The FilZ Protein Contains a Single PilZ Domain and Facilitates the Swarming Motility of Pseudoalteromonas sp. SM9913

Swarming regulation is complicated in flagellated bacteria, especially those possessing dual flagellar systems. It remains unclear whether and how the movement of the constitutive polar flagellum is regulated during swarming motility of these bacteria. Here, we report the downregulation of polar flagellar motility by the c-di-GMP effector FilZ in the marine sedimentary bacterium Pseudoalteromonas sp. SM9913. Strain SM9913 possesses two flagellar systems, and filZ is located in the lateral flagellar gene cluster. The function of FilZ is negatively controlled by intracellular c-di-GMP. Swarming in strain SM9913 consists of three periods. Deletion and overexpression of filZ revealed that, during the period when strain SM9913 expands quickly, FilZ facilitates swarming. In vitro pull-down and bacterial two-hybrid assays suggested that, in the absence of c-di-GMP, FilZ interacts with the CheW homolog A2230, which may be involved in the chemotactic signal transduction pathway to the polar flagellar motor protein FliMp, to interfere with polar flagellar motility. When bound to c-di-GMP, FilZ loses its ability to interact with A2230. Bioinformatic investigation indicated that filZ-like genes are present in many bacteria with dual flagellar systems. Our findings demonstrate a novel mode of regulation of bacterial swarming motility.


Introduction
Motility is important for bacterial nutrient assimilation, growth, virulence, biofilm formation, and cell-cell contact [1]. Among various motility modes, swimming and swarming are vital in flagellated bacteria. Swimming is the movement in aqueous environments that is driven by one or several rotating flagella, with individual cells in motion [2][3][4], while swarming is the flagella-driven rapid movement of a large number of bacterial cells on viscous surfaces [3,[5][6][7]. Flagella are sophisticated nanomachines [8,9]. The number and arrangement of flagella vary among bacterial species. Some bacteria possess both constitutive polar flagella and inducible lateral flagella to support swimming and swarming motilities, respectively [8][9][10]. There is some evidence for interactions between the two flagellar systems. In the deep-sea bacterium Shewanella piezotolerans WP3, mutations interfering with the function of the polar flagellum induce the expression of lateral flagellar genes, while mutations disrupting lateral flagellar genes result in a decrease in the transcription of polar flagellar genes [11]. In Vibrio parahaemolyticus, the production of lateral flagella is inhibited by the expression of the polar flagellar genes and activated by the expression of

Protein Purification
The recombinant cells were pelleted (10,000× g for 10 min at 15 • C), resuspended in 40 mM phosphate buffer (pH 7.5) supplemented with 100 mM NaCl, and then disrupted with passage through a JN-02C French press (JNBIO, Guangzhou, China) at 300 psi. After centrifugation (15,000× g for 10 min at 15 • C), the pellets were exposed to 1% Triton and dissolved in PGE buffer (40 mM phosphate buffer [pH 7.5], supplemented with 50 mM NaCl, 5% glycerol, 0.5 mM EDTA, and 2 mM dithiothreitol) containing 6 M guanidine hydrochloride (GndHCl). The proteins were then dialyzed at 4 • C against PGE buffer. The soluble proteins thus obtained were further purified by chromatography as previously described [28].

Isothermal Titration Calorimetry
Measurements of the affinities and stoichiometries of FilZ and its mutant variants with c-di-GMP were performed with a MicroCal iTC200 system (GE Healthcare, Danderyd, Sweden) as previously described [29,30]. Data were processed using the MicroCal ORIGIN version 7.0 software.

Construction of the Mutants of Strain SM9913
The deletion mutant strains (∆filZ, ∆2230, and ∆0915) were constructed as described by Sheng et al. [27] with some modifications. Pairs of primers (pK18-filZ-up-F/R and pK18-filZdown-F/R; pK18-2230-up-F/R and pK18-2230-down-F/R; and pK18-0915-up-F/R and pK18-0915-down-F/R) were synthesized to amplify the homologous arms of the filZ, PSM_A2230, and PSM_A0915 genes separately. The amplified genes were then ligated into the suicide vector pK18mobsacB-Ery. The plasmids thus constructed were transferred into strain SM9913 for further in-frame deletion. The target sequences were confirmed by PCR with the primer pairs (filZ-screen-F/R, 2230-screen-F/R, and 0915-screen-F/R). All the primers are listed in Table S2.

Construction of Complementary Strains of SM9913
The complementary strains of SM9913 were constructed as previously described [26,27,31,32]. The plasmid pEV was used as a shuttle vector. All plasmids overexpressed the cloned proteins under the control of the plasmid-encoded promoter.
Primers pEV-filZ-F/R were used to construct the plasmid pEV filZ . To construct the plasmid pEV filZ-R13A , site-directed mutagenesis was performed using primers filZ-R13A-F/R and the template pEV filZ . The resulting plasmids were introduced into the corresponding mutants.

Bacterial Motility Assay
The motility of strain SM9913 and its mutant derivatives was assayed as previously described [26]. Briefly, marine LB plates with 0.3% or 0.5% agar (Bacto TM agar, Franklin Lakes, NJ, USA) were used to assay swimming or swarming, respectively. Plates were incubated at 15 • C, and the colony diameters on the swarm plates were measured over 3 days. The scatter plots of the changes in colony diameter were made using OriginPro 8.5 software. The scatterplots were then fit nonlinearly using the Boltzmann function in the Growth/Sigmoidal class. The curves of the colony expansion rates were obtained by plotting differences in diameter as a function of time from 24 to 72 h, using the fitted curves.

Total RNA Extraction and RT-qPCR
The transcription levels of PSM_A2230, filZ, and all the other genes from the lateral flagellar gene cluster were determined by Real-Time quantitative PCR (RT-qPCR). A 5 µL aliquot of an overnight culture of strain SM9913 was spotted at the center of each 0.5% agar plate, and the plates were incubated at 15 • C. Cells were collected from the colony edge at different culture times. The extraction of total RNA, the synthesis of cDNA, and the qPCR reaction were performed as described previously [27]. The relative levels of expression of filZ and PSM_A2230 were normalized using rpoD (the RNA polymerase sigma factor gene) as an internal reference.

Atomic Force Microscopy Imaging
SM9913 cells were collected from the edge of the swarming colonies at different culture times, and atomic force microscopy imaging was performed as previously described [27].

The Construction and Screening of E. coli BACTH Library and Two-Hybrid Assays
The construction of the E. coli BACTH (Bacterial Acetylate Cyclase Two-Hybrid) library and the library screening were performed as previously described with slight modifications [33,34]. Plasmids pKT25 and pUT18C were gifts from Professor Xiaoxue Wang [35]. Fragments of the Sau3AI-digested SM9913 genomic DNA were ligated with the BamHI-digested pKT25 vector, and aliquots of the ligation mixture were introduced chemically into 100 µL competent E. coli DH5α cells (Tsingke Biotechnology Co., Beijing, China). The ligation mixture was mixed with the thawed competent cells, and the cells were incubated on ice for 30 min. Cells were then incubated at 42 • C for 45 s, followed by incubation on ice for 2 min. After incubation, 0.5 mL of LB medium without antibiotics was added, and cells were preincubated at 37 • C for 1 h. Cells were then plated onto LB agar plates containing 50 µg/mL kanamycin, and the plates were incubated at 30 • C for 36 h. All the colonies from the plates were then collected, and the purified plasmids were used for the BACTH plasmid DNA library.
For library screening, protein FilZ-R13A was used as the bait [33,34]. Plasmid pUT18C-filZ-R13A (encoding the T18-FilZ-R13A hybrid protein) was introduced into the E. coli host strain BTH101. The resultant strain, BTH101 (pUT18C-filZ-R13A), was then transformed with the SM9913 DNA library, using 100 ng DNA. The transformed cells were plated on MacConkey medium (Solarbio, Beijing, China) until the red colonies became large enough to isolate. The plasmids from the isolates were sequenced to verify that they contained inserted DNA.

Bacterial Two-Hybrid Assays
The bacterial two-hybrid assay was performed using the BACTH system [35] with slight modifications. Plasmid pKT25 was digested with KpnI, and pUT18C was digested with BamHI and EcoRI. The coding region of the target genes (including the stop codon) was ligated into the digested plasmids. The recombinant pKT25 and pUT18C plasmids were introduced into E. coli BTH101 for the two-hybrid assays.

Pull-Down Experiment
The pull-down experiment was performed as described by Sun et al. [37] with some modifications. Glutathione S-transferase (GST)-tagged FilZ, GST-tagged FilZ-R13A, and the His-tagged A2230 proteins were expressed in E. coli BL21 (DE3) and purified. For the interaction of FilZ (or FilZ-R13A) with A2230, the 2 proteins (0.5 mg each) were mixed and incubated at 4 • C for 1 h. For the interaction of FilZ-c-di-GMP with A2230, 0.5 mg FilZ was incubated with a 10 µM c-di-GMP solution at 4 • C for 1 h. Then, 0.5 mg A2230 was added, followed by further incubation at 4 • C for 1 h. For the interaction of the FilZ-A2230 complex with c-di-GMP, a mixture of 0.5 mg FilZ and 0.5 mg A2230 was incubated at 4 • C for 1 h. Then, 10 µM c-di-GMP was added, followed by further incubation at 4 • C for 1 h. After incubation, the samples were added to a glutathione sepharose 4B (GE healthcare, Amersham, UK) column pre-treated with 40 mM phosphate buffer (pH 7.5) containing 100 mM NaCl. The proteins were eluted with the phosphate buffer containing 20 mM reduced glutathione. The eluted proteins were detected by immunoblotting.

Immunoblot
The proteins FilZ and His-tagged A2230 were separated by SDS-PAGE, and the Histagged A2230 in the electrophoresis gel was transferred (350 mA, 150 min) to the ECL membrane for immunoblotting as described previously [38,39]. The membrane was incubated with TBST buffer containing the primary antibody (Anti 6× His tag antibody, Abcam, Cambridge, UK) at 4 • C overnight. After treatment with the secondary antibody (Sheep anti-rabbit IgG H&L, Abcam, UK), the protein bands on the membrane were developed in a myECL imager (ThermoFisher Scientific, Waltham, MA, USA).

Chemotaxis Assays
Chemotaxis of strain SM9913 and its gene PSM_A2230 deleted derivative were tested by capillary assay with some modifications [40]. Sucrose (1.0 g/L) or casein (2.0 g/L) was used as a chemoattractant. Glucose (2.0 g/L) was used as the positive control. The cells in the capillaries were incubated at 15 • C for 48-60 h. After 30 min incubation at 15 • C, the cells in the capillaries were serially diluted and plated onto nutrient agar. The plates were incubated at 15 • C for 48-60 h, and the number of colonies was counted. Capillaries containing buffer alone were used as the negative control.

Effect of Protein FilZ on the Swarming of Strain SM9913
To determine the timing of the production of lateral flagella, SM9913 cells were inoculated onto a swarm plate (0.5% agar) from liquid medium. Cells at the edge of the colony were observed with an atomic force microscope from 24 h to 72 h. Most of the observed cells had a polar flagellum at all times. Lateral flagella were observed starting at approximately 50 h (Figures 1 and S1). The swarming process can be divided into three periods: the initial period before the colony expanded (0-48 h); the rapid swarming period, during which the swarming rate kept increasing (48-54 h); and the slow swarming period, during which the swarming rate decreased steadily (54-72 h) (Figures 1 and 2b).
An uncharacterized filZ gene is located in the lateral flagellar gene cluster (Figure 2a). The transcription of filZ was upregulated during the rapid swarming period (Figure 2c), suggesting that the filZ gene product may have a role in swarming motility. To determine how FilZ was involved in swarming, we constructed a ∆filZ deletion mutant. The growth in liquid and the temporal order of flagellar production during the swarming process in the ∆filZ mutant were similar to those of WT SM9913 (Figures 1 and S2). However, the expansion pattern of the ∆filZ colony was quite different. Compared to WT SM9913, the ∆filZ colony expanded much more rapidly in the initial period, leading to a precocious swarming phenotype, and then steadily decreased (Figures 1 and 2b). During the rapid swarming period for WT SM9913, the rate of colony expansion of ∆filZ steadily decreased, but that of WT SM9913 kept increasing (Figure 2b). These results suggest that the filZ gene product facilitates the swarming motility of strain SM9913 in the 48-54 h rapid period. Microorganisms 2023, 11, x FOR PEER REVIEW 6 of 16 An uncharacterized filZ gene is located in the lateral flagellar gene cluster ( Figure 2a). The transcription of filZ was upregulated during the rapid swarming period (Figure 2c), suggesting that the filZ gene product may have a role in swarming motility. To determine how FilZ was involved in swarming, we constructed a ΔfilZ deletion mutant. The growth in liquid and the temporal order of flagellar production during the swarming process in the ΔfilZ mutant were similar to those of WT SM9913 (Figures 1 and S2). However, the expansion pattern of the ΔfilZ colony was quite different. Compared to WT SM9913, the ΔfilZ colony expanded much more rapidly in the initial period, leading to a precocious swarming phenotype, and then steadily decreased (Figures 1 and 2b). During the rapid swarming period for WT SM9913, the rate of colony expansion of ΔfilZ steadily decreased, but that of WT SM9913 kept increasing ( Figure 2b). These results suggest that the filZ gene product facilitates the swarming motility of strain SM9913 in the 48-54 h rapid period.  An uncharacterized filZ gene is located in the lateral flagellar gene cluster ( Figure 2a). The transcription of filZ was upregulated during the rapid swarming period (Figure 2c), suggesting that the filZ gene product may have a role in swarming motility. To determine how FilZ was involved in swarming, we constructed a ΔfilZ deletion mutant. The growth in liquid and the temporal order of flagellar production during the swarming process in the ΔfilZ mutant were similar to those of WT SM9913 (Figures 1 and S2). However, the expansion pattern of the ΔfilZ colony was quite different. Compared to WT SM9913, the ΔfilZ colony expanded much more rapidly in the initial period, leading to a precocious swarming phenotype, and then steadily decreased (Figures 1 and 2b). During the rapid swarming period for WT SM9913, the rate of colony expansion of ΔfilZ steadily decreased, but that of WT SM9913 kept increasing ( Figure 2b). These results suggest that the filZ gene product facilitates the swarming motility of strain SM9913 in the 48-54 h rapid period. Cells used to analyze the transcriptional level were collected from the edge of the swarming colony at different times. The fold change in the transcription of filZ was calculated relative to that of the endogenous control gene rpoD (the RNA polymerase sigma factor gene). The graph shows data from triplicate experiments (mean ± SD).

FilZ Has a Single PilZ Domain That Binds c-di-GMP
Sequence analysis indicated that FilZ contains a single PilZ domain (Figures 3a and S3). PilZ domains are known to bind c-di-GMP [17]. We tested the interaction between the recombinant FilZ protein and c-di-GMP using isothermal titration calorimetry (ITC). FilZ displayed a strong c-di-GMP binding ability, with a K d value of 230 nM (Figure 3b,c). This property suggests that FilZ may function as a c-di-GMP effector. To determine the key amino acids involved in binding c-di-GMP, we predicted the tertiary structure of FilZ bound to c-di-GMP using I-Tasser. FilZ consists of an N-terminal β-barrel and a flexible C-terminal α-helix, highly similar to the structure of the C-terminal PilZ domain of YcgR from E. coli (YcgR-PilZ, PDB ID 5Y6F) ( Figure S4). The N-terminal β-barrel contains two motifs, 9 RXXXR 13 and 53 (D/N)XSXXG 58 , that are conserved in PilZ domains [41]. Based on the predicted structure of the complex, c-di-GMP probably interacts with residues R9, R13, K52, D53, G58, F99, and G101 ( Figure S4). To determine the importance of these residues in binding c-di-GMP, we used site-directed mutagenesis to insert alanine at all of these positions. Mutants R9A, R13A, D53A, G58A, and G101A all completely lost the ability to bind c-di-GMP (Figures 3c and S5), indicating that R9, R13, D53, G58, and G101 are important for binding c-di-GMP. These residues are conserved in other PilZ-domain proteins (Figure 4). FilZ-R13A was chosen as a representative for further investigation.
the ΔfilZ mutant, these periods are advanced by about 5 h. (c) The transcription of filZ during swarming. Cells used to analyze the transcriptional level were collected from the edge of the swarming colony at different times. The fold change in the transcription of filZ was calculated relative to that of the endogenous control gene rpoD (the RNA polymerase sigma factor gene). The graph shows data from triplicate experiments (mean ± SD).

FilZ Has a Single PilZ Domain That Binds c-di-GMP
Sequence analysis indicated that FilZ contains a single PilZ domain (Figures 3a and  S3). PilZ domains are known to bind c-di-GMP [17]. We tested the interaction between the recombinant FilZ protein and c-di-GMP using isothermal titration calorimetry (ITC). FilZ displayed a strong c-di-GMP binding ability, with a Kd value of 230 nM (Figure 3b,c). This property suggests that FilZ may function as a c-di-GMP effector. To determine the key amino acids involved in binding c-di-GMP, we predicted the tertiary structure of FilZ bound to c-di-GMP using I-Tasser. FilZ consists of an N-terminal β-barrel and a flexible C-terminal α-helix, highly similar to the structure of the C-terminal PilZ domain of YcgR from E. coli (YcgR-PilZ, PDB ID 5Y6F) ( Figure S4). The N-terminal β-barrel contains two motifs, 9 RXXXR 13 and 53 (D/N)XSXXG 58 , that are conserved in PilZ domains [41]. Based on the predicted structure of the complex, c-di-GMP probably interacts with residues R9, R13, K52, D53, G58, F99, and G101 ( Figure S4). To determine the importance of these residues in binding c-di-GMP, we used site-directed mutagenesis to insert alanine at all of these positions. Mutants R9A, R13A, D53A, G58A, and G101A all completely lost the ability to bind c-di-GMP (Figures 3c and S5), indicating that R9, R13, D53, G58, and G101 are important for binding c-di-GMP. These residues are conserved in other PilZ-domain proteins ( Figure 4). FilZ-R13A was chosen as a representative for further investigation.   The key residues R9, R13, D53, G58, and G101 of FilZ in binding c-di-GMP are highlighted by red asterisks. The most identical residues are highlighted in a black background, followed by a pink and blue background separately.

FilZ Activity Is Negatively Controlled by c-di-GMP In Vivo
We began by determining the intracellular concentration of c-di-GMP during the swarming process. The average intracellular concentration of c-di-GMP increased continuously ( Figure 5a). However, in the rapid swarming period of 48-54 h, the intracellular cdi-GMP concentration was lower than the Kd value of FilZ for c-di-GMP.
To study the interaction between FilZ and c-di-GMP in vivo, we constructed two complementary strains: ΔfilZ(pEV filZ ), which has inducible expression of FilZ, and ΔfilZ (pEV filZ-R13A ) which has inducible expression of FilZ-R13A. The growth of WT SM9913 was decreased by the maintenance of the plasmid pEV ( Figures S2 and S6); thus, all the strains containing the pEV plasmid expanded slower on the swarming plates than those without the plasmid. In contrast to strain ΔfilZ(pEV), both the complementary strains displayed similar increasing swarm expansion rates before 54 h to strain 9913(pEV) (Figure 5b,c). Because the mutant FliZ-R13A had no ability to bind c-di-GMP, this result suggests that FilZ likely functions without binding c-di-GMP in the rapid swarming period. In addition, compared with ΔfilZ(pEV filZ ) and 9913(pEV), ΔfilZ(pEV filZ-R13A ) showed a slower decline in The key residues R9, R13, D53, G58, and G101 of FilZ in binding c-di-GMP are highlighted by red asterisks. The most identical residues are highlighted in a black background, followed by a pink and blue background separately.

FilZ Activity Is Negatively Controlled by c-di-GMP In Vivo
We began by determining the intracellular concentration of c-di-GMP during the swarming process. The average intracellular concentration of c-di-GMP increased continuously ( Figure 5a). However, in the rapid swarming period of 48-54 h, the intracellular c-di-GMP concentration was lower than the K d value of FilZ for c-di-GMP.
To study the interaction between FilZ and c-di-GMP in vivo, we constructed two complementary strains: ∆filZ(pEV filZ ), which has inducible expression of FilZ, and ∆filZ (pEV filZ-R13A ) which has inducible expression of FilZ-R13A. The growth of WT SM9913 was decreased by the maintenance of the plasmid pEV ( Figures S2 and S6); thus, all the strains containing the pEV plasmid expanded slower on the swarming plates than those without the plasmid. In contrast to strain ∆filZ(pEV), both the complementary strains displayed similar increasing swarm expansion rates before 54 h to strain 9913(pEV) (Figure 5b,c). Because the mutant FliZ-R13A had no ability to bind c-di-GMP, this result suggests that FilZ likely functions without binding c-di-GMP in the rapid swarming period. In addition, compared with ∆filZ(pEV filZ ) and 9913(pEV), ∆filZ(pEV filZ-R13A ) showed a slower decline in swarming rate in the slow swarming period, in which the intracellular c-di-GMP concentrations were higher than the K d of FilZ for c-di-GMP (Figure 5b,c). Taken together, these results suggest that FilZ facilitates the swarming motility of strain SM9913 when it is not bound to c-di-GMP but is inactivated when bound to c-di-GMP. Thus, it seems that higher intracellular c-di-GMP concentrations negatively regulate FilZ function to inhibit swarming. swarming rate in the slow swarming period, in which the intracellular c-di-GMP concentrations were higher than the Kd of FilZ for c-di-GMP (Figure 5b,c). Taken together, these results suggest that FilZ facilitates the swarming motility of strain SM9913 when it is not bound to c-di-GMP but is inactivated when bound to c-di-GMP. Thus, it seems that higher intracellular c-di-GMP concentrations negatively regulate FilZ function to inhibit swarming.

FilZ Interacts with the CheW-like Protein A2230
To identify the target protein(s) with which FilZ interacted, we performed a screen using a bacterial two-hybrid assay with FilZ-R13A as the bait. Among the proteins expressed in strain SM9913, only A2230 was verified to interact with FilZ-R13A (Figure 6a). Gene PSM_A2230 is located in the polar flagellar gene cluster. It displayed a stable transcription level during swarming (Figure 6b). Sequence analysis suggested that A2230 is a CheW-like protein, with a high sequence identity (69%) to protein CheW from P. aeruginosa ( Figure S7). Furthermore, a capillary assay to test the chemotaxis behavior of strain SM9913 and its Δ2230 mutant showed that the mutant accumulated only 50% as many cells as WT SM9913. Thus, the A2230 protein contributes to chemotaxis mediated by the polar flagellum.

FilZ Interacts with the CheW-like Protein A2230
To identify the target protein(s) with which FilZ interacted, we performed a screen using a bacterial two-hybrid assay with FilZ-R13A as the bait. Among the proteins expressed in strain SM9913, only A2230 was verified to interact with FilZ-R13A (Figure 6a). Gene PSM_A2230 is located in the polar flagellar gene cluster. It displayed a stable transcription level during swarming (Figure 6b). Sequence analysis suggested that A2230 is a CheW-like protein, with a high sequence identity (69%) to protein CheW from P. aeruginosa ( Figure S7). Furthermore, a capillary assay to test the chemotaxis behavior of strain SM9913 and its ∆2230 mutant showed that the mutant accumulated only 50% as many cells as WT SM9913. Thus, the A2230 protein contributes to chemotaxis mediated by the polar flagellum.
trations were higher than the Kd of FilZ for c-di-GMP (Figure 5b,c). Taken together, these results suggest that FilZ facilitates the swarming motility of strain SM9913 when it is not bound to c-di-GMP but is inactivated when bound to c-di-GMP. Thus, it seems that higher intracellular c-di-GMP concentrations negatively regulate FilZ function to inhibit swarming.

FilZ Interacts with the CheW-like Protein A2230
To identify the target protein(s) with which FilZ interacted, we performed a screen using a bacterial two-hybrid assay with FilZ-R13A as the bait. Among the proteins expressed in strain SM9913, only A2230 was verified to interact with FilZ-R13A (Figure 6a). Gene PSM_A2230 is located in the polar flagellar gene cluster. It displayed a stable transcription level during swarming (Figure 6b). Sequence analysis suggested that A2230 is a CheW-like protein, with a high sequence identity (69%) to protein CheW from P. aeruginosa ( Figure S7). Furthermore, a capillary assay to test the chemotaxis behavior of strain SM9913 and its Δ2230 mutant showed that the mutant accumulated only 50% as many cells as WT SM9913. Thus, the A2230 protein contributes to chemotaxis mediated by the polar flagellum.  To determine the relationships of c-di-GMP and proteins FilZ and A2230, a pull-down experiment was performed. FilZ interacted with A2230 in the absence of c-di-GMP (Lanes 2 to 4 in Figure 6c); however, upon binding c-di-GMP, FilZ could no longer interact with A2230 (Lane 5 in Figures 6c and S8). Thus, in SM9913 swarming cells, FilZ likely exerts its functions through interacting with A2230 when not binding c-di-GMP.
Previous work has shown that CheW is part of the flagella-mediated chemotaxis signal transduction system that consists of the chemotaxis proteins CheW, CheA, and CheY. Phosphorylated CheY acts on the flagellar motor protein FliM to regulate flagellar motility [42,43]. Sequence analysis showed that genes PSM_A2236 and PSM_A2238 located in the polar flagellar gene cluster encode proteins CheA and CheY, respectively (Figure 7a). To determine whether the CheW/CheA/CheY signal transduction pathway interacts with the polar flagellar motor protein FliM p or the lateral flagellar motor protein FliM, we performed two-hybrid experiments using five reporter plasmids: pKT25-cheA, pKT25-fliM, pKT25-fliM p , pUT18C-A2230, and pUT18C-cheY. The results show that protein A2230 (CheW) interacted with A2236 (CheA), and that protein A2238 (CheY) interacted with A2236 and FliM p . Thus, all three proteins may be involved in a chemotaxis signaling pathway directed at FliM p rather than FliM (Figure 7b,c). Together with the finding that FilZ interacted with A2230, it seems likely that FilZ influences swarming indirectly through an effect on polar flagellar motility mediated by the chemotaxis signaling pathway.
in the assay. The interaction was monitored by determining the amount of His-tagged A2230 w His tag antibody by immunoblotting of each mixture.
To determine the relationships of c-di-GMP and proteins FilZ and A2230, a p down experiment was performed. FilZ interacted with A2230 in the absence of c-di-G (Lanes 2 to 4 in Figure 6c); however, upon binding c-di-GMP, FilZ could no longer inte with A2230 (Lane 5 in Figures 6c and S8). Thus, in SM9913 swarming cells, FilZ lik exerts its functions through interacting with A2230 when not binding c-di-GMP.
Previous work has shown that CheW is part of the flagella-mediated chemotaxis nal transduction system that consists of the chemotaxis proteins CheW, CheA, and Ch Phosphorylated CheY acts on the flagellar motor protein FliM to regulate flagellar moti [42,43]. Sequence analysis showed that genes PSM_A2236 and PSM_A2238 located in polar flagellar gene cluster encode proteins CheA and CheY, respectively (Figure 7a) determine whether the CheW/CheA/CheY signal transduction pathway interacts with polar flagellar motor protein FliMp or the lateral flagellar motor protein FliM, we p formed two-hybrid experiments using five reporter plasmids: pKT25-cheA, pKT25-fl pKT25-fliMp, pUT18C-A2230, and pUT18C-cheY. The results show that protein A2 (CheW) interacted with A2236 (CheA), and that protein A2238 (CheY) interacted w A2236 and FliMp. Thus, all three proteins may be involved in a chemotaxis signaling pa way directed at FliMp rather than FliM (Figure 7b,c). Together with the finding that F interacted with A2230, it seems likely that FilZ influences swarming indirectly through effect on polar flagellar motility mediated by the chemotaxis signaling pathway.

FilZ Interferes with the Polar Flagellar Motility
To explore the effect of FilZ on polar flagellar motility further, we constructed the mutant ∆0915, which could not produce the lateral flagella. Protein A0915 is a homolog of LafK, which is the master regulator of the expression of lateral flagella in Vibrio parahaemolyticus ( Figure S9) [14]. As expected, protein A0915 was needed for the expression of most of the genes in the lateral flagellar cluster, and the ∆0915 mutant produced only a polar flagellum and lost swarming motility (Figure 8a). We also constructed strain ∆0915(pEV filZ ) and strain ∆0915(pEV filZ-R13A ) and observed their motility in soft (0.3%) agar. Compared with strains SM9913 (pEV) and ∆0915(pEV), which formed similar spreading colonies, both strains ∆0915(pEV filZ ) and ∆0915(pEV filZ-R13A ) showed impaired spreading (Figure 8b). This result suggests that FilZ interferes with the polar flagellum when it is not bound to c-di-GMP. Presumably, expression of the chromosomal filZ gene is not induced in the free-swimming SM9913 (pEV) and ∆0915(pEV) cells. (CheY) and the lateral flagellar motor protein FliM. The white color indicates that A2238 (CheY) does not interact with FliM.

FilZ Interferes with the Polar Flagellar Motility
To explore the effect of FilZ on polar flagellar motility further, we constructed the mutant Δ0915, which could not produce the lateral flagella. Protein A0915 is a homolog of LafK, which is the master regulator of the expression of lateral flagella in Vibrio parahaemolyticus ( Figure S9) [14]. As expected, protein A0915 was needed for the expression of most of the genes in the lateral flagellar cluster, and the Δ0915 mutant produced only a polar flagellum and lost swarming motility (Figure 8a). We also constructed strain Δ0915(pEV filZ ) and strain Δ0915(pEV filZ-R13A ) and observed their motility in soft (0.3%) agar. Compared with strains SM9913 (pEV) and Δ0915(pEV), which formed similar spreading colonies, both strains Δ0915(pEV filZ ) and Δ0915(pEV filZ-R13A ) showed impaired spreading (Figure 8b). This result suggests that FilZ interferes with the polar flagellum when it is not bound to c-di-GMP. Presumably, expression of the chromosomal filZ gene is not induced in the free-swimming SM9913 (pEV) and Δ0915(pEV) cells.

The Phylogenetic Distribution of FilZ
To learn how prevalent this mode of regulation of swarming is among bacteria, we searched the non-redundant protein database in NCBI. Through sequence alignment and phylogenetic relationship analysis, we found that all bacterial species with FilZ homologs belong to the γ-proteobacteria and that more than half are marine bacteria (Figure 9a and Table S3). Furthermore, part of the lateral flagellar gene cluster, including filZ and another five or six adjacent genes, was present in many marine bacteria possessing dual flagellar systems. These include the genera Pseudoalteromonas, Aeromonas, Shewanella, and others ( Figure 9b). Thus, FilZ-like modulation of swarming may be a common strategy adopted by benthic marine bacteria.

The Phylogenetic Distribution of FilZ
To learn how prevalent this mode of regulation of swarming is among bacteria, we searched the non-redundant protein database in NCBI. Through sequence alignment and phylogenetic relationship analysis, we found that all bacterial species with FilZ homologs belong to the γ-proteobacteria and that more than half are marine bacteria (Figure 9a and Table S3). Furthermore, part of the lateral flagellar gene cluster, including filZ and another five or six adjacent genes, was present in many marine bacteria possessing dual flagellar systems. These include the genera Pseudoalteromonas, Aeromonas, Shewanella, and others (Figure 9b). Thus, FilZ-like modulation of swarming may be a common strategy adopted by benthic marine bacteria.

Discussion
In this study, FilZ, a protein encoded by the filZ gene in the lateral flagellar gene cluster of strain SM9913, was shown to be a c-di-GMP effector that facilitated swarming motility. FilZ consists largely of a single PilZ domain that is responsible for binding c-di-GMP. When not bound to c-di-GMP, FilZ interacted with the CheW homolog A2230, which was part of the chemotaxis signal transduction pathway for the polar flagellum. The R13A variant of FilZ, which could not bind c-di-GMP and thus was able to bind A2230 even in the presence of high levels of c-di-GMP, was constitutively active as a negative regulator of the polar flagellum. Thus, high concentrations of c-di-GMP inhibited swarming because they inactivated a negative regulator of chemotactic control of the polar flagellum. The results indicate that FilZ exerts its function on the abnormally short polar flagellum of the WT SM9913 strain in swarming. In addition, although a previous study showed that the length of the polar flagellum of WT SM9913 has only a small effect on swimming motility [27], the function of FilZ may be different in the strain harboring a normal long polar flagellum, which still awaits further study.
Other PilZ-domain proteins have been reported to be involved in bacterial motility,

Discussion
In this study, FilZ, a protein encoded by the filZ gene in the lateral flagellar gene cluster of strain SM9913, was shown to be a c-di-GMP effector that facilitated swarming motility. FilZ consists largely of a single PilZ domain that is responsible for binding c-di-GMP. When not bound to c-di-GMP, FilZ interacted with the CheW homolog A2230, which was part of the chemotaxis signal transduction pathway for the polar flagellum. The R13A variant of FilZ, which could not bind c-di-GMP and thus was able to bind A2230 even in the presence of high levels of c-di-GMP, was constitutively active as a negative regulator of the polar flagellum. Thus, high concentrations of c-di-GMP inhibited swarming because they inactivated a negative regulator of chemotactic control of the polar flagellum. The results indicate that FilZ exerts its function on the abnormally short polar flagellum of the WT SM9913 strain in swarming. In addition, although a previous study showed that the length of the polar flagellum of WT SM9913 has only a small effect on swimming motility [27], the function of FilZ may be different in the strain harboring a normal long polar flagellum, which still awaits further study.
Other PilZ-domain proteins have been reported to be involved in bacterial motility, including YcgR [18], MotI [19], FlgZ [20], PlzD [21], and MotL [22]. These proteins have a negative regulatory effect on swimming or swarming. For example, YcgR in E. coli interferes with swarming motility via a "backstop brake" mechanism; after binding c-di-GMP, YcgR interacts with the motor proteins MotA and FliG to reduce the efficiency of torque generation on the peritrichous flagella [18,41]. In contrast to YcgR, which affects the function of both polar and lateral flagella in various species, the single-PilZ-domain protein MotL from the lateral flagellar system of Shewanella putrefaciens appears to interact directly with components of the lateral flagellar motors to inhibit the function of lateral flagella when it is bound to c-di-GMP [22]. Compared to these proteins, FilZ has some distinct characteristics. First, its activity is negatively controlled by c-di-GMP. FilZ of strain SM9913 promotes swarming when it is not bound to c-di-GMP. Upon binding c-di-GMP, the function of FilZ may be blocked. Second, although encoded by the lateral flagellar gene cluster, FilZ exerts its effect by interfering with polar flagellar motility to facilitate swarming motility indirectly. This is a direct example of an interplay between the polar and lateral flagellar systems in a bacterium with dual flagellar systems.
Somewhat paradoxically, a ∆filZ mutant showed a precocious swarming phenotype. Precocious swarming has also been observed in bacteria with a single flagellar system, such as Proteus mirabilis [44,45], Salmonella enterica serovar Typhimurium [46], and Serratia marcescens [47][48][49][50]. This phenotype is mediated by the master flagellar regulator FlhDC. However, little is known concerning the regulation of the precocious swarming phenotype in bacteria with dual flagellar systems. Sequence analysis suggested that there are no FlhDC homologs in strain SM9913. Thus, we assume that the precocious swarming phenotype of the ∆filZ mutant may have a very different cause. As FilZ interfered with function of the polar flagellum in strain SM9913, the precocious swarming phenotype of ∆filZ may be driven by the unaffected polar flagellum. Clearly, the mechanism underlying precocious swarming in strain SM9913 requires further study.
Chemotaxis signal transduction has been well-studied in many flagellated bacteria. It influences both swimming and swarming motilities by regulating the direction of flagellar rotation [5,51]. The CheA histidine kinase and the CheY regulator of flagellar switching are central to chemotaxis signal transduction. When coupled with CheW and a chemoreceptor, CheA is activated and phosphorylates CheY. For polar flagella, which can rotate both clockwise and counter-clockwise, phospho-CheY interacts with the FliM protein in the flagellar motor to induce clockwise flagellar rotation, leading to a change in the swimming direction [52,53]. For lateral flagella, which only rotate counter-clockwise, especially from bacteria possessing dual flagellar systems such as V. alginolyticus, this binding slows down the lateral flagella-driven motility [53]. In strain SM9913, we found a complete set of genes encoding the chemotaxis signaling system in the polar flagellar gene cluster, including genes PSM_A2230 (cheW), PSM_A2236 (cheA), and PSM_A2238 (cheY). Pull-down assays demonstrated that proteins FilZ and A2230 interacted, and two-hybrid assays established the link between the proteins A2230, A2236, and A2238 and the FliM p protein of the polar flagellar motor. These results suggest that FilZ may affect the polar flagellum via this chemotaxis signal pathway to promote swarming in strain SM9913. The negative effect of FliZ on the function of the polar flagellum was demonstrated by the impaired spreading of cells with overproduced, plasmid-encoded FilZ or FilZ-R13A in 0.3% agar. The cells move through this agar by swimming driven by the polar flagellum. Cells that cannot reverse the direction of flagellar rotation are defective in chemotaxis and therefore not impaired in spreading [53]. Wild-type cells swam under these conditions because expression of the chromosomal filZ gene was not induced.
In summary, the previously uncharacterized FilZ protein, which has a single PilZ domain that binds c-di-GMP, was found to facilitate surface motility in the rapid swarming period of the deep-sea sedimentary bacterium Pseudoalteromonas sp. SM9913. FilZ binds to the CheW homolog A2230, thereby impairing the function of the polar flagellum. FilZ is unable to bind A2230 when it is bound to c-di-GMP, meaning that c-d-GMP is a negative regulator that inhibits swarming indirectly. FilZ is encoded by a gene in the cluster that encodes the components of the lateral flagella that propel surface swarming. Our study identifies a previously unknown interplay between the polar and lateral flagella and provides new insights into the regulation of motility in bacteria with dual flagellar systems.