Regulation of extracellular matrix components by AmrZ is mediated by c-di-GMP in Pseudomonas ogarae F113

The AmrZ/FleQ hub has been identified as a central node in the regulation of environmental adaption in the plant growth-promoting rhizobacterium and model for rhizosphere colonization Pseudomonas ogarae F113. AmrZ is involved in the regulation of motility, biofilm formation, and bis-(3′-5′)-cyclic dimeric guanosine monophosphate (c-di-GMP) turnover, among others, in this bacterium. The mutants in amrZ have a pleiotropic phenotype with distinguishable colony morphology, reduced biofilm formation, increased motility, and are severely impaired in competitive rhizosphere colonization. Here, RNA-Seq and qRT-PCR gene expression analyses revealed that AmrZ regulates many genes related to the production of extracellular matrix (ECM) components at the transcriptional level. Furthermore, overproduction of c-di-GMP in an amrZ mutant, by ectopic production of the Caulobacter crescentus constitutive diguanylate cyclase PleD*, resulted in increased expression of many genes implicated in the synthesis of ECM components. The overproduction of c-di-GMP in the amrZ mutant also suppressed the biofilm formation and motility phenotypes, but not the defect in competitive rhizosphere colonization. These results indicate that although biofilm formation and motility are mainly regulated indirectly by AmrZ, through the modulation of c-di-GMP levels, the implication of AmrZ in rhizosphere competitive colonization occurs in a c-di-GMP-independent manner.


Results
AmrZ regulates the expression of genes encoding ECM components in P. ogarae F113. As previously observed, amrZ mutants are dramatically impaired in the attachment to surfaces and present a different morphology compared to the wild-type strain in the presence of Congo Red (CR) 7 . Since these observations are consistent with a defect in the production of ECM components, we have studied the transcriptomic profile of F113 and its isogenic amrZ mutant using RNA-Seq approach 32 . The RNA-Seq analyses were done under different growth conditions: exponential and stationary phases in liquid cultures and after rhizosphere colonization. As observed in Fig. 1, the transcriptomic analyses have shown that the expression of several gene clusters encoding proteins implicated in polysaccharide synthesis and the production of extracellular structural proteins is downregulated in the amrZ mutant under all tested conditions compared to the wild-type strain. Differential gene expression was found for 59 genes putatively involved in the production of eight different ECM components, mostly observed during stationary phase, indicating a role for AmrZ in activating the expression of genes related to ECM during this phase (Fig. 1). Fold-Change and p-adjusted values for those genes are shown in Supplementary Table S1.
The AmrZ-regulated clusters include papA-P (PSF113_1970-1955) involved in the production of the putative Pap polysaccharide and pgaA-D (PSF113_0161-0164) encoding PNAG EPS. Some of the genes from the alg cluster (PSF113_4752-4763), involved in alginate synthesis, were differentially expressed in the amrZ mutant compared to the wild-type strain with log 2 Fold-Change values close to the threshold under the tested conditions. The other F113-encoded polysaccharide, levan, is not subjected to AmrZ regulation under any of the tested conditions. Regarding the regulated-clusters related with the production of extracellular proteins or proteinaceous structures, they include fapA-F (PSF113_2680-2685), which are necessary to produce Fap; PSF113_0208, PSF113_1511 and PSF113_3004 encoding the extracellular proteins LapA, MapA, and PsmE, respectively; as well as their associated transport systems (PSF113_0209-0211, PSF113_1508-1510, and PSF113_3005-3007); and PSF113_4178-4192 involved in the Flp/Tad pilus formation. c-di-GMP governs the AmrZ-dependent regulation of ECM components in P. ogarae F113. We have previously shown that AmrZ is a major regulator of c-di-GMP synthesis and that AmrZ-mediated transcriptional activation of several DGCs occurs mainly during stationary phase 7 . On the other hand, almost all the ECM-related clusters have a FleQ-binding site 6 in their promoter regions and only some of them harbor an AmrZ-binding site 24 . Moreover, we have previously described the reciprocal regulation between AmrZ and FleQ TFs 6 . For these reasons, we hypothesized that the observed AmrZ regulation of EMCs could be done through the regulation of c-di-GMP levels or its interrelation with FleQ. In order to test this hypothesis, we first artificially increased the overall c-di-GMP concentration in the wild-type strain, the amrZ mutant, and a double mutant in amrZ and fleQ. The increase of c-di-GMP concentration was obtained through ectopic expression of the mutated version of the pleD* gene from Caulobacter crescentus encoding a constitutive DGC 35,36 . It has been shown that pleD* overexpression stimulates EPSs production, such as cellulose in P. syringae pv. tomato DC3000 13 .
We have also used a qRT-PCR approach to test the expression of selected genes encoding putative ECM components in the wild-type strain and derivatives growing in SA medium at stationary phase either in the absence or presence of pleD* (Fig. 2). As shown in Fig. 2, gene expression was remarkably low, and in some cases, almost undetectable for most of the ECM-related genes tested in amrZand amrZ -fleQcarrying the empty plasmid pJB3Tc19, validating RNA-Seq results. In the case of psmE (PSF113_3004), its gene expression is much higher in the double mutant than in the single mutant, suggesting that FleQ is repressing its expression. Conversely, the presence of pleD* in the amrZ mutant resulted in a significant increase in the expression levels of all tested www.nature.com/scientificreports/ genes. This increase in expression was several folds higher than in the wild-type for genes encoding Pap (papA, PSF113_1970), PNAG (pgaA; PSF113_0164), Fap (fapB; PSF113_2684), MapA (mapA, PSF113_1511), and Flp/ Tad (flp-1, PSF113_4192), demonstrating that the role of AmrZ in the regulation of ECM components occurs indirectly through c-di-GMP. Interestingly, the expression of lapA (PSF113_0208), which encodes the large adhesin LapA was higher in the amrZ -pJBpleD* background than in the amrZ mutant, although expression was still lower than in the wild-type strain, suggesting additional elements involved in its regulation. However, the artificial increase in c-di-GMP levels was not enough to increase gene expression levels in the amrZ -fleQbackground in papA, alg8, fapB, lapA, mapA, and flp-1, also suggesting the dependency of FleQ in their regulation. These results confirm that AmrZ mainly regulates ECM production indirectly, through the increase of c-di-GMP levels and with the interplay of the TF FleQ. The second approach used in this work aims to decipher which DGCs and PDEs are involved in the transcriptional regulation of ECM components in F113. For this purpose, we used mutants affected in the production of DGCs and PDEs with defects in biofilm formation and that have been previously reported as AmrZ-regulated targets: dipA -(PSF113_0499), yfiN -(PSF113_0715), adrA -(PSF113_1982), PSF113_4827, PSF113_3553, and PSF113_4681 7 . DipA and PSF113_4681 contain both GGDEF and EAL motifs, YfiN, AdrA, and PSF113_4827 contain GGDEF motifs, and PSF113_3553 contains an HD-GYP domain. AmrZ activated expression of all the genes encoding these enzymes except for the dipA gene, for which a negative regulation was observed 7 . As shown in Fig. 3, the gene expression of ECM-related genes is also affected by changes in the pool of c-di-GMP produced by the DGCs and PDEs studied. Our results show that gene expression for papA, pgaA and alg8 are decreased in PSF113_4681 mutant. Besides, papA and pgaA gene expression is increased in the dipA mutant; pgaA gene expression is increased in PSF113_4827 mutant and reduced in adrA mutant. Regarding protein components, psmE, fapB, lapA and flp-1 gene expression is decreased in the PSF113_4681 mutant. Moreover, fapB gene expression is increased in the mutant PSF113_3553; lapA gene expression is reduced in yfiN and adrA mutants; flp-1 gene expression is increased in dipA and adrA mutants and reduced in the PSF113_4827 and finally, mapA gene expression is exclusively increased in the dipA mutant (Fig. 3). These results show that different DGCs and PDEs participate in the regulation of the production of ECM components, by affecting the expression of genes involved in the synthesis of the ECM. We have previously reported that amrZ mutants show alterations in CR colony morphology and staining, are hypermotile, and are defective in the attachment to surfaces 7 . All these phenotypes are consistent with the low levels of c-di-GMP. To test whether these phenotypes were caused by the low levels of c-di-GMP that are prevalent in the amrZ mutant 7 , we have used ectopic expression of pleD* from a plasmid in the wild-type and amrZ mutant backgrounds. Furthermore, since FleQ is a regulator of motility and biofilmrelated genes in this bacterium 6 and there is a cross-regulation by AmrZ and FleQ of ECM-related genes expression according to the findings of this work, we have also tested the suppression by c-di-GMP of the CR, biofilm and motility phenotypes in the double mutant amrZ-fleQ. As shown in Fig. 4, the phenotypes associated with amrZand amrZ -fleQare substantially modified when increasing intracellular levels of c-di-GMP by overexpression of pleD*. We analyzed the colony morphology and CR-binding ability of amrZand amrZ -fleQon CR-supplemented plates in the absence or presence of pleD* (pJB3Tc19 and pJBpleD* respectively). As shown in Fig. 4a, amrZand amrZ -fleQ -, showed a light staining, a smooth patch, and an entire border. Conversely, the wild-type strain presented stronger staining, the patch had a rough texture, and its border was dented. In the wild-type, the overproduction of PleD* resulted in a rougher texture and slightly stronger staining. A similar phenotype was observed for all the mutants overexpressing pleD*: an increase in the color intensity, rougher texture, and dented borders. Furthermore, Fig. 4b shows that the defective biofilm formation phenotypes of the amrZand amrZ -fleQare reversed by increasing c-di-GMP intracellular levels, both after the initial attachment at 2 h and in a later development stage at 4 h. Indeed, not only attachment to solid surfaces was increased, but also cell-cell aggregates and pellicle formation (Supplementary Figures S1 and S2) were higher in the pleD* overexpressing strains. Finally, swimming motility analysis has shown the suppression of the hypermotile phenotype of the amrZ mutant caused by pleD* overexpression (Fig. 4c). Moreover, both F113-pJBpleD* and amrZ -pJBpleD* were non-motile after 24 h. As expected, amrZ -fleQis nonmotile in the absence or presence of pleD*, due to the crucial role of FleQ in the synthesis of the flagellar apparatus in this bacterium 4,6 . Altogether, these results suggest that AmrZ regulates motility and biofilm formation mainly through the regulation of c-di-GMP levels.
Overproduction of c-di-GMP does not suppress the defect in rhizosphere competitive colonization in an amrZ mutant. Increased motility in P. ogarae F113 has been associated with enhanced rhizosphere competitive colonization ability 11 . However, despite amrZbeing a hypermotile mutant, it is dramatically impaired in competitive rhizosphere colonization assays 7 . Thus, a similar approach as previously described in Figure 2. AmrZ regulates ECM-related genes in a c-di-GMP-and FleQ-dependent manner. qRT-PCR analysis of selected extracellular matrix-related genes in stationary phase. Gene expression from amrZor amrZ -fleQcarrying pJB3Tc19 or pJBpleD* was normalized to rpoZ and relativized to the wild-type strain carrying the empty plasmid pJB3Tc19. Averages and standard deviation (SD) from two biological replicates with three technical replicas are represented. Means not sharing any letter are significantly different by multiple t-tests with Bonferroni-Dunn method (p value < 0.05). www.nature.com/scientificreports/ this work was carried out to understand the role of c-di-GMP in the AmrZ regulation of the rhizosphere competitive colonization process. To avoid plasmid loss due to the lack of selection, alfalfa rhizosphere competitive colonization assays were performed using F113, the amrZ mutant, and the amrZ mutant harboring a mini-Tn7 based pleD* overexpression system (amrZ -pleD*). The overexpression system was tested for the expected swimming and biofilm formation phenotypes (Supplementary Figure S3). As shown in Fig. 5, amrZwas totally displaced from the alfalfa rhizosphere by F113, consistent with our previous study 7 . Similarly, amrZ -pleD* was displaced entirely by F113, indicating that an increase in c-di-GMP levels was not enough to suppress the competitive colonization phenotype of the amrZ mutant. Furthermore, amrZ -pleD* is dramatically displaced even by the colonization-impaired amrZ mutant, suggesting that bacteria must tightly regulate c-di-GMP levels to be competitive during rizhosphere colonization.

Discussion
Bacterial regulation of the switch between a planktonic lifestyle towards biofilm formation is an important trait for rhizosphere competence and persistence and, therefore, for the performance of rhizobacteria as plant growth promoters. When bacteria form biofilms, they are embedded in a self-produced ECM. The ECM components are subjected to fine control as they are needed not only in adhesion and immobilization of cells providing structure to biofilms but also play an essential role in signaling, genetic exchange, ion-sequestration, and migration 37 .
The second messenger c-di-GMP has been shown to play a crucial role in the transition from motile to sessile lifestyles 10 . We have previously shown that two pleiotropic regulatory proteins, AmrZ and FleQ, are involved in this transition in F113 6 . The involvement of c-di-GMP in this pathway is also evidenced by the regulation of DGC-coding gene expression exerted by AmrZ in this bacterium 7 , while FleQ is a c-di-GMP binding protein, whose activity depends on the binding of c-di-GMP as demonstrated in other pseudomonads 38 .
Here we have shown that in F113, AmrZ is a positive transcriptional regulator of genes related to ECM production, such as polysaccharides, Fap, extracellular adhesins, and the Flp/Tad pilus (Figs. 1 and 2) and that an increase in c-di-GMP levels in the amrZ mutant is enough to increase the expression to wild-type levels or higher (Fig. 2). Interestingly, only two of these genes, lapA and flp-1 were previously found as putative direct qRT-PCR analysis of selected extracellular matrix-related genes in stationary phase. Gene expression from mutants affected in the synthesis of proteins with GGDEF motifs (yfiN -, adrA -, and PSF113_4827), HD-GYP domain (PSF113_3553), or containing both GGDEF and EAL motifs (dipAand PSF113_4681) was normalized to rpoZ and relativized to the wild-type strain. Averages and SD from two biological replicates with three technical replicas are represented. Significant differences in gene expression were determined with multiple t-tests with Bonferroni-Dunn method (*p value < 0.05; **p value < 0.01; ***p value < 0.001; ****p value < 0.0001; ns: not significant). www.nature.com/scientificreports/ regulatory targets for AmrZ due to the presence of an AmrZ-binding site in their promoter region 24 . In fact, in the case of lapA, c-di-GMP levels do not affect at the transcription level, suggesting a direct regulation by AmrZ. Four of the putative ECM components, Pap, Fap, MapA and Flp/Tad pili show a common regulatory pattern. AmrZ appears to regulate all of them via c-di-GMP production. Interestingly, in all of them c-di-GMP effect is FleQ dependent, since PleD* overproduction had no effect in their transcription in the amrZ-fleQ double mutant (Fig. 2). It is interesting to note that we have previously identified FleQ-binding sites in the promoter regions of papA (PSF113_1970),), lapA (PSF113_0208), mapA (PSF113_1511) tadB (PSF113_4182), and rcpA (PSF113_4178) 6 . These results indicate complex AmrZ-FleQ-c-di-GMP interplay participating in the regulation of these ECM components. AmrZ and FleQ have already been described as regulators of the synthesis of different ECM components in other pseudomonads 19,31,[38][39][40] . A role for FleQ has been also observed for lapA expression in P. putida KT2440 6 . Conversely, we have not observed a role for FleQ in psmE transcription and the possible role of c-di-GMP on the regulation of pgaA appears independent of FleQ. Furthermore, in this work, we have shown that the expression of genes related to the production of ECM components responds to changes in the pool of c-di-GMP caused by mutation of distinct DGCs and PDEs that are regulated by AmrZ 7 (Fig. 3). Genes dipA and PSF113_4681, encoding enzymes with both GGDEF and EAL domains appear to be the most important regulators for ECM components production. Mutants affected in either of these genes show an effect on the transcription of papA, pgaA and flp1, the dipA mutation affects mapA expression and the PSF113_4681 mutation has an effect on the transcription of alg8, lapA and psmE. In all cases, mutations in dipA resulted in an increased transcription, while mutations in PSF113_4681, resulted in a transcriptional decrease. Mutations In E. coli, PNAG production is also subjected to c-di-GMP-dependent transcriptional regulation via the c-di-GMP produced by the DGC YddV 41 . Similarly, c-di-GMP contribution to the alginate gene cluster expression has been previously observed in E. coli, in which the ectopic overexpression of the DGC YedQ led to increased alg8 and alg44 gene expression 42 . Although, to our knowledge, no previous data for the relation of c-di-GMP with the fap cluster in pseudomonads is available, regulation of their functional homologs in Salmonella sp. and E. coli, the amyloid curli fibers, is c-di-GMP dependent through the CsgD regulator 43,44 . The results presented here have shown that modulation of c-di-GMP levels is the major pathway for the AmrZ regulation of biofilm formation and motility. In this regard, pleD* overexpression, leading to c-di-GMP overproduction, can suppress the biofilm-forming defects in the amrZ mutant, restoring attachment, Congo Red staining, and bacterial aggregation while suppressing the hypermotile phenotype (Fig. 4). However, the increase of c-di-GMP levels did not restore the rhizosphere colonization defect of the amrZ mutant (Fig. 5). We have previously shown that an amrZ mutant is totally displaced by the wild-type strain in competitive rhizosphere colonization experiments 7 . Here we show that the amrZstrain overproducing c-di-GMP is also displaced from the rhizosphere by the wild-type strain and by the colonization-defective amrZ mutant. These results clearly show that rhizosphere competitive colonization phenotype of the amrZ mutant is not a consequence of c-di-GMP levels, in contrast to biofilm formation and motility in the amrZ mutant, and that high levels of c-di-GMP seem to be detrimental for competitive rhizosphere colonization.
Swimming motility. Motility of P. ogarae F113 and derivatives carrying either the empty vector pJB3Tc19 or the overproducing diguanylate cyclase PleD*: pJBpleD* or miniTn7pleD*Tc was tested in swimming motility assays. Strains were grown overnight in SA agar plates supplemented with Tc at 28 °C. Strains were inoculated in the center of SA agar plates (0.3%) supplemented with Tc. Swimming haloes diameters were measured after 24 h incubation at 28 °C. Experiments were performed at least in duplicate with three replicates in each experiment. Figure 5. c-di-GMP does not restore the defective rhizosphere competitive colonization phenotype of an amrZ mutant. Pseudomonas ogarae F113 wild-type, amrZ -, and the amrZ mutant containing a miniTn7-based pleD* integration were tested in competitive colonization assays. CFUs recovered from the alfalfa rhizosphere after seven days post-inoculation were tested for antibiotic resistance to distinguish the different strains. Percentages of colonies are represented. Each experiment was repeated twice with ten plants per experiment. www.nature.com/scientificreports/ Congo red binding assay. pJB3Tc19 and pJBpleD* or miniTn7pleD*Tc-containing strains were grown overnight in YMB supplemented with Tc and 10 µL of culture were inoculated in triplicate in YMB agar plates supplemented with CR and Tc and incubated at 28 °C. Colony morphology and staining were recorded from day two to seven with a Leica MX125 stereoscope (zoom 1.25x). Every assay was performed three times.
Adherence and biofilm formation assays. Adherence and biofilm formation abilities were assayed following a procedure previously described 52 . Briefly, overnight cultures containing either pJB3Tc19 and pJBpleD* or miniTn7pleD*Tc grown in LB medium supplemented with Tc were adjusted to an optical density (OD) 600  Pseudomonas ogarae F113 wild-type and the amrZ mutant were used as controls. After a week post-inoculation, shoots were removed, and rhizosphere bacteria were recovered by resuspension of soil in 20 mL of NaCl 0.75% (w/v) by vortex. Dilution series were plated onto SA agar plates supplemented with Rif and Chx and allowed to grow for 72 h at 28 °C. CFUs of each strain were plated onto fresh SA plates and distinguished by resistance to antibiotics (Km in the case of the amrZ mutant and KmTc in the amrZ mutant with miniTn7pleD*Tc integration). Assays were performed in duplicate with ten independent plants each time.
RNA isolation, cDNA synthesis, and qRT-PCR assay. Two independent cultures of F113 carrying the empty plasmid pJB3Tc19, amrZ -, amrZ -fleQcontaining pJB3Tc19 or pJBpleD* were grown in SA medium until stationary phase. Then, 1 mL of each culture was centrifuged for 5 min at 12,000 × g at RT. The supernatant was discarded and the cell pellet was resuspended in 100 µL of RNAlater® Solution (Ambion) before storage at 4 °C. RNA isolation, cDNA synthesis, and qRT-PCR assays were custom made by Plataforma de Genómica de la Fundación Parque Científico de Madrid (Madrid, Spain) with the primers listed in Supplementary Table S3. Gene expression was calculated using the Ct values. Data were normalized using rpoZ expression as housekeeping and relativized to F113 pJB3Tc19 following the 2 −ΔΔCt method 54 .
RNA-Seq. RNA isolation, sequencing, and bioinformatic analysis from F113 and amrZcultures grown in SA medium at exponential (OD 600 = 0.6) and stationary phase (OD 600 = 1.2), or from rhizosphere colonization experiments were performed as described in 32 . F113 genes predicted to encode putative extracellular matrix components were selected. Differential gene expression was considered for values meeting a threshold of log 2 Fold-Change (mutant/wild-type) ≥ 1 or ≤ -1 and a p-adjusted value ≤ 0.001. Heat-map was made using pheatmap R package version 1.0.12 55 .
Statistical analysis. GraphPad Prism version 7.00 for Windows (GraphPad Software, San Diego, California USA, www. graph pad. com) was used in the statistical analysis and representation of swimming motility, biofilm formation assays, and qRT-PCR data using multiple t-tests for independent samples with Bonferroni-Dunn method.