps. fluorescens Strain X and Derived Mutants
Nine mutants (k36, W139, R48, B161, B91, A150, r26, d40 and r93) were isolated out of a mutant library of 12000 derived from random Tn5 insertion mutagenesis of Ps. fluorescens strain X. All mutants had lost the ability to inhibit P. ultimum radial growth on Potato Dextrose Agar (PDA) and retained the same growth rate with the wild type (data not shown). When the wild type was incubated on minimal medium (M9) supplemented with glucose (2% w/v) as the sole carbon source, it acidified the substrate, lowering the pH from 6 to 5. On the contrary, mutants k36, W139, R48, B139, B91, A150 and r26 increased the pH of the medium from 6 to 8, while d40 and r93 did not alter the pH (Table 1). Furthermore, cell-free filtrates of the wild type strain and the complemented mutants (described in a subsequent section) after growth in either Potato Dextrose Broth (PDB) or Luria Bertani Broth (LB), were treated overnight with proteinase K and pronase and were tested for inhibition of P. ultimum radial growth (Table 1). No alterations in the antimicrobial activity for any of the treated filtrates were observed (Table 1). These results indicate that the antimicrobial activity might be attributed to a molecule either of non-peptidic nature or containing non-proteinaceous aminoacids.

Cloning and Sequencing of the Suppression Genes
Southern hybridization confirmed that each of the mutants used in this study contained a single transposon insertion in the chromosome. Mutants containing more than one insertion were not analysed further (data not shown). Subsequent cloning and sequencing of the chromosomal regions flanking the transposon insertion in the selected mutants revealed that the transposon was localized in three different genomic regions. In mutants k36, W139 and R48 the transposon integration was in a 2,418 bp gene, designated sup1. The transposon insertion in mutants B163, B91 was in sup2, (2,424 bp). In mutants A150 and r26 the transposon was in two genes, sup3 and sup4, (276 bp and 996 bp respectively). Genes sup2, sup3 and sup4 were neighbouring in a genomic region of 7119 bp (Fig. 1). Finally, mutants d40 and r93 had a transposon integration in genes sup5 (777 bp) and sup6 (348 bp) respectively. These genes were localized in a third genomic region of 6,782 bp (Fig. 2).

Identification and Characterization of the Chromosome Regions Containing the Suppression Genes
The deduced product of sup1 (805 amino acids; 85.9 kDa) exhibits the highest similarity (95% identical) to the putative PQQdependent glucose dehydrogenase (Gcd) of the fully sequenced biocontrol strain Ps. fluorescens SBW25 encoded by PFLU1086 (GenBank accession no. YP_002870745.1). The product of sup1 was 70% identical to the Gcd from the well-characterized strain Ps. fluorescens CHA0 (ACN53518.1), encoded by locus FJ694890 (10), as well as to the Gcd from the biocontrol strain Ps. fluorescens Pf-5, encoded by PFL_4916 (AAY94145.1).
The proteins encoded by the neighbouring genes sup2, sup3 and sup4, show high similarity to PqqF, PqqD and PqqE respectively, involved in the biosynthesis of pyrroloquinoline quinone (PQQ). Sup2 (807 amino acids; 87.9 kDa) exhibits relatively low similarity (61% identical) to the putative peptidase PqqF of Ps. fluorescens CHA0 (CAA60730.1) [17]. The putative product of sup3 (91 amino acids; 10.3 kDa) has the highest similarity (96%) with PqqD of Ps. fluorescens SBW25 (YP_002875096.1). The deduced product of sup4 (331 amino acids; 37.4 kDa) showed 99% identity to the PqqE of Ps. fluorescens SBW25 (YP_002875097.1). The genes sup5 and sup6 are located in a genomic region similar to the Ps. syringae pv.syringae UMAF0158 mangotoxin biosynthesis region. The deduced product of gene sup5 (258 amino acids; 28.7 kDa) is a protein of unknown function with highest similarity (73% identical) to the protein encoded by orf3 from Ps. syringae pv. syringae UMAF0158 (ABG00044.1). Interestingly, the deduced product of sup5 is less similar to the proteins encoded from other Ps. fluorescens strains. Sup5 is 69% similar to the protein encoded by Pfl01_0128 from Ps. fluorescens Pf0-1 (YP_345861.1) and 53% similar to the protein encoded by PFLU_0121 from Ps. fluorescens SBW25 (YP_002869817.1). Furthermore, sup6 encodes a protein (115 amino acids; 13.3 kDa) that belongs to the cupin superfamily. The cupin superfamily, whose name comes from a conserved bbarrel fold structure (deriving from the Latin term 'cupa' that stands for a small barrel), is a functionally diverse family that comprises enzymatic and non-enzymatic members [36]. Among all Pseudomonas species, the only strain with a gene homologous to sup6 is the opportunistic human pathogen Ps. aeruginosa PA7. Sup6 is 58% identical to the protein encoded by PSPA7_2111 (YP_001347484.1), which bears two cupin domains (data not shown), thus structurally categorised in the bicupins group of the cupin protein superfamily.

Complementation of Sup 2 Mutants
Mutants k36, R48 and W139 were complemented by a 2771 bp fragment amplified with the primers set gcd1-gcd3, which included the gene sup1 and a 356 bp region upstream. By promoter prediction analysis of the 356 bp region upstream, a unique promoter site was found (P = 0.92). Complemented mutants regained the ability to inhibit radial growth of P. ultimum on PDA and to acidify the medium like the wild type.
In order to investigate if sup2, sup3 and sup4 were separately involved in biological control and acidification of PDA, several DNA fragments containing parts of the PQQ genomic locus were tested for their ability to complement mutants B91, B163, r26 and A150. Among these, the only one which restored the suppressive phenotype in all four mutants was a 6383 bp fragment which included five genes (sup2, orf1, orf2, orf3, sup3 and sup4) and a 504 bp region upstream of sup2 (Fig. 1). The 6383 bp region was PCR-amplified with primers FOR2 and pqqE2. Smaller fragments of this area and individual genes tested could not restore the sup phenotype. Complemented mutants had fully restored the ability Table 1. Characteristics of sup 2 mutants and complementation analysis.

Strain Acidification a
In vitro inhibition of P. ultimum radial growth by PDB filtrates of bacteria In vitro inhibition of P. ultimum radial growth by LB filtrates of bacteria Antimicrobial activity of PDB filtrates after treatment with proteinase K and pronase b X Acidification was observed on solid minimal medium M9 supplemented with 2% w/v glucose, as described before [14]. b Enzyme treatment was performed for filtrates from incubation in PDB, as described before [47]. doi:10.1371/journal.pone.0061808.t001 Figure 1. Complementation analysis of the PQQ biosynthesis region in Ps. fluorescens strain X. PCR fragments of this region (1-5), with different sets of genes from Ps. fluorescens X, and their ability to complement sup 2 mutants. Ability to complement is noted with plus (+) or minus (2). The direction of the plasposon Tn5-RL27 insertion in the derivative mutants B91, B163, A150 and r26 is indicated with a flag beneath the sequence. The predicted site for the unique putative promoter is also marked. doi:10.1371/journal.pone.0061808.g001 to inhibit the radial growth of P. ultimum on PDA and to acidify the medium. Complementation of sup 2 mutants d40 and r93, which bear a transposon integration in genes sup5 and sup6 respectively, could only be achieved by a 6,496 bp region which includes five genes (sup5, sup6, orf8, orf9 and orf7) and the 423 bp region upstream of sup5. Smaller DNA fragments containing only parts of this 6,496 bp region did not complement the sup-mutations in mutants d40 and r93 (Fig. 2). The region that complemented the sup phenotype was PCR-amplified from Ps. fluorescens strain X genomic DNA with the primer set supFor-supRev.

Analysis of the sup5 and sup6 Genomic Locus
The genomic locus containing genes sup5 and sup6 has not been previously implicated in the biocontrol activity of any bacterial antagonists of plant pathogens. Therefore, we studied this locus in more detail. By primer walking upstream and downstream of the transposon insertion, a total of 6,782 bp was cloned and sequenced from the region flanking these two genes in strain X. Figure 2. Arrangement of the genes in the genomic locus of sup5 and sup6, compared to other Pseudomonas strains, and complementation analysis of the region. The lines beneath the genomic of Ps. fluorescens X represent regions of this locus that were PCRamplified, cloned into pBBR1MCS5 and tested for complementation. Ability to complement is noted with plus (+) or minus (2). Putative ORFs are indicated by thick coloured arrows on a line. Genes that might be organised in a putative operon are enclosed by a grey frame. The direction of the plasposon Tn5-RL27 insertion in mutants d40 and r93 is indicated with a black arrow beneath the sequence (c). Predicted sites for the unique putative promoter and operator are also marked ( ). Size, genomic location and locus tag of the different ORFs sequenced in Ps. aeruginosa PA7, Ps. fluorescens Pf0-1, Ps. fluorescens SBW25, Ps. entomophila L48, Ps. syringae pv. syringae B728a,Ps. syringae pv. tomato DC3000 and Ps. syringae pv. syringae UMAF0158 are indicated. doi:10.1371/journal.pone.0061808.g002 Several ORFs (open reading frames) were identified (Fig. 2), including a non-ribosomal peptide synthase (NRPS), orf8; a polyketide cyclase/dehydrase, orf9; a protein of unknown function, orf7; and a partial sequence of a EmrB/QacA family drug resistance transporter, orf6 (data not shown). The organization of these genes in strain X is similar to the homologous loci of all fully sequenced Pseudomonas strains (Fig. 2) and shows high similarity with the locus attributed to mangotoxin biosynthesis by Ps. syringae pv. syringae UMAF0158 [37]. A detailed analysis of the 1,158 amino acids from the deduced product of the orf8 gene determined to contain a sole module [38,39] consisting of an adenylation, a thiolation, a condensation and a reductase domain (no other domain was detected). The domains and the conserved core motifs of the NRPS encoded by orf8 had minimal differences with those from MgoA of the strain UMAF0158 (data not shown).
Prediction analysis of sites for putative promoters resulted in the presence of a unique promoter upstream of sup5 (P = 0.96). The uniqueness of this putative promoter implies that the genes downstream might be transcribed together. Also, data resulting from the synteny search of this region revealed the existence of an operon consisting of four to five genes (Fig. 2), depending on the strain.
In this genomic locus and in the region upstream of the putative four-gene operon, most Pseudomonas strains examined in this study (except Ps. aeruginosa PA7) have a gene encoding a putative transcription factor of the GntR family. Since sequencing of the Ps. fluorescens strain X chromosome has not been completed for this area, it is not known whether this transcription factor exists in this strain. The potential existence of a gene encoding this transcription factor in strain X may be an indication that this factor may regulate the transcription of supX in Ps. fluorescens strain X. Analysis of the region between the site of the unique putative promoter and the gene sup5, revealed a putative operator site belonging to transcriptional factors of the FadR subfamily of GntR transcriptional regulators, with the sequence GAGTGGTCAGCGT-TAAC. The consensus sequence for FadR operators is AACTGGTCNGACCAGTT [40]. The existence of a putative operator site in the particular region, suggests that a factor of the FadR subfamily might control the transcription of the genes downstream. The region containing the site of the putative promoter and the putative operator site of the transcriptional factor extends 423 bp upstream of the putative operon supX (Fig. 2).
Expression of sup5, sup6 and orf8 in the Wild Type and in Mutants r26 and k36 Quantitative gene expression analysis was performed to study the expression levels of genes sup5, sup6 and orf8 during growth of the wild type strain X and two selected mutants (k36, mutation in gene sup1 for gcd and r26, mutation in gene sup4 for PQQ) in two different liquid media (LB and PDB) during the exponential and stationary phases of growth. The media were selected because strain X inhibits P. ultimum radial growth on PDA but not on LA. We decided to study the expression of genes sup5 and sup6 because they are involved in the biocontrol activity of strain X, as shown by mutagenesis and complementation. We also decided to study the expression of orf8, a gene in the same putative operon, because it encodes a non-ribosomal peptide synthase, a class of genes responsible for the biosynthesis of antimicrobial compounds which suppress plant pathogens.
RNA from cells of the wild type and mutants was isolated from an identical stage of their growth, based on standard growth curves and a specific optical density at mid-exponential (OD 600 = 0.6) and stationary (OD 600 = 1) phase.
During the mid-exponential phase, transcription levels of sup5, sup6 and orf8 in the wild type were low and very similar in both media (Fig. 3E, 3F). At the stationary phase, transcription of supX genes was low when the wild type strain was grown in LB, but much higher in PDB (Fig. 3A, 3B), in correlation with the production of an antifungal compound in PDA and the resulting reduction of radial growth. These genes exhibit the highest expression levels during the stationary phase in strain X, much like most bacterial genes responsible for the biosynthesis of secondary metabolites known to suppress plant pathogens.
Expression of sup5, sup6 and orf8 in the two mutants compared to the wild type was significantly decreased during the stationary phase, when bacteria were grown in PDB (Fig. 3C, 3D). Mutants r26 and k36 exhibited similar expression levels of genes sup5, sup6 and orf8 during stationary and mid-exponential phase in both nutrient media (data not shown). These results indicate that the insertional mutations in genes gcd and pqqE, which consequently result in an inactive glucose dehydrogenase, decreased the transcription of genes sup5, sup6 and orf8 which are responsible for the biocontrol activity of Ps. fluorescens strain X.

Discussion
To elucidate the mechanism of antagonism towards the oomycete P. ultimum, nine mutants of the bacterial antagonist Ps. fluorescens strain X deficient in the in vitro inhibition of P. ultimum radial growth were isolated after mutagenesis with the miniplasposon Tn5-RL27. Sequencing of the DNA areas flanking the transposon insertion and complementation of the mutated genes showed that three separate genomic regions are involved in the biological control traits of strain X. The biocontrol ability of Ps. fluorescens strain X depends on glucose dehydrogenase, its co-factor PQQ, and the proteins encoded by two additional genes (sup5 and sup6). A gene for a non-ribosomal peptide synthase (NPRS) which is located together with sup5 and sup6 in the same putative operon may also be responsible for the biocontrol activity of strain X, since NRPSs are known to synthesize antimicrobial compounds involved in the suppression of plant pathogens in planta.
In previous studies it has been reported that biological control traits of bacterial antagonists, such as the production of antifungal compounds, are intimately related to Gcd and its cofactor PQQ, but not in the same way for every strain. A Tn5 insertion mutant of Ps. fluorescens CHA0, in which the transposon had integrated in one of the PQQ biosynthetic genes (pqqF), showed increased production of pyoluteorin (PLT) and decreased production of 2,4diacetylphloroglucinol (DAPG) [17]. The authors of this paper concluded that the increase in the PLT production was caused by the Tn5 integration, while the loss of DAPG production was attributed to a second spontaneous mutation that led to subsequent loss of the ability to suppress black root rot in tobacco plants and take-all in wheat. Moreover, a mutant of CHA0 carrying an in-frame deletion in the gene encoding Gcd had lost the ability to produce organic acids and to solubilise inorganic phosphate, but exhibited increased production of the antifungal metabolites DAPG and PLT. Consequently, the Dgcd mutant was more effective in the biological control of G. graminis var. tritici [10]. A similar study was performed to characterize the antagonistic activity of Rahnella aquatilis HX2. Insertional mutations in gcd and pqqE as well as the Dgcd in-frame deletion mutant were impaired in acidification of the medium and in the production of an antibacterial substance, resulting in reduced biological control of grapevine crown gall [13]. This finding shows a different effect of the same mutations in this strain. In the biocontrol strain Serratia marcescens W1 the genomic region of the PQQ biosynthetic genes was functionally expressed in E. coli. The transformed strain gained the ability to inhibit the in vitro radial growth of Magnaporthe grisea and Cercospora citrullina [18].
In Ps. fluorescens strain X, mutational inactivation of genes encoding PQQ biosynthesis proteins or Gcd resulted in loss of the in vitro inhibition of P. ultimum. This result is similar to the report on R. aquatilis HX2 [13] but different from the findings on Ps. fluorescens CHA0 [17,10]. In accordance with the results from both HX2 and CHA0, are the results presented for the Ps. fluorescens strain X mutants impaired in the acidification of PDA, a glucosecontaining medium ( Table 1). Loss of the acidification ability in PQQ or Gcd mutants has been attributed to lack of oxidation of glucose to gluconic acid, a reaction catalysed by membrane-bound PQQ-dependant Gcd [10]. Although it has been suggested that gluconic acid is the key molecule for the biocontrol activity of strains AN5 and CHA0 [11,10], the exact mechanism of biocontrol remains undefined in these strains.
The study of the mutants Ps. fluorescens d40 and r93 was intriguing due to the similarity of the genes disrupted by the Tn5 insertion (sup6 and sup5) and the flanking region with the homologous genomic locus in Ps. syringae pv. syringae UMAF0158, which is involved in the biosynthesis of mangotoxin [37]. The putative product of sup5 was found to be similar to ABG00044.1, the product of orf3 in the strain UMAF0158, while sup6 was absent from the region involved in mangotoxin biosynthesis in the strain UMAF0158 [37]. Among all fully sequenced Pseudomonas strains, the only strain with a gene homologous to sup6 is the opportunistic human pathogen Ps. aeruginosa PA7. However, the putative protein encoded by sup6 exhibits relatively low aminoacid similarity (58%) with protein YP_001347484.1 from strain PA7, whose function is unknown.
Interesting findings were obtained from the in silico analysis of the genomic region in which sup5 and sup6 are localized. Synteny search concluded that these genes seem to be organized in a putative operon, designated supX, together with genes orf7, orf8 (which encodes a NRPS) and orf9. The homologous genes in strain UMAF0158 have also been suggested to be organised in an operon [37]. From the synteny search, we also located a gene encoding a transcriptional regulator of the GntR family, upstream of this putative operon in most Pseudomonas strains. We do not yet have sequence data of the region upstream of supX in Ps. fluorescens strain X, therefore we cannot confirm the presence of a GntR transcriptional regulator in this strain. However, we found a putative FadR operator site in this region, which is recognised by a transcriptional factor of the GntR family. Future work will be to sequence more of the region upstream of supX and to determine the existence of a GntR-type transcriptional regulator, its operator site and the type of regulation on the transcription of the putative operon supX.
The family of GntR transcriptional regulators is divided into four major subfamilies (FadR, HutC, MocR, and YtrA). DNAbinding domains, operator sequences [41,40], as well as functions and types of regulation [41,42] for each subfamily have been previously described. It has been demonstrated that glucose and gluconic acid hold a key role in the inactivation of regulators belonging to this superfamily [43]. Moreover, glucose inactivates agl3R, a GntR transcriptional repressor in Streptomyces coelicolor, resulting in the expression of genes involved in the ABC excretion system and antibiotic biosynthesis [44]. On the contrary, in Serratia sp. ATCC 39006 gluconic acid inactivates the PigT transcriptional regulator, of the GntR superfamily, preventing the expression of the prodigiosin biosynthetic genes [8].
The existence of a putative FadR operator site upstream of the putative operon supX is a possible indication that transcription of supX might also be regulated by a factor of the GntR family. Thus, trascriptional regulation of supX might be subjected to regulation by glucose, possibly through gluconic acid, the product of the catalytic action of Gcd on glucose. In order to investigate whether this hypothesis is valid, we tested the expression of three genes of the putative operon supX in the wild type and in two mutants (r26 and k36) with deficient Gcd activity. Transcription of supX correlated with the in vitro inhibition data of P. ultimum by strain X, since elevated transcription levels of sup5, sup6 and orf8 in the wild type were observed in PDB and not in LB. (Table 1). Expression analysis of sup5, sup6 and orf8 in mutants r26 and k36 (impaired in Gcd activity) supports the role of glucose on the antagonistic properties of strain X. Transcript levels of all three genes were significantly decreased in these mutants. The decrease was evident during the mid-exponential and stationary phase in both nutrient media tested. Finally, the expression of supX was elevated during stationary phase, a typical condition during the biosynthesis of secondary metabolites. Together with the mutational and complementation analysis of supX, this demonstrates that the putative operon supX is responsible for the antagonistic activity of Ps. fluorescens strain X through the biosynthesis of an antimicrobial secondary metabolite (the chemical characterization of this metabolite is in progress).
The exact role of sup5 and sup6 in the biosynthesis of the antimicrobial metabolite is currently under investigation. Our primary hypothesis suggests that sup6, contributes either to the final folding of the metabolite or in the transcriptional regulation of the genes in the operon supX, due to its high similarity to the functionally diverse cupin superfamily [36]. The product of sup5, which exhibits similarity to a heme-oxygenase-like protein, may catalyse the oxidation of the peptide chain synthesized by the NRPS encoded by orf8. Since the putative operon supX includes a NRPS, the novel metabolite might be of peptidic nature. The results of the protease treatment of the filtrates (Table 1) do not necessarily contradict this hypothesis. Resistance of a peptidic antimicrobial compound to protease may be due to the presence of non-proteinaceous amino acids and D-amino acids, or the lack of optimum conditions for protease and proteinase cleavage (pH, final concentration of the enzyme or the antimicrobial molecule, denaturing buffer).
The results of this study demonstrated that the putative operon supX is related for the first time with the biological control of a soilborne phytopathogen by a Ps. fluorescens strain. Genes with some homology to those in supX exist in other pseudomonads as well, including known strains with biocontrol activity, as well as plant and human pathogens. However, these genes have never been related to the biocontrol of plant pathogens before, so their role in strain X is a novel finding. Ongoing and future work will focus on the characterization of the antimicrobial compound(s) produced by strain X, the regulation of the biosynthesis of this compound by glucose and the exact role of the genes in supX in the biosynthesis of this antimicrobial metabolite.

Bacterial Strains, Plasmids, and Mutagenesis
All bacterial strains and plasmids used in this study are listed in Table 2. Ps. fluorescens strain X was grown at 28uC, and E. coli strains was grown at 37uC. All strains were routinely grown on LB agar, unless otherwise stated. A spontaneous derivative of Ps. fluorescens strain X resistant to rifampicin was used for transposon mutagenesis. This derivative had identical biochemical, growth and biocontrol characteristics with the wild type (data not shown). Transposon mutagenesis of Ps. fluorescens strain Xrif was carried out by triparental mating using the mini plasposon Tn-RL27, as described previously [45] and E. coli HB101 carrying plasmid pRK2013 [46] as the helper. This mini plasposon carries the origin of replication from plasmid R6K (oriR6K) to allow cloning of transposon insertion sites. Derivative strains were isolated after 48 h at 28uC on King's B medium under selection of kanamycin (100 mg/ml) and rifampicin (40 mg/ml). In total, a library of Ps. fluorescens strain Xrif mutants containing over 12000 strains was constructed.

Biochemical Characterization of Ps. fluorescens Strain X Mutants
The growth of Ps. fluorescens strain Xrif mutants and their ability to suppress fungal growth was tested with dual cultures on PDA at 23uC. After a first screening, potential mutants were further tested in three replicate plates together with the wild-type to confirm the loss of P. ultimum growth inhibition. Loss of suppression of fungal growth was assessed 2 days later.
Filtrates of strain X and selected mutants grown in PDB and LB 48 h at 23uC were tested for growth inhibition of P. ultimum as previously described [35].
Enzymatic treatment by pronase and proteinase K, of the filtrates was performed according to Arrebola et al. [47]. Observations of the culture pH were performed using different pH indicators [14].

DNA Manipulation and Sequencing
DNA digestion, ligation reactions, and transformation of E. coli were performed according to standard protocols [48]. Genomic DNA isolation was performed using the GenElute TM Bacterial Genomic DNA Kit (Sigma-Aldrich Co. LLC., Germany). Plasmid mini-preps were done using the Qiaprep spin miniprep and midiprep kit (Qiagen GmbH, Düsseldorf, Germany). For sequence analysis of the regions flanking the miniplasposon insertions, published primers were used [45]. Automated DNA sequencing of rescue plasmids was carried out by Macrogen Inc. (Korea) and VBC-Biotech Service GmbH (Austria).

Isolation and Characterization of Genomic Loci Carrying a Transposon Insertion
Plasmid rescue was performed to clone the genomic locus of the insertion in every mutant. Genomic DNA was digested with BamHI and subsequently treated with T4 DNA ligase. The ligation mix was transformed into E. coli DH5a/lpir, where circularized fragments containing the transposon replicate as plasmids, allowing the selection only of chromosomal fragments containing the transposon [45]. Southern blot analysis was performed (DIG High Primer DNA labelling and detection starter kit TM , Roche Applied Science, Germany) in BamHI-digested genomic DNA from the selected mutants using the Tn-RL27 as probe to determine the uniqueness of the insertion in the genome and HB101 hsdR, recA,proA,leu-0,ara-l4 gaiK2, lacYl, xyl-5, mtl-1 str-2, thi-1, supE44 [48] determine the size of the genomic fragments containing the transposon element.

Complementation of Mutants
Complementation of the wild type phenotype was accomplished for all isolated strains. Specific primers designed for PCR amplification of the genes and loci in which the transposon had integrated, are listed in Table 2. Amplification of genomic loci from genomic DNA of the wild type, in which the insertion was localized in the sup 2 mutants, was possible using the Expand Long Range dNTPack (Roche Diagnostics Gmbh, Germany). The broad host range vector pBBR1MCS5, carrying the gene for gentamycin resistance, was used for cloning single genes or genomic loci [49]. The plasmids that were constructed were transformed into E. coli DH10b and were subsequently inserted into the respective mutants by triparental conjugation.

Transcriptional Analysis
For the transcriptional analyses, RNA was isolated from bacterial cells according to standard procedures [48], followed by DNase I (Promega) treatment. Total RNA was extracted from Ps. fluorescens strain Xrif and two mutant strains k36 and r26 grown in PDB and LB medium, at midlog and stationary phase. The concentration and purity of the RNA samples were measured by using an ND100 spectrophotometer (Nanodrop Technologies, USA) according to the manufacturer's protocols. RT-PCR was performed by the SYBRH FAST One-Step qRT-PCR Kit Universal (Kapa Biosystems, USA). In every reaction, 50 ng of total RNA was used, according to the manufacturer's protocol. Real-time quantitative PCR (Q-PCR) was conducted with the Mx3005P TM system from Stratagene (USA). The concentration of the primers was optimized (200 nM final concentration for the all genes analysed) according to the manufacturer's technical data sheet. The primers used for the Q-PCR were chosen with the help of the Beacon Designer v 9.1software and are listed in Table 2. Control reactions in which cDNA synthesis was circumvented, ensured that DNA products resulted from the amplification of cDNA rather than from DNA contamination.
Normalization of the results was performed by using rpoD as the housekeeping gene [50]. The rpoD gene was used to provide an internal control cDNA that was amplified with oligonucleotides rpoDf/rpoDr ( Table 2) and used to normalize the sample data. After the PCR, a melting curve was generated to confirm the amplification of a single product. The cycle in which the SYBR green fluorescence crossed a manually set cycle threshold (CT) was used to determine transcript levels. For each gene, the threshold was fixed based on the exponential segment of the PCR curve.
Relative transcript levels of genes of interest were analyzed using the 'Delta-delta method' as described previously [51]. Q-PCR analysis was performed in three (technical) replicates on two independent RNA isolations (biological replicates).

Bioinformatics
Database searches were performed with the BLAST 2.0 service of the National Center for Biotechnology Information (Bethesda, MD, U.S.A.). Amino acid sequences were aligned with MultAlin [52]. Synteny analysis was performed by using the BioCyc Database [53]. The putative promoter site prediction was performed with two bioinformatic applications, BPROM software (Softberry Inc., Mount Kisco, NY, U.S.A.) and NNPP 2.2 software (Berkeley Drosophila Genome Project-BDGP, Berkeley, CA, U.S.A.). Analysis of the cores and domains of NRPS, translated by orf8, was accomplished according to previous work [38,39]. Primer design for PCR and RT-PCR was carried out using DNASTAR and BeaconDesign, respectively.

Accession Numbers
The nucleotide and deduced amino acid sequences for the sup1 gene, encoding the PQQ-dependent Gcd in Ps. fluorescens strain X, has been deposited in the GenBank database under accession no. HQ383687. The nucleotide and deduced amino acid sequences for the sup2, sup3 and sup4 gene, encoding the PQQ biosynthesis proteins F, D and E, as parts of the flanking region in Ps. fluorescens strain X, have been deposited in the GenBank database under accession no. JQ039398. The nucleotide and deduced amino acid sequences of the genes located in the supX putative operon, as well as parts flanking this region, have been deposited in the GenBank database under accession no. JQ039399.