PP4397/FlgZ provides the link between PP2258 c-di-GMP signalling and altered motility in Pseudomonas putida

Bacteria swim and swarm using rotating flagella that are driven by a membrane-spanning motor complex. Performance of the flagella motility apparatus is modulated by the chemosensory signal transduction system to allow navigation through physico-chemical gradients – a process that can be fine-tuned by the bacterial second messenger c-di-GMP. We have previously analysed the Pseudomonas putida signalling protein PP2258 that has the capacity to both synthesize and degrade c-di-GMP. A PP2258 null mutant displays reduced motility, implicating the c-di-GMP signal originating from this protein in control of P. putida motility. In Escherichia coli and Salmonella, the PilZ-domain protein YcgR mediates c-di-GMP responsive control of motility through interaction with the flagellar motors. Here we provide genetic evidence that the P. putida protein PP4397 (also known as FlgZ), despite low sequence homology and a different genomic context to YcgR, functions as a c-di-GMP responsive link between the signal arising from PP2258 and alterations in swimming and swarming motility in P. putida.


Results and Discussion
pp4397/flgZ is co-transcribed with flagellar associated genes dependent on σ 54 and σ FliA . In silico searches of genome sequenced P. putida KT2440 identified PP4397/FlgZ as the protein exhibiting highest identity to YcgR (17% and 24% identity with E. coli and Salmonella, respectively). Despite this low homology, similar to YcgR, PP4397/FlgZ also possesses a type I c-di-GMP binding PilZ and PilZN-like domain (Figs 1 and S1). The assembly and operation of Pseudomonad flagella depends on more than 50 genes arranged in clusters that are controlled by a four-tiered transcriptional regulatory cascade 26 . As depicted in Fig. 2A (upper), enteric ycgR lies in a monocistronic operon, while the P. putida flgZ/pp4397 gene lies within a cluster of genes encoding proteins involved in flagellar biosynthesis ( Fig. 2A, lower). We therefore wanted to clarify if pp4397 is co-transcribed with any of these genes.
Reverse transcription polymerase chain reaction (PCR) assays with RNA isolated from P. putida KT2440 and primer pairs spanning pp4397 and adjacent genes showed that pp4397 is co-transcribed with the σ FliA anti-σ-factor gene flgM and pp4396 -a gene encoding an FlgN family protein (Fig. 2B). An appropriately located σ FliA -dependent promoter is found immediately upstream of the flgM-flgN-flgZ/pp4397 genes cluster (see Fig. 2A). Hence, this data is consistent with the idea that pp4397 lies within a tri-cistronic flgM-flgN-pp4397 operon, but does not exclude the possibility that internal promoters within flgM and/or flgN may also contribute to transcription of pp4397. Recent analysis of transcription of pp4397/flgZ counterparts in P. fluorescens F113 and P. putida KT2440 could only document a partial dependence on σ FliA as assessed using FliA null strains 10,24 . Therefore, we extended the analysis to include the upstream flgA gene that encodes a protein involved in flagellar P-ring formation. As shown in Fig. 2B, transcription of the flgM-flgN-pp4397 operon also appears to be mediated by read-through transcription from a σ 54 -dependent promoter located upstream of the flgA gene.
To further substantiate the above findings, we performed in vivo and in vitro transcription assays. In vivo transcription of pp4397/flgZ was monitored using a luxAB (luciferase) transcriptional fusion generated downstream The amino acid sequence of PP4397 is shown with nine highly conserved residues of PilZ domains (highlighted in bold and underlined) that encompass the RXXXR and (D/N)XSXXG motifs of c-di-GMP binding-proficient type I PilZ domains 34 . Residues, which when substituted by alanine, essentially abolish c-di-GMP binding by PP4397/FlgZ 25 are highlighted in grey. A complete alignment of PP4397 and YcgR as in 15,25 using ESPript 35 is shown in Fig. S1, while the extensive homology with FlgZ proteins of representative Pseudomonads is shown in Fig. S2.
ScIentIfIc REPORts | (2018) 8:12205 | DOI:10.1038/s41598-018-29785-w of pp4397 in wild-type, FliA null, and RpoN (σ 54 ) null P. putida backgrounds. Consistent with co-dependence on both these σ-factors, transcriptional output was decreased but not abolished in both of the null strains as compared to wild-type when grown Luria-Bertani (LB) broth (Fig. 2C). Functionality of the identified σ FliA promoter located upstream of the flgM-flgN-pp4397 tri-gene cluster (P flgM ) was verified by single-round in vitro transcription assays with σ FliA -RNA polymerase reconstituted from purified P. putida components (Fig. 2D). As anticipated by its high identity to the optimal consensus for P. putida FliA dependent transcription, the near consensus flgM promoter produces high levels of transcripts as compared to the previously analysed suboptimal σ FliA -dependent promoter for aer2 27 .
The difference in σ-dependence for the flgA promoters in Pseudomonads as compared to enterics is due to differences in the hierarchical expression of flagellar genes in these two species. Transcription of genes needed early, e.g. flgA, are dependent on σ 54 in Pseudomonas and σ 70 in E. coli and Salmonella 26,28 . Thus, while the genomic context of pp4397/flgZ is different from the moncistronic context of E. coli ycgR gene, transcriptional control of the flgA and flgMN counterparts is conceptually similar, with a promoter upstream of flgA generating read-through transcription of downstream genes within a σ FliA -dependent operon. Given that the flgZ gene is highly conserved  Table S1). Graphed values are the average +/− standard deviation of six independent determinations from cultures grown to the stationary phase in LB (OD 600 ~5.0 for wild-type and FliA null; ~2.1 for RpoN null). (D) Single-round in vitro transcription assays using 10 nM supercoiled DNA templates harbouring the σ FliA -dependent P. putida P aer2 promoter (4; pVI1011) or P flgM (5; pVI2368) in the presence of 10 nM σ FliA -RNAP. Inset shows images from one of two independent experiments used to obtain the graphed average values (P flgM upper; P aer2 lower). A comparison of the P aer2 and P flgM promoter sequences with the optimal consensus 27 for P. putida σ FliA is shown to the right.
ScIentIfIc REPORts | (2018) 8:12205 | DOI:10.1038/s41598-018-29785-w in sequence and synteny in all sequenced Pseudomonads 24 , including P. fluorescens F113, P. aeruginosa PA14, and P. aeruginosa PAO1 (Fig. S2), co-dependence on both σ FliA and σ 54 as found here for P. putida KT2440 is likely the case for these and other Pseudomonads. It is interesting to note that this regulatory arrangement would ensure σ 54 -dependent transcription of the flgM-flgN-pp4397 genes even in the absence of σ FliA . Therefore, it would also result in σ FliA -independent production of PP4397/FlgZ that serves to slow-down flagella (as detailed below) and production of the anti-σ FliA -factor FlgM to block new de novo flagella production -two steps needed for preparation to enter the biofilm mode of growth 9 .
Lack of PP4397 results in altered swimming and swarming motility. Having established that PP4397 is co-ordinately regulated with genes of the flagella regulon, we next addressed its involvement in c-di-GMP responsive control of flagella-mediated motility. To test potential involvement of PP4397 in motility control, we first generated a null mutant of P. putida KT2701 (a streptomycin resistant derivative of the genome sequenced P. putida KT2440) in which the majority of the pp4397/flgZ gene was replaced by a tetracycline resistance cassette. When tested on rich (LB) 0.3% soft agar swimming motility plates or 0.5% agar swarming motility plates, the Δpp4397::Tc strain displayed only a slightly enhanced motility phenotype as compared to the wild type (Fig. 3).
The above data is consistent with previous findings for E. coli and Salmonella in which a YcgR null counterparts only exhibited increased motility in a high c-di-GMP background -a condition that normally results in decreased motility 20 . Therefore, we also introduced the Δpp4397::Tc allele in P. putida KT2701 Δpp2258::Km, a previously analysed strain known to have elevated c-di-GMP levels and a swimming motility defect on soft agar plates despite having a wild type number of polar flagella 12 .
Both the Δpp2258::Km and the Δpp2258/Δpp4397 double mutant strains exhibit prolonged lag phases upon outgrowth from overnight cultures on rich media (LB), but once they attain exponential growth, have doubling times (41.3 +/− 2.9 min) similar to, but slower, than the wild type and the Δpp4397::Tc strains (36.2 +/− 1.7 min; see Fig. S3A). While the exact level of c-di-GMP in the Δpp2258::Km PP2258 null strain is unknown, elevated c-di-GMP levels in this strain and the Δpp2258/Δpp4397 double null strain are insufficient to provoke altered biofilm production or dispersal phenotypes ( Fig. S3B-D), as judged using a microtitre dish-based assays that employs serial dilution to recapitulate biofilm growth and dispersal kinetics 29 . As detailed in Fig. S3, in both cases biofilm production and dispersal rates appear similar to wild type, despite a delay as a consequence of growth kinetics.
In contrast, the reduced motility seen for the Δpp2258::Km strain in both swimming and swarming abilities was significantly rescued in the double mutant [compare PP2258 null with the PP2258/PP4397 double null in Fig. 3A,B]. Even though exponentially growing cells were used for inoculation of the motility assay plates (see Methods) reduced growth rates as a consequence of elevated c-di-GMP levels probably, at least in part, underlies why full motility comparable to the wild-type strain could not be achieved. Taken together, the data in Figs 3 and S3 consolidate a role for PP4397/FlgZ of P. putida in swimming and swarming motility (but not biofilm production or dispersal), and provide the first evidence that PP4397/FlgZ functions downstream of PP2258 in response to modulation of c-di-GMP levels.
The c-di-GMP binding property of PP4397 is required to mediate motility control. To verify that the phenotype for the Δpp2258/Δpp4397 double null strain was not attributable to indirect effects on upstream genes within the flgM-flgN-flgZ/pp4397 operon, this strain was complemented with plasmids carrying either a native version of the pp4397/flgZ gene or a C-terminally FLAG-tagged version under control of the IPTG inducible lacI Q /P tac promoter. The lacI Q /P tac system of the expression plasmid is leaky and produced sufficient PP4397/FlgZ to reverse the effect  (Table S1) on 0.3% soft agar LB plates. Representative swim rings are shown above. (B) Relative swarming motility of the same strains on 0.5% agar plates. Representative swarm zones are shown to the right. Graphed values in all cases are averages with standard deviations calculated from three independent colonies and were normalized by setting the values of the wild type strain as 1. P-values shown for relevant comparisons were calculated with two-tailed student t-test (***P < 0.001; **P < 0.01).
ScIentIfIc REPORts | (2018) 8:12205 | DOI:10.1038/s41598-018-29785-w of lack of PP4397 in the double Δpp2258/Δpp4397 strain -i.e. expression of PP4397 or PP4397-FLAG in the double mutants resulted in a reduced motility phenotype approximating that of the PP2258 null strain (Fig. 4). The motility phenotypes shown in Fig. 4 were unaffected by addition of 0.5 mM IPTG (data not shown). These results confirm that the motility rescue phenotype of the double mutant is due to lack of PP4397/FlgZ. Biochemical analysis of PP4397 has shown that alanine substitutions of arginine 127 (R127A) or glycine 162 (G162A) both abolish the capacity of PP4397 to bind c-di-GMP. Arginine 127 is directly involved in binding of c-di-GMP, while glycine 162 is conserved among PilZ domain proteins and is probably needed for correct folding 25 . To confirm that c-di-GMP binding is required for PP4397 to exert its phenotypic effects, equivalent expression plasmids for native and C-terminally FLAG-tagged PP4397-R127A and PP4397-G162A derivatives were generated and tested as described for wild type PP4397. Neither of these c-di-GMP binding defective derivatives mediated a reduced motility phenotype (Fig. 4), even though they were expressed at the same levels as the wild type protein (Figs 4B and S4A). Thus, these results demonstrate that PP4397 is c-di-GMP responsive in vivo, and that c-di-GMP binding by PP4397/FlgZ is a prerequisite for its effects on P. putida motility.
YcgR, like PP4397, restores a motility defect in P. putida Δpp2258/Δpp4397. To ascertain if PP4397 and YcgR showed cross-species functionality, plasmids expressing PP4397-FLAG and E. coli YcgR-FLAG under control of an araC/P BAD promoter were introduced into E. coli MG1655-ΔyhjH/ΔycgR (which has elevated c-di-GMP levels due to the lack of the PDE YhjH) and P. putida Δpp2258/Δpp4397 (which also has elevated c-di-GMP levels due to the lack of PP2258). Relative swimming motilities were assayed on LB soft 0.3% agar swimming motility plates containing 0 to 1.0% L-arabinose. As anticipated, motility of the E. coli ΔyhjH/ΔycgR strain was greatly reduced by expression of YcgR-FLAG induced with either 0.2% or 1% L-arabinose, but not by expression of PP4397-FLAG (Fig. 5A). However, Western analysis revealed that expression levels of PP4397-FLAG were notably lower than those of YcgR-FLAG, which likely underlies the inability of PP4397 to cause an altered motility phenotype in this strain (expanded Western Fig. S4B). In marked contrast, both YcgR-FLAG and PP4397-FLAG greatly reduced motility of P. putida Δpp2258/Δpp4397 when expressed at similar levels (induced with 1% L-arabinose Figs 5B, and S4B). No reduction in motility was observed with c-di-GMP binding-deficient derivatives of either protein (PP4397-R127A-FLAG and its corresponding YcgR-R118A-FLAG counterpart, data not shown). These results lend strong support to the idea that despite their limited identity (Fig. 1), PP4397/FlgZ and YcgR are functional c-di-GMP responsive counterparts that act to control motility in P. putida and E. coli, respectively. uted throughout their surface, and fluorescently tagged YcgR has previously been found to localize to puncta on the cells together with the flagellar apparatus 17,19 . Similar puncta have been observed for P. fluorescens when fluorescently tagged FlgZ was overexpressed in cells with elevated c-di-GMP 23 . In the case of P. aeruginosa, which possesses a single polar flagellum, mono-copy fluorescently tagged FlgZ exhibits co-polar localization with the motility apparatus, and could be observed in a higher percentage of cells when c-di-GMP levels were elevated 24 . P. putida KT2701 used here possess a bundle of 6 to 10 flagella located at a single pole 27 and, therefore, co-localization of its FlgZ/PP3497 counterpart with flagella would be anticipated to result in a polar localization.
To determine if PP4397, similarly to YcgR and other FlgZ counterparts, co-localizes with the flagellar machinery, PP4397-EYFP fusions were introduced into P. putida, both in mono-copy in its native location on the chromosome, and in multi-copy on an araC/P BAD expression plasmid (as used in the motility assays in Fig. 5). Functionality of the PP4397-EYFP fusion, designed to have the same intervening residues as the YcgR-EYFP fusion, was confirmed by its maintenance of the reduced motility phenotype of the PP2258 null strain (Fig. S5). Western analysis was performed on cells harvested at the same time as cells were fixed for microscopy to facilitate correlation between images and corresponding protein expression levels.
In contrast to a mono-copy polar localization control (Aer2-EYFP 13 ), mono-copy PP4397-EYFP was expressed at a higher level and localized to the cytoplasmic compartment in P. putida (Fig. 6, compare B to C). This apparent cytoplasmic localization was maintained in strains lacking PP2258 (Fig. 6, compare C and D) -i.e. under elevated c-di-GMP levels that results in altered swimming and swarming motility (Figs 3 and S5). This contrasts data for P. fluorescens and P. aeruginosa 23,24 , where cytosolic FlgZ counterparts could be visualized as puncta or at the pole under conditions where cellular c-di-GMP levels were elevated.
Cytosolic localization was also observed with PP4397-EYFP expressed from a multi-copy plasmid under inducing (1% L-arabinose) conditions (Fig. 6, compare C and E), which also alter P. putida motility (Figs 3 and S6C). While present at lower levels, multi-copy expression of YcgR-EYFP likewise showed a cytoplasmic location in P. putida (Fig. 6, compare E and F) and had a corresponding reduced effect on motility (Fig. S6C). This contrasts its punctate localization in E. coli, where it is expressed at similar levels as PP4397 (Fig. 6, compare F and H).
Taken together, the data in Figs 5 and 6 suggests that interaction between PP4397 (and likely YcgR) with the flagella motility apparatus of P. putida is weaker and/or more transient than that of YcgR with the motility apparatus of E. coli; and further, that a constant strict association with the flagella motor is not required for functionality. Although the interaction target of PP4397/FlgZ is unknown, based on the findings with the highly homologous FlgZ counterpart of P. aeruginosa 23 , it appears likely that one predominant target would be MotC and that functional replacement by YcgR relies on regions bearing common features between P. putida MotC and E. coli MotA proteins. Determining the interaction partner(s) for PP4397 is the subject of future studies.

Concluding Remarks
As for other bacteria, artificial increase of c-di-GMP levels by expression of native or heterologous DGCs results in reduced flagella-mediated motility in P. putida KT2440 12 . Here we identify the PilZ domain containing PP4397/FlgZ protein as the effector relay protein that responds to elevated c-di-GMP levels resulting from lack of the signalling protein PP2258. Because P. putida harbours multiple c-di-GMP turnover proteins, it is likely that other c-di-GMP signalling pathways could also feed in to fine tune flagella performance through PP4397/ FlgZ. Amongst the forty two P. putida c-di-GMP turnover proteins, PP2258 is the only one currently identified . Western analysis of EYFP-tagged proteins expressed from mono-copy chromosomal translational fusions present in 50 and 25 µg of crude extract. Cell were harvested for Western analysis at the same time as fixing for imaging (after 2 to 2.5 hrs of growth; OD 600 0.5 to 0.7), which contrasts those shown for the motility assays in Fig. 5 (harvested after 5 hr of growth; OD 600 ~3.5). Note that Aer2-EYFP, although clearly visible at the pole of the cell in panel B, is expressed at much lower levels than PP4397-EYFP and is not detected at the exposure shown. The cropped image is derived from the same experiment processed in parallel on the same gel, and are shown alongside molecular size markers in Fig. S6A. Lower panels [E to H] and cognate western are of strains cultured on LB in the presence of 1% L-arabinose. Boxed images are differentially exposed cells for comparison of the presence or lack of puncta. (E) P. putida KT2701-Δpp2258/Δpp4397 (double PP2258/PP4397 null strain) carrying the multi-copy araC/P BAD pp4397-eyfp expression plasmid (pVI2374). Cells examined: n = 271, 0% with puncta. (F) P. putida KT2701-Δpp2258/ΔPP4397 carrying the multi-copy araC/P BAD ycgR-eyfp expression plasmid (pVI2375). Cells examined n = 282, 0% with puncta. (G) E. coli MG1655-ΔyhjH/ΔycgR (double YhjH/YcgR null strain) carrying the multi-copy araC/P BAD pp4397-eyfp expression plasmid (pVI2374). Cells examined: n = 343, 0% with puncta. to possess both c-di-GMP degrading (PDE) and synthesising (DGC) activities 12 . However, the mechanism that controls the two opposing activities of PP2258 is unknown. One possibility is suggested by the genetic context of the pp2258 gene, which is located in a bicistronic operon downstream of aer1 that encodes a polar-localized receptor 13 . Both PP2258 and Aer1 possess PAS domains that are renowned for facilitating protein-protein interactions. Because the PAS domain of PP2258 is critical for its DGC activity 12 , it appears plausible that direct or indirect interaction between Aer1 and PP2258 could trigger a switch in its activities. Our current dissection of the signal transduction cascade from PP2258 to PP4397/FlgZ should greatly facilitate future work to determine if Aer1 controls PP2258 c-di-GMP signalling to ultimately control the ability of PP4397/FlgZ to act as an active hand-brake on the flagella motor.

Methods
Bacterial strains, growth conditions and general procedure. E. coli and P. putida strains (Table S1) were grown at 37 °C and 30 °C, respectively. E. coli DH5 30 was used for construction and maintenance of expression plasmids. The specialised replication-permissive E. coli S17λpir host, which expresses the Pir protein essential for replication of R6K 31 was used for maintenance and conjugation of R6K-based suicide plasmids. P. putida strains used are all based on the genome sequenced KT2440 32 or a spontaneous streptomycin resistant derivative of KT2440 (KT2701 33 ). Plasmids (Table S2) were constructed by standard molecular techniques, as detailed in supporting information, and were introduced into P. putida by either electroporation or conjugation. Strains were cultured in Luria-Bertani (LB) broth (AppliChem GmbH) or on agar solidified plates supplemented with appropriate antibiotics. Concentrations used for E. coli were carbenicillin (Cb) 100 µg/ml, kanamycin (Km) 50 µg/ml, and tetracycline (Tc) 5 µg/ml, while those for P. putida were Cb 1 mg/ml, Km 50 µg/ml, and Tc 50 µg/ml. PCR determination of the genome organisation of pp4397. Generation of cDNA from total RNA isolated from P. putida was as previously described 13 . After cDNA synthesis, mRNA was removed by 15 min incubation at 37 °C in the presence of 0.23 M NaOH and then neutralized by adding HEPES to a final concentration of 625 mM. The cDNA was subsequently buffer exchanged to 10 mM Tris-HCl (pH 8.5) using High Pure PCR product purification kit (Roche) before being subjected to PCR using the primer sets listed in Table S3 and depicted in Fig. 2B.
Generation of P. putida strains lacking PP4397. The pp4397 gene replacement cassette (Δpp4397::Tc) was introduced into the chromosome of P. putida KT2701 and its PP2258 null derivative 13 via conjugation of pVI2299 (Table S2) from E. coli S17λpir and subsequent double-site recombination as previously described 13 . Growth in medium containing Tc and 10% sucrose was used to select for recombinants. Diagnostic PCR of the resulting strains was used to confirm loss of the native intact pp4397 gene and the presence of a fragment encompassing novel junctions of the Tc gene replacement and DNA upstream and downstream of the gene fragment of the suicide plasmid.

Generation of P. putida mono-copy chromosomal transcriptional and translational fusions.
Fusions were introduced into the chromosome of P. putida strains via single site recombination as previously described 13 . Suicide plasmids carrying 3′-regions of target genes with cognate transcriptional fusions to either the promoter-less luxAB genes or in-frame translational fusions to eyfp, were introduced by conjugation as described above. Recombinants were selected using the antibiotic resistance marker(s) of the vector. Since the suicide plasmids carry only 3′-portions of the target genes, the resulting strains contain one functional (fused) copy and one inactive truncated copy of the gene separated by plasmid DNA. Diagnostic PCR was used to confirm correct recombination using primers homologous to DNA upstream of the gene fragment on the suicide plasmid and the DNA of the fusion partner.
In vivo luciferase transcriptional reporter assay. P. putida strains harboring mono-copy transcriptional fusions to luxAB were cultured in LB supplemented with appropriate antibiotics. To ensure balanced growth, overnight cultures were diluted in pre-warmed media and cultured into the exponential phase prior to a second dilution (to OD 600 ~ 0.04) and initiation of the experiment. Growth and luciferase activity were monitored every 45 minutes for >9 hrs. Light emission was determined using 100 μl of culture after addition of decanal (1:2000 dilution) using a Infinite M200 (TECAN) luminometer.
Motility swimming and swarming plate assays. E. coli and P. putida strains were inoculated in LB supplemented with appropriate antibiotics and grown overnight. The next day, cultures were grown into early exponential phase, diluted to an OD 600nm of 0.1 and grown once again for 5 hours. Cultures were then adjusted to