Complex Interplay between FleQ, Cyclic Diguanylate and Multiple σ Factors Coordinately Regulates Flagellar Motility and Biofilm Development in Pseudomonas putida

Most bacteria alternate between a free living planktonic lifestyle and the formation of structured surface-associated communities named biofilms. The transition between these two lifestyles requires a precise and timely regulation of the factors involved in each of the stages that has been likened to a developmental process. Here we characterize the involvement of the transcriptional regulator FleQ and the second messenger cyclic diguanylate in the coordinate regulation of multiple functions related to motility and surface colonization in Pseudomonas putida. Disruption of fleQ caused strong defects in flagellar motility, biofilm formation and surface attachment, and the ability of this mutation to suppress multiple biofilm-related phenotypes associated to cyclic diguanylate overproduction suggests that FleQ mediates cyclic diguanylate signaling critical to biofilm growth. We have constructed a library containing 94 promoters potentially involved in motility and biofilm development fused to gfp and lacZ, screened this library for FleQ and cyclic diguanylate regulation, and assessed the involvement of alternative σ factors σN and FliA in the transcription of FleQ-regulated promoters. Our results suggest a dual mode of action for FleQ. Low cyclic diguanylate levels favor FleQ interaction with σN-dependent promoters to activate the flagellar cascade, encompassing the flagellar cluster and additional genes involved in cyclic diguanylate metabolism, signal transduction and gene regulation. On the other hand, characterization of the FleQ-regulated σN- and FliA-independent PlapA and PbcsD promoters revealed two disparate regulatory mechanisms leading to a similar outcome: the synthesis of biofilm matrix components in response to increased cyclic diguanylate levels.


Introduction
The alternation between a free-swimming planktonic lifestyle and the formation of structured highly cooperative polymer-encased sessile communities, named biofilms, is a common feature of bacterial life in the environment [1]. Biofilm formation is a form of coordinated collective behavior that has been regarded as an evolutionary precursor of developmental processes. Biofilm development proceeds through stages of adhesion, proliferation and microcolony formation and maturation, and is terminated by programmed biofilm dispersal [2,3]. Transition between the planktonic and biofilm lifestyles and through the different stages of biofilm development requires the timely production of different factors in response to environmental and physiological signals, involving a variety of signal transduction and regulatory pathways to connect such cues to the adequate physiological responses [3].
The nucleotide cyclic diguanylate (c-di-GMP) is a second messenger that is ubiquitously used in bacteria for signaling the transition between the planktonic and sessile lifestyles. c-di-GMP is synthesized from GTP by diguanylate cyclase (DGC) activities, associated to proteins with GGDEF domains, and degraded by specific phosphodiesterase (PDE) activities, associated to proteins with EAL or HD-GYP domains. Bacterial genomes often encode multiple proteins containing one or more of these domains, implying that the c-di-GMP levels are likely regulated in a complex fashion. Changes in c-di-GMP concentration are sensed by effectors, which in turn regulate a variety of processes, generally related to motility, biofilm development or virulence, acting at the transcriptional, translational or posttranslational levels. The biology of cdi-GMP signaling has been extensively reviewed in recent years [4][5][6][7][8].
FleQ has long been known as an enhancer-binding protein belonging to the NtrC/NifA family of σ N -dependent promoter activators, and as the master regulator of flagellar biogenesis in Pseudomonas aeruginosa and other bacteria [9][10][11]. In P. aeruginosa, flagellar promoters are regulated by a four-tiered cascade. Class I promoters drive the expression of the two major regulatory elements, the transcriptional activator FleQ and the alternative σ factor FliA [9,10,12]. FleQ directly activates a set of σ N -dependent Class II promoters, and indirectly activates Class III promoters, which are also σ N -dependent and activated by Class II gene products FleSR, and Class IV promoters, which are transcribed by FliA-loaded RNA polymerase. FleQ activation of several flagellar promoters is antagonized by c-di-GMP [13,14], likely through high-affinity interaction with the AAA+ ATPase domain of FleQ [15,16]. FleQ also represses transcription of σ N -independent promoters driving the synthesis of biofilm matrix components, and repression is released by interaction of FleQ with c-di-GMP [13,17,18]. In addition, FleQ positively regulates some of these operons at high c-di-GMP concentrations [17,18]. The auxiliary protein FleN forms a complex with FleQ to antagonize its activity and enhance the action of c-di-GMP on both FleQ-activated [15,19,20] and repressed promoters [13,17].
Pseudomonas putida is a well-characterized Gram-negative soil and rhizosphere bacterium. Due to its metabolic versatility, P. putida is a model organism for biodegradation of organic toxicants, and a key asset in bioremediation [21]. The ability to form biofilms on both biotic and abiotic surfaces is a key to P. putida survival in its natural environment, and several factors relevant to biofilm development in P. putida have been identified [22]. The high molecular weight adhesin proteins LapA and LapF are important determinants for cell-surface and cellcell interactions in early and mature biofilms, respectively [23]. Flagella have been shown to contribute to initial surface attachment and the transition from microcolonies to the mature biofilm [22]. The extracellular matrix of P. putida biofilms contains extracellular DNA [24] and a mixture of exopolysaccharide (EPS). Four gene clusters involved in EPS synthesis and export are present in P. putida and the contribution of different types of EPS to the extracellular biofilm matrix and biofilm stability has been determined [25][26][27][28]. In addition, P. putida biofilms undergo rapid dispersal in response to nutritional stress [29]. Studies performed in P. putida and the related species Pseudomonas fluorescens have determined that dispersal is effected by proteolytic cleavage of LapA by the periplasmic protease LapG [6,30,31].
Despite the growing understanding of the structural and mechanistic details of the biofilm cycle, little is known of the signals and regulatory mechanisms governing biofilm development in P. putida. A role for c-di-GMP as a key regulator of biofilm formation is inferred from the aggregative hyper-biofilm forming phenotype obtained by overexpression of enzymes with diguanylate cyclase (DGC) activity [32,33]. Although the P. putida KT2440 genome bears 43 genes encoding GGDEF, EAL or HD-GYP domain proteins [34,35], only the putative DGC MorA and the PDE BifA have been shown to be unequivocally involved in biofilm development [36,37]. Synthesis of the adhesin LapF, involved in biofilm maturation, is dependent on the stationary phase sigma factor σ S [38], and a role of c-di-GMP and FleQ in the expression of lapA and the cellulose synthesis bcs operon was recently described [21,39]. In addition, starvation-induced biofilm dispersal is triggered by a decrease in the intracellular c-di-GMP concentration effected by BifA [37]. Such decrease releases the inhibition of the protease activity of LapG by the membrane-bound c-di-GMP effector LapD, thus allowing LapA proteolysis [6,30,31].
Recently, we undertook the isolation of biofilm-deficient mutants of P putida KT2442 [40]. Five of our biofilm-deficient mutants were found to bear insertions in fleQ. Here we describe the phenotypic characterization of a null fleQ mutant and the impact of FleQ and c-di-GMP on the regulation of biofilm development and flagellar biogenesis using a novel approach based on the screening of a partial P. putida KT2440 ordered promoter library, containing 94 promoters potentially related to these two processes.

Biofilm growth and quantification
For procedures involving biofilm growth, overnight cultures grown in LB broth were diluted in the same medium to an A 600 of 0.1 and 150 μl were dispensed into wells of Costar 96 microtiter polystyrene plates (Corning). The plates were incubated at 25°C with moderate shaking (150 rpm) for the desired period of time and processed for planktonic and biofilm growth quantification, essentially as described [53]. Dilution-based planktonic and biofilm growth curves were performed as previously described [47]. For each experiment, at least 3 biological replicates were assayed in octuplicate.
For pellicle and aggregate detection, fresh colonies were inoculated in glass tubes containing 5 ml of 1% bacto-tryptone (Difco) broth [54]. Cultures were incubated overnight at 30°C with shaking, after which the tubes were placed on a rack for 10 minutes and documented by digital photography. Plasmids used in this work are summarized in Table 1 and S1 Table. Sequences of oligonucleotides used in PCR reactions are available upon request. All DNA manipulations were performed following standard procedures [41]. Restriction and modification enzymes were used according to the manufacturers instructions (Fermentas, Roche and NEB). When required, blunt ends were generated using the Klenow fragment or T4 DNA polymerase. E. coli DH5α was used as a host in cloning procedures. All cloning steps involving PCR were verified by commercial sequencing (Secugen). Plasmid DNA was transferred to E. coli and P. putida strains by transformation [41], triparental mating [55] or electroporation [56]. Site-specific integration of miniTn7 derivatives was performed essentially as described [51]. Two Gateway 1 expression vectors, pMRB2 and pMRB3, were constructed to host the ordered promoter library. The gfpmut3 allele was PCR-amplified, and the resulting product was cleaved with BamHI and BglII and cloned in the single BamHI site of pMPO234 to yield the gfpmut3-lacZ fusion vector pMRB1. Then, the Gateway 1 Vector Conversion System attR1-Cm r -ccdB-attR2 cassette was cloned at the SmaI site in pMRB1 in both orientations to generate pMRB2. For construction of a miniTn7-based fleQ complementation plasmid, a DNA fragment containing the complete fleQ and its promoter region was PCR-amplified, the PCR product cleaved with EcoRI and XbaI and ligated into EcoRI-and SpeI-digested pUC18Sfi-miniTn7BB-Gm to yield pMRB73.
For generation of a bifA deletion in the fleQ::miniTn5-Km mutant, the gentamycin resistance cassette from miniTn7BB-Gm was excised with NcoI and cloned into the sole NcoI site in pMRB30 to disrupt the kanamycin resistance gene, yielding pMRB90. The disruption vectors were transferred to the corresponding hosts (KT2442 or MRB35, respectively) by electroporation. Selection of integration, allelic replacement and FLP-mediated excision of the kanamycin resistance marker was performed essentially as described [57,58]. The structure of the deleted bifA locus was verified by PCR and Southern blot.
For the construction of pMRB95, a plasmid overproducing a chitin binding domain (CBD)intein-FleQ fusion for protein purification, the fleQ open reading frame was PCR amplified and cloned into NdeI-and EcoRI-cleaved pTYB12 (New England Biolabs).

Construction of an ordered promoter library
For construction of an ordered library of biofilm-and motility-related promoters, intergenic regions of the complete P. putida KT2440 chromosome sequence [59] harbouring promoters were identified based on experimental evidence in the literature and/or an operon prediction algorithm [Database of prOkaryotic OpeRons (DOOR)] [60, 61] (S1 Table). A subset of promoter-bearing intergenic regions potentially related to motility and/or biofilm development were selected and PCR-amplified using a 5'CACC tag in the forward primer, directionally cloned into pENTR-D/TOPO and subsequently transferred to pMRB2 or pMRB3. According to the orientation of the attR1-Cm r -ccdB-attR2 cassette in pMRB2 and pMRB3, forward strand promoters (as displayed in the P. putida KT2440 genome sequence) were cloned in pMRB2 and reverse strand promoters were cloned in pMRB3 to grant proper orientation relative to the reporter genes. Intergenic regions containing two divergent promoters of interest were cloned in both vectors. The library was constructed and stored in E. coli DH5α.

Phase-contrast microscopy
Phase contrast microscopy of surface adhesion was performed on a Leica DMI4000B inverted microscope using 20x objective and 1.6x ocular magnification. Cells from mid-exponential (A 600 = 0.2-0.3) LB cultures of the selected P. putida strains were serially diluted (10 2 −10 4 ) in LB and samples were transferred to wells of Costar 96 microtiter polystyrene plates (Corning). Attachment was allowed to proceed for 30 minutes, planktonic cells were removed by washing twice with 150 μl LB, and then 50 μl LB was added to the wells. For short-term assessment of surface attachment, plates were incubated at room temperature for 3 hours, and then cells deposited on the plane of the well surface were video recorded for 1 minute. For long-term assessment of surface attachment, a field containing a single cell was selected and recorded in a time-lapse series at 2-minute intervals during 16 hours.

Swimming and swarming assays
Swimming assays were adapted from [62]. Tryptone motility plates containing 0.3% Bactoagar (Difco) were toothpick-inoculated with fresh colonies and incubated for 12 h at 30°C. Digital photographs were taken, and the swimming zone diameter was measured and normalized that of the wild-type. At least 3 biological replicates were assayed for each strain.
Swarming assays were performed essentially as described by [63], with some modifications. Two and a half microliters of overnight LB cultures were spotted onto the center of plates containing 0.5% PG-agar [proteose peptone No. 3 (Difco 212693), 0.5%, and glucose, 0.2%, with Difco Bacto-Agar] and plates were incubated at 25°C for 40 hours. At least two biological replicates per experiment were assayed in three separate experiments.

Congo Red (CR) binding assay
To assess cellulose-like polysaccharide production the CR binding assay was adapted from [54]. Bacterial cells from 2 ml overnight cultures in LB were collected by centrifugation and resuspended in 1 ml 1% bacto-tryptone (Difco) broth containing 40 μg/ml CR. The mixtures were incubated at 30°C with shaking at 180 rpm for 180 min. The bacterial cells and bound CR were removed by centrifugation (1 min, 13000 rpm), and the amount of CR remaining in the supernatants was determined by measuring the absorbance of the supernatant at 490 nm, using the same medium supplemented with 40 μg/ml CR as a control.

Promoter library screening
For screening the ordered promoter library, the fusion plasmids were transferred to the appropriate P. putida strains by mating. Exconjugant colonies were inoculated in deep-well microtiter plate wells containing LB and grown overnight as described above. Cultures were diluted 100-fold and re-grown for 24 h in the same conditions, except that 2 mM salicylate was added to the strains bearing pMRB89. Gene expression was assessed by GFP fluorescence measurements. To this end, 200 μl culture samples were transferred to Costar 96-well microtiter polystyrene plates to determine the A 600 , from which 150 μl were subsequently transferred to Microfluor 1 Black 96-well plates (Thermo LabSystems) to register fluorescence using a 485 nm band pass excitation filter and a 520 nm emission filter on a POLARstar Omega Fluorometer (BMG Labtech). Specific fluorescence was calculated by dividing the fluorescence reading by the A 600 reading.

β-galactosidase assays
For β-galactosidase assays of lacZ fusions, overnight cultures bearing the corresponding fusion plasmids were diluted in 5 ml LB to an A 600 of 0.01 and incubated for 24 hours at 30°C with shaking. Growth was then stopped, and β-galactosidase activity was determined from sodium dodecyl sulfate-and chloroform-permeabilized cells as described [64].
RNA preparation and real time qRT-PCR. Total RNA was prepared from stationary phase cultures as previously described [65]. Reverse transcription (RT) of total RNA (5 μg) was carried out using the high-capacity cDNA Archive kit (Applied Biosystems), with random hexamers as primers. Negative and positive controls were performed with total RNA or genomic DNA as templates, respectively. qPCR was performed using specific primers for the lapA and bcsD coding sequences and SYBR Green technology in an ABI Prism 7000 Sequence Detection System (Applied Biosystems) essentially as described [66].

FleQ purification and gel mobility shift assays
FleQ was overproduced from the T7 promoter as a N-terminal CBD-intein fusion. T7 RNA polymerase-expressing strain NCM631 bearing pIZ227 was used as a host. Overproduction and purification using the IMPACT kit, featuring chitin affinity chromatography followed by intein self-cleavage induced by reducing conditions, were performed essentially as described for P. aeruginosa FleQ [13]. The purified protein was dialized against FleQ storage buffer (20 mM Tris-HCl, pH 8.0; 250 mM KCl; 0.1 mM EDTA; 2 mM DTT; 50% glycerol), and stored in aliquots at -80°C. Protein concentration was determined by means of the Bio-Rad Protein assay (Bio-Rad), and purity was assessed by SDS-PAGE.
Gel mobility shift assays were performed essentially as described [13]. The PcdrA, Ppp1155, PlapA and PbcsD promoter regions were PCR amplified from pCdrA C , and clones PP1155, PP0168 and PP2629 from the promoter library, repectively. The resulting products were subsequently cleaved with appropriate restriction enzymes to yield fragments <500 bp. The cleaved probes were challenged with different FleQ concentrations in a binding buffer containing 10 mM Tris, pH 7.8; 8 mM magnesium acetate; 50 mM KCl; 5% glycerol, and 250 ng ml -1 BSA for 30 minutes at room temperature, resolved in native 5% PAGE at 4°C, stained with ethidium bromide and documented by digital photography.

FleQ is required for flagellar motility, surface attachment and biofilm formation
In order to characterize the contribution of FleQ to the biofilm developmental cycle of P. putida, we examined the biofilm development kinetics of strains KT2442 (wild-type) and MRB35 (fleQ) in LB by means of a serial dilution-based growth curve method we described recently, in which a dilution series is used to recapitulate the time course of planktonic and biofilm growth in microtiter plate wells [47] (Fig 1A). The wild-type strain displayed a steady increase in biofilm growth to reach a maximum coincident with the onset of stationary phase, and then biofilm biomass decreased, consistent with nutrient starvation-induced biofilm dispersal in this organism [47]. In contrast, surface-attached biomass of the fleQ mutant MRB35 was negligible during the whole growth curve. Complementation with a site-specifically inserted miniTn7 transposon derivative expressing FleQ from its natural promoter restored a normal biofilm formation and dispersal pattern in MRB35 (S1 Fig). These results strongly suggest that FleQ integrity is critical for biofilm formation.
The fleQ mutant is non-motile (S2 Fig), as expected from its assumed role as a regulator of flagellar biogenesis. On the other hand, the lack of flagella has been shown to correlate with surface adhesion defects in P. putida [67][68][69]. Thus, to clarify whether the biofilm formation defect of the fleQ mutant is related to the lack of flagellar motility, the evolution of planktonic and biofilm growth of the non-motile mutant MRB47, bearing a miniTn5-Km insertion in fliG, which encodes a flagellar motor protein, was analyzed by means of dilution series-based growth curves as above (Fig 1B). Planktonic and biofilm growth of MRB47 was slow compared to that of the wild-type strain. However, this phenotype was relatively mild compared to that of the fleQ mutant (Fig 1A), as substantial levels of biofilm biomass were achieved. Analysis of other non-motile mutants bearing insertions in the flagellar structural genes flgG, fliF and fliN and fliP yielded similar results [40]. These results indicate that the loss of flagellar motility does not suffice to explain the severe biofilm formation defect of the fleQ mutant.
To determine whether the lack of FleQ has an impact on bacterial adhesion to surfaces, we used phase-contrast microscopy to analyze the adhesion phase of biofilm formation. To this end, cells of the wild-type, fleQ and fliG strains were allowed to attach to the bottom of polystyrene microtiter plate wells, and were then recorded in 1-minute videos (Fig 2). Observation of the images obtained revealed that the majority of the wild-type and fliG cells were immobilized on the polystyrene surface, strongly suggesting that both strains are capable of irreversible attachment. In contrast, the fleQ mutant cells displayed characteristic Brownian motion and were not immobilized in these conditions. Furthermore, monitoring the evolution of a single fleQ cell offspring for 16 hours in a 2-minute time-lapse movie failed to reveal any significant attachment (data not shown), strongly suggesting that FleQ is absolutely required for strong, irreversible interaction with the surface. Taken together, these results indicate that the lack of flagellar motility does not impair surface attachment, and therefore the strong adhesion defect of the fleQ mutant must be due to an additional role of FleQ in the regulation of cell-surface interactions.

FleQ is required for multiple phenotypes associated with high c-di-GMP levels
Increased c-di-GMP levels have been shown to stimulate surface adhesion and biofilm formation in P. putida [33,63]. We have recently described that mutational inactivation of bifA, encoding a c-di-GMP phosphodiesterase, results in elevated intracellular c-di-GMP concentration [37]. To test whether high c-di-GMP levels may boost biofilm formation in the absence of fleQ, the ΔbifA fleQ mutant MRB50 was constructed and its biofilm development kinetics was assessed as above. The behavior of MRB50 was indistinguishable from that of the fleQ mutant MRB35 (Fig 1C), strongly suggesting that FleQ is required for c-di-GMP-dependent stimulation of biofilm growth.
Increased c-di-GMP levels have also been shown to promote cell aggregation in liquid medium, biofilm formation in the medium-air interphase (pellicle) and increased adsorption of Congo Red (CR) to the cell surface [37,63]. These phenotypes were also tested in the wildtype strain KT2442 and its ΔbifA, fleQ and ΔbifA fleQ derivatives MRB32, MRB35 and MRB50 (Fig 3). The wild-type and fleQ mutant strains showed little pellicle biomass and no detectable aggregation (Fig 3A). In contrast, most of the ΔbifA mutant biomass was in the form of glassattached pellicle and cell aggregates, consistent with the high c-di-GMP levels present in this strain. Interestingly the double ΔbifA fleQ mutant MRB50 did not display an aggregative behavior, and showed decreased levels of pellicle formation. On the other hand, the ΔbifA mutant MRB32 bound twice as much CR as the wild-type (Fig 3B). In contrast, CR retention by the fleQ mutant MRB35 was 2-fold reduced relative to the wild-type. CR retention by the double ΔbifA fleQ mutant MRB50 was also decreased (4-fold). The fact that FleQ is required for an array of phenotypes generally associated to sessile growth and induced by high intracellular c-di-GMP concentrations strongly suggests the involvement of FleQ in c-di-GMP regulation of adhesion and biofilm growth.

FleQ and c-di-GMP coordinately regulate motility-and biofilm-related promoters
To determine the extent of FleQ-mediated regulation of P. putida lifestyle changes, we screened a library of 94 P. putida KT2440 promoters fused to the reporters gfpmut3 and lacZ. This library includes promoters directing the synthesis of adhesion factors, extracellular matrix components, appendages, and known or assumed regulatory elements of biofilm development and motility, such as a variety of transcription factors and potential c-di-GMP DGC or PDE proteins (See Materials and Methods and S1 Table for details). To this end, the library fusion plasmids, along with the empty vector pMRB1, were transferred to the wild-type P. putida KT2442 and its fleQ derivative MRB35, and expression was determined from fluorescence measurements of stationary phase planktonic LB cultures grown in deep-well microtiter plate wells (Fig 4A). Twenty promoters of the library, displaying at least 1.5-fold differential expression in the fleQ mutant relative to the wild-type strain, were initially chosen for further analysis. All but one promoter were upregulated in the wild-type relative to the fleQ mutant, suggesting that FleQ acts preferentially as an activator in P. putida. For confirmation, individual cultures of the wild-type and fleQ strains bearing the FleQ-regulated fusions obtained in the screening were re-tested by means of β-galactosidase assays. Sixteen promoters were confirmed to show significant FleQ-dependent regulation (p<0.05) (Fig 5A). The set included seven promoters of the flagella/chemotaxis gene cluster, located upstream from fliC, flgB, fliS, flgA, flhA, cheV3 and flgM and predicted to transcribe the fliC-fleL, flgBCDE-pp4387, fliST, flgA, flhAF-fleN-fliA, cheV3R and flgMN operons, and additional promoters upstream from PP4328, predicted to transcribe the pp4328-4329 operon, encoding a FliK homolog and a FlhB domain protein, and mcpG, encoding a gamma-amino butyric acid-responsive methyl-accepting chemotaxis protein (MCP) [70]. All these promoters were positively regulated by FleQ to various extents, ranging from 1.8-fold (Ppp4328) to 54-fold (PfliC). These results support the notion that FleQ is the master regulator of the synthesis of the flagellar motility and chemotaxis apparatus in P. putida.  . Red and blue lines represent 2-fold differential expression. Data points showing over 2-fold differential expression are shown in dark red or dark blue. Data points showing over differential expression greater than 1.5-fold but lower than 2-fold are shown in orange or cyan. Data from the experimental controls (KT2442 and KT2442 bearing the empty vector pMRB1) are shown in yellow and pink,respectively. Promoters selected for confirmation are indicated in each plot. The original data for these plots are shown in S2  Two additional FleQ-regulated promoters, PlapA, directing the synthesis of the high molecular weight adhesin LapA, and PbcsD, predicted to direct transcription of the bcsDEFGRQABZC operon, encompassing the cellulose synthesis and export bcs genes, were related to the biofilm lifestyle [28,38]. While PlapA was positively regulated (8-fold), PbcsD displayed modest negative regulation (~1.5-fold) by FleQ. It should be noted that PbcsD was initially picked in the screening as a FleQ-activated promoter. However, further verification indicated that the expression levels obtained in the screening were atypical, and confirmed negative regulation of this promoter. Promoters upstream from PP1144, PP3711 and morA, encoding GGDEF/EAL proteins were positively regulated (~3-fold) by FleQ, while Ppp3932, directing the synthesis of a putative GGDEF domain protein was negatively regulated, albeit modestly (1.4-fold). Finally, Ppp4100, predicted to direct transcription of the pp4100-gacA-uvrC-pgsA operon, encoding a Cro/cI family regulator, response regulator GacA, ABC excinuclease C subunit and CDP-diacylglycerol-glycerol-3-phosphate 3-phosphatidyltransferase, was 3-fold upregulated by FleQ, indicating that GacA, a regulatory element known to be relevant to biofilm formation in P. aeruginosa [71], is also a part of the FleQ regulon. These results are consistent with the phenotypic analysis above showing that FleQ is required for both flagellar motility and biofilm development, and suggests that it may do so by means of a complex network involving additional regulators and modulation of c-di-GMP synthesis and hydrolysis. Similarly, the promoter library was screened for differential expression in response to changes in c-di-GMP levels. To this end, we overproduced the E. coli DGC YedQ or the PDE YjhH from plasmids pYedQ or pMRB89, to increase or decrease, respectively, the intracellular c-di-GMP concentration. In pMRB89 yhjH is under control of the Psal promoter, and expression was induced by adding 2 mM salicylate to the growth medium. In pYedQ, yedQ is transcribed from the Plac promoter. However, IPTG-induced YedQ overexpression inhibited growth and even in the absence of IPTG, YedQ production provoked strong aggregation that hampered fluorescence measurements (data not shown). Thus, the non-aggregating lapAderivative of KT2442 MRB34 was used in this experiment, and IPTG was not added to the growth medium. Accordingly, the library fusion plasmids, along with the empty vector pMRB1, were transferred to MRB34 bearing pYedQ or pMRB89, and expression was determined from fluorescence measurements as above (Fig 4B).
Eighteen promoters displaying >1.5-fold differential expression in response to changes in the c-di-GMP levels were chosen for further analysis, fourteen of which were repressed by cdi-GMP and four were induced. Thirteen of these promoters were confirmed to be significantly regulated by c-di-GMP by means of β-galactosidase assays, as above ( Fig 5B). Interestingly, a great deal of overlap was observed between the set of FleQ-regulated and the set of cdi-GMP-regulated promoters (Fig 6). Nine FleQ-activated promoters were downregulated in the presence of c-di-GMP. This set includes flagella and motility-related promoters (PfliC, PflgB, PfliS, PflgA, PcheV3, PflgM and Ppp4328), as well as PmorA and Ppp4100. The PlapA promoter was upregulated by both FleQ and c-di-GMP, while Ppp3932 was downregulated by FleQ and upregulated by c-di-GMP. Two additional promoters, Ppp2357, predicted to direct transcription of the type I pili biogenesis operon pp2357-pp2358-2359-2360-csuCDE, and PbifA, directing the synthesis of the PDE BifA, involved in starvation-induced dispersal [37], were negatively regulated by c-di-GMP in a FleQ-independent fashion. Of these two, at least PbifA is expected to be FleQ-regulated, as it was shown to be dependent on the flagellar σ factor FliA [46]. Since FleQ responds to changes in the c-di-GMP levels, we also tested FleQ regulation by comparing the expression of the corresponding fusions in the wild-type and fleQ strains bearing plasmids pMRB89 or pYedQ (Fig 6). Expression of the Ppp2357 promoter was 2-fold repressed by c-di-GMP in both the wild-type and fleQ backgrounds, thus confirming that this promoter is FleQ-independent (Fig 6A). In contrast, 2-fold repression of the PbifA promoter by c-di-GMP was only observed in the wild-type background, while the fleQ mutant displayed basal expression levels regardless of the c-di-GMP concentration (Fig 6B). These results indicate that PbifA is positively regulated by FleQ, but such regulation is only observed at low c-di-GMP concentrations. σ factor dependency of FleQ-regulated promoters FleQ is an enhancer-binding protein known to activate flagellar promoters dependent on the alternative σ factor σ N . On the other hand, FleQ has also been shown to indirectly regulate transcription from promoters dependent on the flagella-specific alternative σ factor FliA [10]. To address the σ factor dependence of our FleQ-regulated promoter set, expression from the gfp-lacZ fusions was assessed from the wild-type strain KT2440 (the parent strain of KT2442), its rpoN mutant derivative KT2440rpoN and its fliA mutant derivative by means of β-galactosidase assays of stationary phase LB cultures (Fig 5C and 5D).
All nine flagellar motility and chemotaxis-related promoters were upregulated in the wildtype strain relative to the rpoN mutant. Promoters PflgB, PfliC, PmcpG, PfliS and PflhA displayed strict dependency on the alternative σ factor, as their basal expression levels in the rpoN mutant were comparable to those in the vector construct and expression was elevated 5-to 58-fold (for PflhA and PflgB, respectively) in the presence of σ N . A lesser extent of regulation (2-to 3-fold) and higher σ N -independent basal levels were observed for PflgA, Ppp4328, PcheV3 and PflgM, indicating partial σ N -dependence of these promoters. The latter pattern was also observed with the Ppp1144, Ppp3711, PmorA promoters, driving the synthesis of three GGDEF/EAL domain proteins, and Ppp4100, directing the synthesis of the transcription factor GacA and other proteins, while Ppp3932 was inversely regulated: its expression was increased 2-fold in the absence of σ N . In contrast, PlapA and PbcsD, responsible for the production of biofilm matrix components were not affected by the rpoN mutation.
Six of the flagellar motility and chemotaxis-related promoters were upregulated in the wildtype strain relative to the fliA mutant. Promoters PfliC, PfliS and PmcpG displayed strict dependency of the alternative σ factor, as their basal expression levels in the fliA mutant were comparable to those in the vector construct and expression was elevated 4-to 30-fold (for PmcpG and PfliC, respectively) in the presence of FliA. A lesser extent of regulation (2-to 3-fold) and higher FliA-independent basal levels were observed for PcheV3, Ppp4328, and PflgM, indicating partial FliA-dependence of these promoters. The latter pattern was also observed with the Ppp1144, Ppp3711 and PmorA promoters. High basal expression in the fliA mutant suggests that these promoter regions may also be recognized by RNA polymerase loaded with a σ factor other than FliA, but the biological significance of this is unclear. Expression of three flagella-related promoters (PflhA, PflgA and PflgB), the two biofilm matrix-related promoters PlapA and PbcsD, Ppp4100 and Ppp3932 was not stimulated by the presence of FliA, and three of them (Ppp3932, PflgA and PflgB) were moderately (2-fold) downregulated in the wild-type relative to the fliA mutant. Taken together, our results strongly suggest that FleQ is a high-ranked regulator in a cascade that involves the transcriptional regulation of σ N -dependent, FliA-dependent and σ Nand FliA-independent promoters in P. putida. In addition, σ N -and FliA-dependency appear to be associated with flagellar motility and chemotaxis-associated functions, while synthesis of biofilm matrix components is regulated by FleQ in a σ N -and FliA-independent fashion.
FleQ directly regulates a subset of flagellar promoters. To determine whether FleQdependent regulation of the P. putida target genes identified is direct or indirect, plasmids bearing gfpmut3-lacZ fusions to the complete set of FleQ-regulated promoters, along with the empty vector pMRB1 were transformed into E. coli ET8000, a heterologous background not encoding a FleQ ortholog, and ET8000 bearing the FleQ-producing plasmid pMRB99, and expression was assessed by means of β-galactosidase assays of LB-grown stationary phase cultures (Fig 7). Expression from the PflgB, PflhA and PflgA fusions was increased 4-to 7-fold in the presence of FleQ, indicating that FleQ directly activates transcription from these three promoters. In contrast, no significant change in expression was observed with the FliA-regulated PfliC, PfliS, PflgM, PcheV3, Ppp4328, PcheV3, Ppp4100, PmcpG, PmorA, Ppp1144, Ppp3711 and Ppp3932 promoters, consistent with the notion that the effect of FleQ on these promoters is indirect, and mediated by FleQ control of the flagellar cascade.
FleQ and c-di-GMP regulation of the PlapA and PbcsD promoters. For a deeper understanding of the role of FleQ and c-di-GMP in the regulation of biofilm development, we studied further the regulation of the PlapA and PbcsD promoters. To this end, we assessed the observed FleQ-dependent regulation by qRT-PCR. Assays performed with cDNA obtained from the wild-type and fleQ strain showed that the lapA mRNA levels were 2-fold (p<0.01) greater in the wild-type strain relative to the fleQ mutant, while bcsD mRNA levels were 32-fold (p<0.05) elevated in the fleQ mutant relative to the wild-type. These results confirm that FleQ is a positive regulator of PlapA and a negative regulator of PbcsD. In addition, we evaluated the combined effect of fleQ and c-di-GMP regulation by assaying expression from the PlapA and PbcsD fusions in the wild-type and fnr strains bearing the YedQ-and YhjH-producing plasmids pYedQ and pMRB89, as shown above for Ppp2357 and PbifA (Fig 6). Under a low c-di-GMP regime, PbcsD expression was 1.4-fold increased in the absence of FleQ (Fig 6C). However, when c-di-GMP levels were elevated, PbcsD expression was high and unresponsive to FleQ. These results suggest that PbcsD is repressed by FleQ at low c-di-GMP concentrations, and high c-di-GMP levels antagonize FleQ-dependent repression. In contrast, PlapA expression was low in the fleQ mutant, and was elevated 4-and 10-fold in the wild-type strain under the low and high c-di-GMP regimes, respectively (Fig 6D), suggestiing that FleQ is an activator of PlapA transcription, and high c-di-GMP levels further stimulate FleQ-dependent activation.
We also used the heterologous E. coli ET8000 background, as shown above, to assess whether FleQ regulation of the two biofilm-related promoters is direct or indirect (Fig 7). When assayed in these conditions, the PbcsD promoter was 2-fold downregulated in the presence of FleQ, suggesting that FleQ is a direct regulator of PbcsD transcription. However, no regulation was observed with the PlapA promoter. While this result may suggest that the effect of FleQ on PlapA is indirect, alternative explanations, such as the requirement for additional P. putida factors other than FleQ for in vivo PlapA activation, are compatible with direct regulation of PlapA. Recently, a consensus sequence for the P. aeruginosa FleQ binding motif was derived from the analysis of thirteen verified FleQ binding sites [18]. Bioinformatics search for this motif using the FIMO software [72] revealed three highly significant (P<10 −4 ) matches at the PlapA promoter region and two at PbcsD (S3 Table), suggesting that FleQ may bind these two regions directly. Finally, to clarify this issue, we resolved to determine the ability of FleQ to interact with the PlapA and PbcsD promoter regions by means of gel mobility shift analysis. To this end, P. putida KT2440 FleQ was overproduced as a N-terminal CBD-intein-FleQ fusion, purified by chitin affinity chromatography, and the tag was subsequently self-cleaved to release the native FleQ protein (S3 Fig). Activity of this FleQ preparation was initially tested in nonradioactive gel mobility shift assays with the PCR-amplified P. aeruginosa PcdrA, bearing three experimentally verified FleQ binding sites [18], as a probe (Fig 8). This PcdrA probe was fully retarded at both FleQ concentrations used (0.45 and 4.5 μM). As a negative control, a PCR fragment bearing the intergenic region of P. putida ORF PP1155, which was negative for FleQ regulation in our screening above, was digested with SacII and used as a probe. None of the two PP1155 promoter bands were substantially retarded, suggesting that our preparation of P.putida FleQ is and active capable of specific interaction with the P. aeruginosa PcdrA promoter region. Non-radioactive gel mobility shift assays were also performed using restriction enzyme-digested PCR products containing the PlapA and PbcsD promoter region fragments used for expression analysis above (Fig 8). At PlapA, 4.5 μM FleQ caused a full shift of the 550 and 297 bp bands, predicted to contain two and one FleQ binding sites, respectively. No alteration was observed in the mobility of the 152 bp band, not bearing any putative FleQ binding motifs. A similar result was obtained with the PbcsD promoter region: the 263 bp fragment, containing two putative sites, was fully retarded at 4.5 μM FleQ, while mobility of the additional 200 bp fragment was not altered. These results indicate that FleQ interacts specifically with the PlapA and PbcsD promoter regions. In addition, the high degree of correlation between the retarded fragments and the bioinformatics predictions is consistent with the notion that the predicted FleQ binding motifs indeed correspond to bona fide FleQ binding sites.

Discussion
Our results provide evidence that FleQ is the master regulator of the flagellar cascade in P. putida. Phenotypic analyses revealed that a fleQ mutant is strongly defective in flagella-mediated swimming and swarming motility. On the other hand, gene expression analyses revealed seven FleQ-activated promoters within the flagellar biogenesis gene cluster, and two additional FleQ-activated promoters driving the synthesis of flagellar and chemotaxis proteins. The set of FleQ-activated motility-related promoters could be further divided into two subsets, three promoters directly activated by FleQ and strongly dependent on σ N , and six promoters indirectly activated by FleQ, and co-dependent on σ factors σ N and FliA. The regulatory patterns in these two subsets are entirely consistent with those in Class II and Class IV flagellar promoters as defined in P. aeruginosa [10]. Even though our analysis is far from complete, we propose that a regulatory cascade reminiscent of (although arguably not identical to) that described in P. aeruginosa controls flagellar synthesis in P. putida. Additional evidence in this vein includes the conservation of the regulatory elements involved (FleQ, FleN, [73]. Flagellar gene regulation in P. putida was for the most part so far unexplored, and this study provides a fresh framework for further research on the regulation of the P. putida flagellar genes. Our results also support the notion that c-di-GMP interferes with FleQ in the activation of the flagellar cascade. Increased c-di-GMP levels have been shown to negatively regulate flagellar function and flagellar biogenesis in multiple bacteria by a variety of mechanisms [74], and c-di-GMP has been shown to interact directly with P. aeruginosa FleQ to inhibit its ATPase activity, essential for σ N promoter activation [13,15]. This phenomenon is relevant to the transition from a motile to a sessile lifestyle, during which c-di-GMP stimulates the synthesis of biofilm matrix components [74] while preventing flagellar motility, which would otherwise contribute to destabilize the biofilm structure [75][76][77]. Four additional promoters driving the synthesis of proteins involved in c-di-GMP signaling (PmorA, Ppp1144, Ppp3711 and PbifA) showed a regulatory pattern consistent with those of Class IV flagellar promoters. Regulation of c-di-GMP signaling by the flagellar cascade was previously documented in L. pneumophila [78]. As c-di-GMP is an effector of FleQ, the regulation of these factors may operate as feedback loops to fine-tune FIeQ activity. It is interesting that two of these genes provide additional connections with the biofilm developmental cycle: MorA is a DGC that regulates both flagellar motility and biofilm development [36,79], and BifA is a PDE responsible for the decrease of the c-di-GMP levels that trigger starvation-induced biofilm dispersal [35]. Regulation of BifA synthesis by the flagellar cascade is a clear-cut example of how a positive feedback loop may operate: decreased c-di-GMP levels induce the flagellar cascade, which in turn leads to FliAdependent activation of BifA synthesis to decrease the c-di-GMP levels further, simultaneously stimulating flagellar synthesis and biofilm dispersal. This mechanism likely acts as a checkpoint to foster quick dispersal, provided that synthesis of functional flagella to faciltitate escape of the cells released from the biofilm is underway.
We also provide evidence of the involvement of FleQ as a major regulator of biofilm development. A fleQ mutant was strongly deficient in surface attachment and biofilm formation. FleQ was also responsible for mediating certain responses to elevated c-di-GMP levels, as a fleQ null mutation suppressed the aggregation, pellicle formation and CR adsorption phenotypes of a ΔbifA strain, displaying high intracellular c-di-GMP concentrations. This is in sharp contrast with the situation in P. aeruginosa, in which lack of FleQ leads to the formation of wrinkly colonies as well as overproduction of the Pel and Psl exopolysaccharides and the high molecular weight adhesin CdrA [10,13]. However, a P. fluorescens Pf0-1 mutant lacking the FleQ ortholog AdnA was defective in adhesion and biofilm formation in a flagellum-independent fashion [80,81], suggesting a regulatory scheme similar to that in P. putida, although AdnA regulation of biofilm matrix components has not yet been documented.
Two promoters driving the synthesis of important components of the biofilm matrix are targets for coordinate FleQ-and c-di-GMP-dependent regulation: PlapA, responsible for the synthesis of the high molecular weight adhesin LapA and PbcsD, directing the synthesis of the cellulose synthase complex. The direct involvement of FleQ in the regulation of lapA and bcs transcription was described very recently [39]. Our In silico analysis identified three and two putative FleQ binding sites at the PlapA and PbcsD promoter regions, respectively, and gel mobility shift analysis revealed that fragments containing these regions were specifically bound by FleQ. Therefore, our observations confirm and expand these results, providing evidence of the location of the FleQ binding elements.
Our results show that FleQ represses PbcsD transcription, and repression is prevented by high c-di-GMP levels. This regulatory pattern is qualitatively equivalent to that described recently by means of qRT-PCR using bcsQ-specific primers [39]. However, this work a greater extent of repression (~5-fold vs.~1.5-fold) was observed, and our qRT-PCR results using bcsDspecific primers yielded even greater (>30-fold) repression. A second PbcsD-gfp-lacZ fusion including additional sequences (from -674 to +388), also yielded~1.5-fold regulation (data not shown). The reasons for this discrepancy between the qRT-PCR and promoter fusion results are currently unknown, and may be be related to differences between the two experimenrtal systems, or to as of yet unrevealed regulatory features of the bcs cluster. FleQ repression that is antagonized by c-di-GMP is reminiscent of the regulation of the pel, psl and cdrA operons [13], encoding biofilm matrix components in P. aeruginosa. Stimulation of cellulose synthesis by cdi-GMP may also operate by a second mechanism. Inspection of the P. putida BcsA sequence revealed high conservation of the PilZ domain in the catalytic subunit BcsA. This domain has been shown to interact with c-di-GMP to induce cellulose synthase activity in other organisms [82][83][84], and therefore similar posttranscriptional regulation is likely to occur in P. putida.
Our results also show that FleQ activates PlapA transcription, and activation is stimulated by high c-di-GMP levels, again consistently with recently published results [39]. Synergistic activation of LapA synthesis by c-di-GMP and FleQ provide a rationale to the high pellicle production and aggregative phenotypes of a ΔbifA mutant and the suppression of these phenotypes by inactivation of fleQ. FleQ is a member of the well-characterized bacterial enhancer-binding protein (bEBP) family involved in activation of σ N -bearing RNA polymerase-dependent transcription. However, unlike the FleQ-activated flagellar promoters, PlapA is not σ N -dependent. Previously, FleQ has been shown to act as an activator of the σ N -independent promoters for the P. aeruginosa pel and cdrA operons in the presence of high c-di-GMP concentrations [17,18]. A unique feature of PlapA is that it is activated by FleQ under both high and low c-di-GMP regimes, although c-di-GMP clearly stimulates activation. In contrast, pel and cdrA are repressed in the presence of low c-di-GMP levels, and activated when c-di-GMP levels are increased. Activation of σ N -dependent promoters by bEBPs involves a number of unique, highly specific mechanisms (binding to distant sites, oligomerization of the DNA-bound activator and ATP hydrolysis coupled to rearrangement of the transcription initiation complex) widely divergent from those observed in promoters dependent on other σ factors [85]. FleQ is unique among bEBPs in its ability to activate both σ Ndependent and σ N -independent transcription, and the elucidation of the mechanisms underlying the ability to interact with both forms of RNA polymerase is a main goal of our future research.
Taken together, the results presented in this work can be integrated with the available knowledge in a working regulatory model in which c-di-GMP levels control the P. putida lifestyle switch by means of a complex regulatory network involving multiple regulatory and signal transduction elements (Fig 9). FleQ acts as a positive regulator of flagellar promoters, both directly, and indirectly, and its activity is antagonized by c-di-GMP. We believe it is safe to assume that a four-tiered flagellar cascade similar to that described in P. aeruginosa likely occurs in P. putida [10]. FleQ also indirectly activates genes encoding several putative c-di-GMP synthesizing and degrading proteins, which may generate regulatory feedback loops in the system. Finally, c-di-GMP positively regulates biofilm development by two separate mechanisms: firstly, by modulating the activity of FleQ to promote the synthesis of key components of the biofilm matrix, and secondly, by inhibiting the signal transduction circuit leading to starvation-induced dispersal [29,30,37]. While some of the details of this circuit need to be clarified, we believe it represents a valid framework for further research on the regulation of the lifestyle switch in P. putida.  [72] and [46], respectively. (PDF) S1 Table. Promoters included in the ordered promoter library. 1 ORF identification code according to the Pseudomonas genome annotation. 2 Operon structure, according to the DOOR algorythm prediction for the P. putida KT2440 genome. 3 Description of the relevant gene products and the criteria used for selection of each promoter. (PDF) S2 Table. Results of the screenings for FleQ-and c-di-GMP-dependent regulation. The results shown are the averages and standard deviations of at least 3 independent experiments. (PDF) S3 Table. Predicted FleQ binding sites at the PlapA and PbcsD promoter regions. Prediction was performed according to [18]. 1 Coordinates are relative to the start codons. 2 A plus (+) sign designates the strand corresponding to the lapA or bcs transcript 3 Positions matching the consensus are indicated in bold. (PDF)