ABSTRACT
A core phosphorelay pathway that directs developmental transitions and cellular asymmetries in Agrobacterium tumefaciens putatively includes two overlapping, integrated phosphorelays. One of these phosphorelays putatively includes at least four histidine sensor kinase homologues, DivJ, PleC, PdhS1, and PdhS2, and at least two response regulators, DivK and PleD. Previously we demonstrated that PdhS2 reciprocally regulates biofilm formation and swimming motility. In the current study we further dissect the role and regulatory impact of PdhS2 in A. tumefaciens revealing that PdhS2-dependent effects on attachment and motility require the response regulator, DivK, but do not require PdhS2 autokinase or phosphotransfer activities. We also demonstrate that PdhS2 regulation of biofilm formation is dependent upon multiple diguanylate cyclases, including PleD, DgcA, and DgcB, implying that PdhS2 regulation of this process intersects with pathways regulating levels of the second messenger cyclic diguanylate monophosphate (cdGMP). Finally, we show that upon cell division a GFP fusion to PdhS2 dynamically localizes to the new pole of the bacterium suggesting that PdhS2 controls processes in the daughter cell compartment of predivisional cells. These observations suggest that PdhS2 negatively regulates DivK, and possibly PleD, activity to control developmental processes in the daughter cell compartment of predivisional cells, as well as in newly released motile daughter cells.
IMPORTANCE Bacterial developmental processes, including morphological transformations as well as behavioral transitions, are tightly regulated. In many Alphaproteobacteria cell division and development are coordinated by a suite of conserved histidine kinases and their partnered regulatory proteins. Here we describe how the histidine kinase PdhS2 genetically interacts with a single-domain response regulator, DivK, and the intracellular signal cyclic diguanylate monophosphate. PdhS2 dynamically localizes to the new pole of recently divided cells and negatively regulates processes that ultimately lead to attachment and subsequent biofilm formation in Agrobacterium tumefaciens. These findings expand our understanding of the links between cell division and developmental control in A. tumefaciens and related Alphaproteobacteria.
INTRODUCTION
Bacteria are sometimes considered to be elementary life forms, with simple body plans, streamlined reproductive cycles, and monolithic behavior when compared with higher eukaryotes. To the contrary, many bacteria can exhibit a remarkable diversity of developmental complexity, both temporal and morphological (1, 2). Even bacterial species whose cells appear morphologically uniform, such as rod-shaped Escherichia coli or coccoid Staphylococcus aureus, possess distinct cellular architectures as well as intricately timed cell division programs, and a large number of bacteria can form multicellular biofilms (3, 4). Developmental processes in bacteria, as in higher eukaryotes, are driven by factors that may be considered both cell-intrinsic and cell-extrinsic. Intrinsic factors include genomic and proteomic content, while extrinsic factors comprise environmental conditions, such as pH and temperature, which cells sense and to which they respond (5).
Members of the Alphaproteobacteria class include host-associated pathogens (e.g. Brucella sp., Bartonella sp.), host-associated commensals (e.g. Sinorhizobium sp., Bradyrhizobium sp.), and free-living aquatic bacteria (e.g. Caulobacter sp., Rhodobacter sp.). Recent work has revealed that several Alphaproteobacteria divide asymmetrically, in which cells elongate, duplicate and segregate their genomic content between two non-equivalent compartments of predivisional cells, and finally generate two cells by cytokinesis (6, 7). Notably, cellular components are unevenly distributed between the two daughter cells during cell division, including surface structures (e.g. flagella and holdfasts), cell wall components (e.g. peptidoglycan), and even cytoplasmic complexes (e.g. heat shock proteins). For example, there may be a clear segregation of existing organelles to one daughter cell while the second cell generates these structures de novo (6, 8-10). Although the specific details may vary among species, the end result is the production of a young daughter cell and a comparatively older mother cell. Not only does this uneven division partition senescence among the products of cell division, but it also allows for the generation of functionally distinct cell types. For example, in Caulobacter crescentus the non-motile stalked cell type can attach to surfaces using its polar adhesin called the holdfast (11). This stalked cell then serves as the mother cell during multiple rounds of cell division, generating and releasing motile swarmer cells upon each cytokinetic event (12). Motile swarmer cells are prohibited from entering the cell division cycle until differentiation into the non-motile stalked form (13, 14).
Underlying asymmetric cell division is subcellular differentiation that includes localization of specific regulatory proteins to programmed locations within each cell (15). Prominent among these are components of two overlapping phosphorelays, the first which functions through the response regulators DivK and PleD (the DivK-PleD relay) and the second which functions primarily through the response regulator CtrA (the CtrA relay). The pathways are connected through DivK, which controls initiation of the CtrA relay by regulating the CckA sensor kinase (16, 17). Collectively we refer to these two relays as the DivK-CtrA pathway. In C. crescentus the sensor histidine kinases PleC and DivJ control the phosphorylation state of DivK and PleD, and localize to opposing poles of the predivisional cell (18-21). Through antagonistic kinase and phosphatase activities on the same target proteins, DivK and PleD, PleC and DivJ inversely manifest their activity on the most downstream component of the DivK-CtrA pathway, the response regulator CtrA (22-25). DivJ is retained at the stalked cell pole and serves as a DivK/PleD kinase, resulting in increased DivK~P concentration and decreased CtrA~P concentration in this region of the cell (Fig. 1A). Conversely, PleC localizes to the pole distal to the stalk, where the single polar flagellum is assembled, and where PleC dephosphorylates DivK, leading downstream to increased CtrA~P levels and activity. Phospho-CtrA binds to the replication origin thereby preventing DNA replication and also acts as a transcriptional regulator for many genes, including activating those for assembly of the flagella (26-28). The CtrA relay is also influenced by the DivK-PleD relay through levels of the second messenger cyclic diguanylate monophosphate (cdGMP). DivJ-dependent phosphorylation of PleD at the stalk pole of the predivisional cell stimulates its diguanylate cyclase activity, resulting in higher levels of cdGMP at this end of the cell. The CckA kinase that initiates the CtrA relay is biased away from its kinase and towards its phosphatase activity by direct allosteric control through high levels of cdGMP, thereby reinforcing a CtrA~P gradient, relatively low at the stalk pole and increasing towards the distal pole (29-34) (Fig. 1A).
Agrobacterium tumefaciens is a plant pathogen of the Alphaproteobacteria group that is not stalked, but like C. crescentus, divides asymmetrically generating a motile daughter cell from a mother cell (6). As a facultative pathogen, the A. tumefaciens lifestyle substantially differs from that of the freshwater oligotroph C. crescentus. Nonetheless core components of the DivK-CtrA pathway are well conserved in A. tumefaciens, including the three non-essential PleC/DivJ homologue sensor kinase (PdhS) homologues PleC (Atu0982), PdhS1 (Atu0614), and PdhS2 (Atu1888). The divJ gene (Atu0921) is essential in A. tumefaciens (35, 36) (Fig. 1B). We have previously shown that the three non-essential PdhS homologues have distinct roles in the normal cellular development of A. tumefaciens (Kim, 2013). PleC and PdhS1, as well as the A. tumefaciens DivK homologue, all manifested marked effects on both cell division and complex behaviors including motility and biofilm formation. In genetic backgrounds with reduced or absent levels of these proteins cells are elongated and branched, and these strains also exhibit both reduced motility and reduced biofilm formation. To date the essentiality of divJ has precluded exhaustive phenotypic analysis (35) although preliminary depletion studies suggest it similarly impacts these same phenotypes (Heindl et al. in prep). The fourth PdhS family member, PdhS2, does not appear to participate in regulation of cell division as all cells are morphologically wild-type in appearance (35). Loss of PdhS2, however, results in dramatically increased attachment and biofilm formation and simultaneous loss of motility. Reciprocal regulation of these phenotypes is often a hallmark of regulation by cdGMP (37). In this work we further explore the mechanism by which PdhS2 regulates adhesion and motility. Our results link DivK downstream of PdhS2. We also show a clear intersection of PdhS2 activity and the activity of several diguanylate cyclases, suggesting that PdhS2 and cdGMP coordinately regulate biofilm formation and motility in A. tumefaciens. Finally, we find that PdhS2 dynamically localizes to the newly generated poles of both the new daughter cell and the mother cell following cytokinesis. This pattern suggests that PdhS2 activity is specifically required for proper development of motile daughter cells.
RESULTS
DivK is epistatic to PdhS2
Members of the PdhS family of sensor kinases were originally identified based on homology with their namesakes DivJ and PleC of C. crescentus (38, 39) (Fig. 1B). Based on this homology all PdhS family members are predicted to interact with the single domain response regulator DivK (39) and also interact with the diguanylate cyclase response regulator PleD. Prior work from our laboratory has shown that both swimming motility and adherent biomass are diminished in the ΔdivK mutant, implying that DivK activity is required for proper regulation of these phenotypes in A. tumefaciens (35). In contrast, the PdhS–type kinase PdhS2 inversely regulates these phenotypes; a ΔpdhS2 mutant is non-motile but hyperadherent (35). To determine whether PdhS2 genetically interacts with DivK we constructed a ΔdivKΔpdhS2 mutant and compared swimming motility and biofilm formation in this strain to wild-type and parental single deletion strains (Fig. 2A). As seen before, loss of either divK or pdhS2 reduced swimming motility as measured by swim ring diameter on motility agar. Biofilm formation on PVC coverslips in the ΔdivK mutant was reduced relative to the wild-type C58 strain while for the ΔpdhS2 mutant it was dramatically increased. The ΔdivKΔpdhS2 mutant phenocopied the ΔdivK mutant in both assays, with no significant difference in the efficiency of either swimming motility or biofilm formation between the two strains.
Swim ring diameters of the ΔdivK and ΔdivKΔpdhS2 mutants were decreased by roughly 20% compared to wildtype while the decrease in ΔpdhS2 swim ring diameters was roughly 40% compared to wildtype, suggesting that the nature of the defect in swimming motility differs between these two classes of mutants and that loss of divK partially restores motility in the absence of pdhS2. Indeed, it was earlier noted that while both the ΔdivK and the ΔpdhS2 single deletion mutants produce polar flagella few ΔpdhS2 mutant bacteria were observed to be motile under wet-mount microscopy implying that the swimming defect is due to diminished flagellar activity rather than flagellar assembly (35). The ΔdivK mutant, however, was readily observed to be motile under wet-mount microscopy. Similarly, the ΔdivKΔpdhS2 mutant generates polar flagella and its motility is readily observed under wet-mount microscopy (data not shown). Both the ΔdivK and ΔdivKΔpdhS2 mutants, and not the ΔpdhS2 mutant, generate aberrant cell morphologies including elongated and branched cells (35) (Data not shown). These data support the hypothesis that PdhS2 interacts with DivK and that the phenotypes of the ΔpdhS2 mutant are at least partly due to altered regulation of DivK activity. Moreover, the reciprocal nature of the phenotypes between the ΔdivK and ΔpdhS2 mutants implies that PdhS2 negatively regulates DivK.
Further support for PdhS2 activity proceeding through DivK was provided by expressing wild-type PdhS2 ectopically from a Plac promoter. Induced expression of pdhS2 rescues swimming motility and returns biofilm formation closer to wild type levels in the ΔpdhS2 mutant, albeit incompletely (Fig. 2B). However, as predicted from the ΔdivKΔpdhS2 phenotypes, plasmid-borne provision of PdhS2 in the ΔdivK mutant had no significant effect on either biofilm formation or swimming motility (Fig. 2C).
PdhS2 kinase and phosphatase activities coordinately regulate developmental phenotypes
Members of the PdhS family of sensor histidine kinases contain a conserved HATPase_c catalytic domain at their carboxyl termini and an upstream conserved HisKA dimerization/phosphoacceptor domain (Fig. 1C). Many sensor kinases exhibit bifunctional catalytic activity, alternately acting as kinase or phosphatase, and in C. crescentus PleC is one such example (18, 22, 40). Multiple sequence alignment of the HisKA domain from the A. tumefaciens and C. crescentus PdhS family kinase homologues highlights the high level of conservation of this domain including the phospho-accepting histidine residue (H271 of PdhS2) and a threonine residue important for phosphatase activity (T275 of PdhS2) (Fig. 1B).
To test the requirement of the conserved phospho-accepting histidine for PdhS2 activity we mutated this residue to alanine (H271A) and evaluated the ability of this mutant pdhS2 allele to complement ΔpdhS2 phenotypes (Fig. 2B). Ectopic expression of PdhS2H271A (Plac-pdhS2 on a low copy plasmid) efficiently complemented the attachment and motility phenotypes of the ΔpdhS2 mutant. These data indicate that this histidine residue is not required for PdhS2 regulation of swimming motility and biofilm formation, and imply that PdhS2 autokinase and phosphotransfer are not required to regulate these phenotypes. Instead, we hypothesize that PdhS2 acts primarily as a phosphatase towards its cognate response regulators, including DivK.
The conserved HisKA dimerization/phosphoacceptor domain is also primarily responsible for the phosphatase activity of these two-component system kinases (41). Phosphatase activity requires a conserved threonine residue roughly one α-helical turn (4 residues) downstream of the phospho-accepting histidine residue. To test the requirement for PdhS2 phosphatase activity in regulating developmental phenotypes we mutated this conserved threonine residue to alanine (Thr275A) and evaluated the ability of this mutant pdhS2 allele to complement ΔpdhS2 phenotypes. In contrast to the PdhS2H271A mutant protein, equivalent ectopic expression of the PdhS2T275A allele failed to complement the ΔpdhS2 motility and attachment phenotypes, and in fact exacerbated them (Fig. 2B). Expression of either the kinase-null or the phosphatase-null allele of PdhS2 in the ΔdivK background had no effect on biofilm formation or swimming motility (Fig. 2C). A PdhS2 double mutant with both the histidine and threonine residues mutated had no effect on biofilm formation or swimming motility (Fig. S1). Together these data are most consistent with the hypothesis that PdhS2 phosphatase activity is primarily responsible for regulating these phenotypes in A. tumefaciens. The data are also consistent with PdhS2 negatively regulating DivK, and DivK being downstream of PdhS2 in the DivK-CtrA pathway. In wild-type A. tumefaciens induced expression of PdhS2 modestly but significantly increases biofilm formation and decreases swimming motility (Fig. 2B). Induced expression of the PdhS2H271A protein in the wild-type background, however, significantly reduces biofilm formation. These data suggest that PdhS2 kinase activity may also play a role during the normal life cycle of A. tumefaciens and that the balance of PdhS2 kinase and phosphatase activities is coordinated to match dynamic cellular conditions.
PdhS2 and DivJ localize to the pole of A. tumefaciens
One mechanism for establishing and maintaining developmental asymmetries in bacteria is the differential polar localization of proteins with opposing functionalities (42, 43). Several members of the PdhS family of sensor kinases localize to one or both bacterial poles (18, 22, 44). Using a full length PdhS2-GFP fusion that retains wild type functionality we tracked localization of PdhS2 in A. tumefaciens following induction of expression with IPTG (Fig. 3A and Movie S1). In both wild-type and ΔpdhS2 backgrounds PdhS2-GFP localized primarily to the budding pole of predivisional cells. Time-lapse microscopy of the ΔpdhS2 mutant expressing PdhS2-GFP revealed apparent dynamic relocalization of PdhS2-GFP to the newly generated pole of the daughter cell coincident with cytokinesis. In the mother cell a new focus of PdhS2-GFP subsequently accumulates at the new budding pole generated following septation and daughter cell release. These time-lapse experiments clearly indicate that PdhS2-GFP localizes to the actively budding pole of the cell, and is lost at that pole as it matures and the cell proceeds to the predivisional state. Following cytokinesis PdhS2-GFP localizes to the newly generated, younger poles of both the daughter cell and the mother cell. This dynamic localization is consistent with PdhS2 activity being restricted to the motile-cell compartment of predivisional cells and to newly generated motile cells.
Since PdhS2 localizes to the budding pole of dividing cells and primarily requires its phosphatase activity there likely exist one or more non-budding pole-localized kinases opposing PdhS2 activity. The most obvious candidate for this activity is DivJ. DivJ localizes to the old pole in C. crescentus and acts as a DivK kinase in both C. crescentus and S. meliloti (39, 45). Time-lapse microscopy of a full length DivJ-GFP fusion in wild-type A. tumefaciens reveals localization to the non-budding pole in mother cells that is not redistributed over the course of multiple cell division cycles (Fig. 3B and Movie S2). The interplay between PdhS2 and DivJ is a topic of current study, one made difficult by the essential nature of divJ. The remaining PdhS kinases also likely affect DivK or PleD activity. Localization patterns for PdhS1 and PleC have not been determined. However, phenotypic evaluation of pdhS1, pdhS2, and pleC double and triple mutants reveals complicated and non-redundant regulation of biofilm formation by these PdhS kinases (Fig. S2).
PdhS2 and DivK affect CtrA activity
To determine the effect of PdhS2 and DivK activity on CtrA activity and stability in A. tumefaciens we evaluated its steady-state levels and turnover in unsynchronized cells. Steady-state levels of CtrA protein measured in stationary phase cultures were modestly increased in the ΔpdhS2 background and modestly decreased in the ΔdivK background, consistent with PdhS2 and DivK inversely regulating CtrA accumulation or stability (Fig. 4A). Because steady-state protein levels reflect the balance between protein synthesis and degradation CtrA protein stability was evaluated following treatment of cultures with chloramphenicol to inhibit translation. In wild-type cultures, translation inhibition leads to a decline in CtrA abundance over the course of 2-3 hours, diminishing to roughly 30% of the steady-state levels. Loss of pdhS2 had no effect on either final steady-state levels of CtrA following inhibition of translation or the observed rate of turnover when compared with wild-type cultures (Fig. 4B). Taken together these data suggest that PdhS2 and DivK are likely to primarily affect CtrA activity or accumulation rather than CtrA stability.
In C. crescentus, CtrA is known to directly regulate at least 55 operons, acting as either an activator or repressor of transcription (22, 28). A. tumefaciens CtrA is predicted to act similarly to C. crescentus CtrA, binding to DNA in a phosphorylation-dependent manner and regulating DNA replication and transcription. A. tumefaciens CtrA is 84% identical to C. crescentus CtrA at the amino acid level and purified C. crescentus CtrA binds to a site upstream of the A. tumefaciens ccrM gene (46). Furthermore, computational analysis of multiple Alphaproteobacterial genomes uncovered numerous cell cycle regulated genes preceded by a consensus CtrA binding site (47). We therefore evaluated CtrA activity by determining the transcriptional activity of several known and hypothesized CtrA-dependent promoters from both C. crescentus and A. tumefaciens. The ccrM, ctrA, and pilA promoters from C. crescentus were chosen to represent CtrA-activated promoters that we predicted would be similarly regulated in A. tumefaciens (13, 14, 48, 49). In the ΔpdhS2 background, expression levels from both the ctrA and pilA promoters from C. crescentus were significantly reduced while transcription from the C. crescentus ccrM promoter was unchanged (Table 1). In the A. tumefaciens ΔdivK background the C. crescentus ccrM and ctrA promoters exhibited increased activity while the pilA promoter was unchanged (Table 1). These data are largely consistent with A. tumefaciens CtrA regulating transcription of known CtrA-dependent promoters, and with PdhS2 and DivK inversely regulating CtrA activity in A. tumefaciens.
From A. tumefaciens the ccrM promoter is the only promoter for which experimental data suggest CtrA-dependent regulation, thus this promoter was selected for analysis (46). Curiously, the same computational analysis that identified possible CtrA-dependent promoters among a suite of Alphaproteobacterial genomes failed to find a significant CtrA binding motif upstream of the A. tumefaciens ccrM gene (47). In addition to ccrM, putative promoters for ctrA and pdhS1 were selected for analysis based on the presence of at least one predicted CtrA binding site as well as hypothesized cell cycle regulation of these loci. Transcriptional activity from the A. tumefaciens ctrA and pdhS1 promoter constructs showed inverse regulation in the ΔpdhS2 and ΔdivK backgrounds, with expression decreased from the ctrA promoter and increased at the pdhS1 promoter in the ΔpdhS2 mutant, and exactly reversed in the ΔdivK mutant. Although absence of pdhS2 had little effect on the A. tumefaciens ccrM promoter, transcription from this promoter was significantly increased in the ΔdivK background (Table 1). These data are congruent with the above data for C. crescentus CtrA-dependent promoters and further support CtrA regulation of cell cycle-responsive genes in A. tumefaciens.
PdhS2 activity intersects with cyclic-di-GMP pools
In C. crescentus DivJ and PleC positively regulate, via phosphorylation, a second response regulator, PleD, as well as DivK (22). In C. crescentus and A. tumefaciens the divK and pleD coding sequences form one operon and transcriptional regulation of both genes is linked. Since PdhS kinases are predicted to interact with both DivK and PleD we analyzed the effect of loss of PleD activity in the ΔpdhS2 background. As reported previously, deletion of pleD alone has minimal effect on either swimming motility or adherent biomass (35). Loss of pleD in the ΔpdhS2 background had only a minor effect on swimming motility (Fig. S3A). Adherent biomass, however, was reduced by approximately 30%, indicating that PleD contributes to the increased attachment phenotype of the ΔpdhS2 mutant (Fig. 5A).
PleD is a GGDEF motif-containing diguanylate cyclase, and thus it is likely that the attachment phenotype of the ΔpdhS2 mutant requires increased levels of cdGMP. Earlier work from our lab identified three additional diguanylate cyclases that are relevant to attachment and biofilm formation: DgcA, DgcB, and DgcC (50) (Wang unpublished). As seen in wild-type C58, deletion of dgcA or dgcB in the ΔpdhS2 background significantly decreased attachment and biofilm formation, whereas loss of dgcC did not (Fig. 5A). These data suggest that increased biofilm formation by the ΔpdhS2 mutant is dependent on cdGMP pools, generated through PleD, DgcA, and DgcB. Swimming motility was equivalent in either wild-type C58 or ΔpdhS2 backgrounds in combination with mutations in pleD, dgcA, dgcB, or dgcC (Fig. S3A).
Previously we found that mutants lacking either dgcA, dgcB, or dgcC show no difference in average levels of cdGMP (50). Nonetheless, loss of either dgcA or dgcB, or mutation of the GGDEF catalytic site of either enzyme, significantly reduced biofilm formation (50), implicating these enzymes in controlling the pool of cdGMP, and affecting attachment. We compared cytoplasmic cdGMP levels among wild-type C58, the ΔpdhS2 mutant, and the ΔpdhS2 ΔdgcB mutant strain and found no change in these levels (Fig. S4). To verify that the diguanylate cyclase (DGC) activity is responsible for the increased biofilm formation in the ΔpdhS2 background we expressed an allele of dgcB with a mutation in its GGDEF catalytic motif (GGAAF; dgcB*). We have previously shown that this mutation abrogates cdGMP formation by DgcB and that production of this mutant protein does not complement a ΔdgcB mutant for either cdGMP formation or attachment phenotypes (50). Plasmid-borne expression of wild-type dgcB (Plac-dgcB) results in a massive increase in attachment and biofilm formation in either the wild-type C58 or ΔpdhS2 background (Fig. 5B). Expression of dgcB* from a Plac promoter, however, did not increase biofilm formation in either background. In the ΔpdhS2 ΔdgcB mutant background expression of wild-type dgcB increased biofilm formation to the same degree seen in the wild-type and ΔpdhS2 mutant. Expression of the mutant dgcB* allele in the ΔpdhS2 ΔdgcB mutant did appear to modestly increase biofilm formation, although far less than with the wild-type dgcB allele. Swimming motility was modestly but significantly reduced when the wild-type dgcB allele was provided in trans (Fig. S3B). Together these data are consistent both with PdhS2 negatively regulating one or more diguanylate cyclases, and for increased synthesis of cdGMP mediating the attachment and biofilm formation phenotypes of the ΔpdhS2 mutant and contributing, albeit less so, to the ΔpdhS2 motility phenotype.
Increased attachment in a pdhS2 mutant requires the UPP polysaccharide
We have previously reported that PleD-stimulated attachment was due to increased levels of the unipolar polysaccharide (UPP) and cellulose (50). In addition to UPP and cellulose, A. tumefaciens produces at least three other exopolysaccharides: succinoglycan, cyclic β-1, 2 glucans, β-1, 3 glucan (curdlan), as well as outer membrane associated lipopolysaccharide (LPS) (51). Of these only LPS is essential for A. tumefaciens growth (36). The ΔpdhS2 strain was tested for the impact of each of the non-essential exopolysaccharides for biofilm formation. The upp mutation completely abolished attachment in both the wild type and the pdhS2 mutant (Fig. S5). The chvAB mutant, known to have pleiotropic effects (52), was diminished in adherence overall, but was still elevated by the pdhS2 mutation. None of the other exopolysaccharide pathways impacted adherence in either background. These results confirm that biofilm formation in the ΔpdhS2 strain is dependent primarily on UPP and that the motility phenotype of the ΔpdhS2 mutant is not dependent on any of the known exopolysaccharides.
Global transcriptional analysis of PdhS2 activity
Prior work on the DivK-CtrA pathway in other Alphaproteobacteria has shown that one mechanism by which these bacteria switch between the developmental programs of mother and daughter cells is by differential gene expression in the two cell types (27, 39, 53). The PdhS kinases may affect gene expression through two mechanisms. First, members of the PdhS kinase family have been demonstrated to affect the activity of the global transcriptional regulator, CtrA, in several Alphaproteobacteria, including C. crescentus, S. meliloti, and B. abortus (32, 39, 44, 54). Transcriptome effects may also result from partitioning during cytokinesis of factors regulating transcription independent of CtrA, a process that is itself indirectly regulated by the PdhS family kinases. To determine the effect of PdhS2 activity on the A. tumefaciens transcriptome we used whole genome microarrays. Gene expression was compared between WT and ΔpdhS2 strains grown to exponential phase in minimal media. Of 5338 unique loci represented on the arrays 39 genes were differentially regulated above our statistical cut-offs (P values, ≤ 0.05; log2 ratios of ≥ ± 0.50; Table 2). Of these, 24 genes were significantly upregulated, indicating negative regulation by PdhS2. Upregulated genes included dgcB, shown above to contribute to biofilm formation in the ΔpdhS2 mutant, and Atu3318, a LuxR-type transcriptional regulator previously identified as regulated by the VisNR regulatory system involved in motility and biofilm formation (50). Downregulated genes included six succinoglycan biosynthetic genes, consistent with previously published results showing positive regulation of succinoglycan production by PdhS2 (35).
To determine whether any of these 39 genes were putatively regulated by CtrA we scanned a sequence window from 500 bp upstream of the start codon to 100 bp into the coding sequence for plausible CtrA binding sites. CtrA binding sites were defined using the conserved Alphaproteobacterial CtrA recognition sequence 5’-TTAANNNNNNGTTAAC-3’ (46, 47). Sequences containing seven or more of the conserved nucleotides in this motif were deemed plausible candidates. Using these parameters 16 differentially transcribed loci have matching upstream sequences and are thus putatively regulated by CtrA, including dgcB and Atu3318, as well as all five of the downregulated succinoglycan biosynthetic genes (Table 2).
To verify our microarray results we measured transcription of translational fusions to β-galactosidase from two selected genes, dgcB and Atu3318, in wild-type and ΔpdhS2 backgrounds (Table 1). In both cases beta-galactosidase activity increased in the ΔpdhS2 mutant, corroborating the microarray results. Overall these results are consistent with PdhS2 impacting the motile cell developmental program through transcriptional control. The data are also consistent with PdhS2 negatively regulating DivK and thereby indirectly affecting CtrA activity during the A. tumefaciens cell cycle.
DISCUSSION
PdhS2 is a key regulator of developmental phenotypes
Regulation of the developmental program of many Alphaproteobacteria centers on the global transcriptional regulator CtrA (42, 47, 55, 56). CtrA activity is controlled, indirectly, through a series of phosphotransfer reactions dependent on one or more PdhS-type histidine kinases. Here we show that PdhS2, one of at least four PdhS family kinases from A. tumefaciens, regulates developmental phenotypes at least in part through fine-tuning of the activity of CtrA. We demonstrate that null mutations of the single domain response regulator divK are epistatic to pdhS2 in A. tumefaciens and that PdhS2 is likely to negatively regulate DivK activity. Null mutation of a second response regulator, pleD, is also epistatic to the pdhS2 mutation. The phosphatase activity of PdhS2 predominates in vivo, although it is apparent that its kinase activity is also required for proper regulation of PdhS2 phenotypes. Determination of the timing, specificity, and regulation of both phosphatase and kinase activities await further experimentation. PdhS2 dynamically localizes to the budding pole of A. tumefaciens cells following cytokinesis while DivJ, another PdhS-type kinase, localizes to the non-budding pole of each cell. We propose that together the antagonistic activities of DivJ and PdhS2 (and perhaps additional PdhS homologues), coupled with their distinct localization patterns, generate a spatiotemporal gradient of phospho-DivK and phospho-PleD and an opposing gradient of phospho-CtrA, thus differentially regulating the developmental program of A. tumefaciens (Fig. 1A). Computational models of asymmetric cell development in C. crescentus support this notion, with the important caveat that phospho-DivK and phospho-PleD may not be distributed in a gradient but rather locally restricted (23, 25, 57).
Plasticity and robustness of the DivK-PleD signaling module and PdhS kinases
The PdhS kinase homologues are all predicted to interact with two response regulators, DivK and PleD. Through modulating DivK and PleD activity, PdhS family members ultimately affect activity of CtrA. Among the Alphaproteobacteria, regulation of CtrA activity has diversified such that the number of PdhS homologues has increased and/or decreased in different species. Species have been identified that putatively contain as many as 12 distinct members, while other species have lost all PleC/DivJ homologues (39, 47). The A. tumefaciens genome encodes four PdhS–type proteins, all of which impact one or more developmental phenotypes. The selective forces leading to expansion or contraction of the number of PdhS family members in different species are unknown. It is evident, however, that simple functional redundancy is not a sufficient explanation as the essentiality and genetic interactions between PdhS family members and their downstream targets (DivK and PleD) varies among different Alphaproteobacteria. In A. tumefaciens only divJ is essential among the PdhS-type kinases, and neither divK nor pleD are essential, leading to the hypothesis that at least one unidentified target for DivJ exists, the activity of which is required for viability, or that dysregulation of multiple targets is synthetically lethal (35). All possible single, double, and triple mutant combinations of the remaining PdhS family members in A. tumefaciens are viable, as is the ΔdivK ΔpleD mutant strain (35) (Fig. S2; and data not shown). The divK gene is essential in both C. crescentus and S. meliloti but essentiality varies for the PdhS-type kinases (39, 58). Neither pleC nor divJ are essential in C. crescentus, whereas pleC is essential in S. meliloti. While neither S. meliloti divJ nor cbrA (another PdhS family member) are independently essential, cells must have at least one copy of either gene for viability (39). The varied essentiality of module components together with the known effects from loss of multiple components highlight both the genetic plasticity and signaling robustness evident in the PdhS-DivK regulatory circuit among Alphaproteobacteria.
PdhS2 influences cdGMP-dependent phenotypes
The effects of the pdhS2 mutation on biofilm formation and swimming motility by PdhS2 differ from those observed for either the ΔpleC or ΔpdhS1 mutants (35) (and data not shown). In these mutants, biofilm formation is inhibited and cell morphology is elongated and branched, similarly to the ΔdivK mutant. Reduced swimming motility in these morphologically aberrant cells is most likely due to a combination of their aberrant cell shape and flagellar placement, rather than a defect in assembly or function of the flagella (35). The motility defect in the ΔpdhS2 strain, however, appears fundamentally distinct. The increased biofilm formation and diminished motility in a pdhS2 mutant is most similar to the inverse regulation frequently achieved by modulating internal pools of cdGMP. Indeed, our data demonstrate a strong dependence on cdGMP levels for mediating the ΔpdhS2 phenotypes. Although the overall levels of cdGMP are not significantly affected by the pdhS2 mutation, it is clear that mutations in the diguanylate cyclases PleD, DgcA and DgcB that disable their ability to synthesize cdGMP diminish the biofilm phenotype of pdhS2 null mutants. This suggests that increased cdGMP synthesis via these enzymes may impart the effects on motility and biofilm formation. PdhS2 is likely to affect PleD enzymatic activity through the phosphorylation state of its receiver domain, similar to other PleD homologues (22, 59, 60). Additionally, mutation of pdhS2 increases dgcB transcription providing a direct link to elevated cdGMP production. Whether PdhS2 also regulates DgcA activity remains to be determined.
Indirect regulation of CtrA by PdhS2
A primary candidate for ultimately mediating the effects of PdhS2 on A. tumefaciens physiology is the transcriptional regulator CtrA. Our data support altered CtrA activity, and not abundance, as primarily responsible for pdhS2- and divK-dependent transcriptional responses. There are several examples in C. crescentus where altered CtrA activity is observed without apparent significant changes in CtrA abundance (36, 61, 62). There are also several examples in both C. crescentus and S. meliloti where altered PdhS kinase activity results in both perturbed activity and altered abundance of CtrA (39, 63). Our expression analyses suggest that A. tumefaciens CtrA functions to both activate some promoters (e.g. succinoglycan synthetic genes and ctrA) while repressing others (e.g. dgcB, Atu3318, and pdhS1).
Convergence of PdhS2 and VisNR regulatory pathways
Many phenotypes regulated by PdhS2 mirror those regulated by the LuxR-type transcriptional motility regulators VisR and VisN (50). Loss of either visN or visR results in strong inhibition of motility and a dramatic increase in attachment and biofilm formation, that is dependent on cdGMP production. Moreover, VisNR effects on attachment require the diguanylate cyclases DgcA and DgcB. Both PdhS2 and VisNR positively regulate succinoglycan production at the transcriptional level, and negatively regulate transcription of the LuxR-type regulator Atu3318. The promoter region for Atu3318 as well as several of the succinoglycan genes contain a consensus CtrA binding site, and PdhS2 control is most likely mediated through CtrA. It is possible that the increased activity of one or more diguanylate cyclases in the ΔvisR/ΔvisN mutants ultimately results in decreased abundance or activity of CtrA. In both C. crescentus and A. tumefaciens, cdGMP allosterically switches the bifunctional hybrid histidine kinase CckA from kinase to phosphatase mode, downregulating CtrA phosphorylation and DNA-binding activity (30, 31). In contrast, VisNR motility control is mediated directly through transcriptional control of the Rem activator of flagellar gene expression (50, 64). Our pdhS2 expression profiling does not reveal a decrease in flagellar gene transcription, suggesting distinct mechanisms. Furthermore, the increase in UPP and cellulose production in visR and visN mutants is cdGMP-dependent, and likely proceeds via allosteric control of the UPP biosynthesis machinery and cellulose synthase (which has a PilZ-type cdGMP binding motif). It is quite plausible that a significant fraction of the pdhS2 hyper-adherent phenotype reflects similar allosteric control of UPP and cellulose.
Segregation of antagonistic signaling activity promotes asymmetric development
The asymmetric division of A. tumefaciens and other Alphaproteobacteria, producing two genetically identical but phenotypically distinct daughter cells, requires well-coordinated regulation of two developmental programs. The mother cell remains in a terminally differentiated state, proceeding through distinct synthesis (S) and growth (G2) phases of the cell cycle (51, 55). During G2 phase the cell elongates into a predivisional cell and establishes a functional asymmetry between its two cellular poles by differential localization of antagonistic homologues of the PdhS kinases. At least one PdhS kinase, DivJ in C. crescentus and A. tumefaciens, PdhS in B. abortus and CbrA in S. meliloti, localizes to the attached pole (18, 38) (Fig. 1A and 3B). At this site these kinases can act to phosphorylate DivK and PleD, indirectly inactivating CtrA (as reported for Caulobacter). At the opposite pole at least one PdhS kinase, PleC in C. crescentus and PdhS2 in A. tumefaciens (Fig. 1A and 3A), localizes and acts primarily to dephosphorylate DivK and PleD, ultimately promoting CtrA stability and activity. Upon cytokinesis, then, the motile daughter cell is released in a G1 growth phase with high levels of CtrA activity establishing a distinct transcriptional program.
Pole specification may also be generated as a natural consequence of cell division. Following each cytokinetic event a new, younger pole is generated from the division septum. Cell elongation in many Rhizobiales, including A. tumefaciens, occurs in the absence of obvious homologues to DivIVA or MreB, important regulators of cell elongation in many rod-shaped bacteria (65). Instead, cell elongation in these bacteria occurs primarily at the younger pole of the bacterium in a process of zonal insertion of peptidoglycan and cell membrane components (6, 7). The canonical cell division proteins FtsZ and FtsA both localize at the growing pole of A. tumefaciens until just prior to cytokinesis at which point they relocalize to the cell division septum. Several other A. tumefaciens proteins, including PopZ and PodJ, act as dynamic fiduciary markers localizing to one or both bacterial poles throughout the cell cycle (66, 67). While these cell cycle, elongation, and division components make use of polar localization the role played by any of these proteins in differentiation of old versus new poles is unclear.
Our data are consistent with PdhS2 acting in the motile daughter cell to prevent premature activation of cell attachment processes such as UPP production, as well as to inhibit motility. Our results suggest that ΔpdhS2 mutant cells are blocked from completely exiting the dividing cell phase and differentiating into the motile cell type. Alternatively, ΔpdhS2 mutant cells may have an abridged motile cell (G1) phase and prematurely differentiate into the dividing cell type. In either case, pdhS2 mutants assemble polar flagella in daughter cells, but these seem to be largely blocked for rotation, hence the swimming deficiency in the pdhS2 mutant.
The subtle effects on CtrA abundance coupled with the modest, but significant, effects on several CtrA-dependent promoters, suggest that either PdhS2 regulation of CtrA activity is restricted to a very tight window during the cell cycle or to a specific location within the developing cell Alternatively, the complement of remaining PdhS kinases buffer CtrA activity in the absence of pdhS2. Microarray results using unsynchronized cultures of S. meliloti lacking either the divJ or the cbrA PdhS kinase were similar to those in our own study (39, 68). All three datasets uncovered a restricted set of differentially regulated genes, several of which were common to one or more experiment (Table 2). Transcriptional assays of the dgcB and Atu3318 promoters are consistent with the microarray data in both directionality and magnitude. It is telling that reporter activity of these potential CtrA-regulated promoters in the ΔpdhS2 mutant is two to three times higher than the remaining regulated promoters (Table 1). This lower overall activity may explain why more putative CtrA-dependent promoters were not identified in the transcriptional profiling. A reassessment for transcriptional effects of the PdhS kinases will be a high priority once synchronization and cell type enrichment procedures are developed.
All of the PdhS-type kinases in A. tumefaciens including PleC and DivJ are hypothesized to function through the DivK and PleD response regulators. However, it is clear that they can have starkly distinct regulatory outputs. In A. tumefaciens pleC and pdhS1 mutants manifest defects in budding and form branched cells, whereas the pdhS2 mutant has no apparent defect in budding, and is instead down for motility and elevated for attachment. These differences may in large measure reflect the spatial and temporal control of their activity. We hypothesize that spatial restriction of PdhS2 activity to the budding poles of mother cells, that rapidly transition to become the old pole of newly formed daughter cells, imparts PdhS2 control of motility and attachment processes, without strongly influencing the budding process per se.
MATERIALS AND METHODS
Strains and plasmids
Bacterial strains, plasmids, and oligonucleotides used in these studies are listed in Tables S1 through S3. A. tumefaciens was routinely cultivated at 28°C in AT minimal medium plus 1% (w/v) glucose as a carbon source and 15 mM (NH4)2SO4 as a nitrogen source (ATGN), without exogenous FeSO4 (69, 70). For biofilm assays 22 μM FeSO4 was included in the media. E. coli was routinely cultivated at 37°C in lysogeny broth (LB). Antibiotics were used at the following concentrations (A. tumefaciens/E. coli): ampicillin (100/100 μg·mL-1), kanamycin (150/25 μg·mL-1), gentamicin (150/30 μg·mL-1), spectinomycin (300/100 μg·mL-1), and tetracycline (4/10 μg·mL-1).
Non-polar, markerless deletion of pdhS2 (Atu1888) in all genetic backgrounds used in this work was accomplished using splicing by overlap extension (SOE) polymerase chain reaction (PCR) followed by homologous recombination, as described (35). Suicide plasmid pJEH040 carries an approximately 1 kb SOE deletion fragment of pdhS2 on a pNPTS138 vector backbone. pNPTS138 is a ColE1 plasmid and as such is unable to replicate in A. tumefaciens. pJEH040 was delivered to recipient strains by either transformation or conjugation followed by selection on ATGN plates supplemented with 300 μg·mL-1 Km, selecting for A. tumefaciens cells in which pJEH040 had integrated at the chromosomal pdhS2 locus by homologous recombination. Recombinants were then grown overnight at 28°C in ATGN in the absence of Km and plated the following day onto ATSN (ATGN with sucrose substituted for glucose) agar plates to select for sucrose resistant (SucR) allelic replacement candidates. After three days’ growth at 28°C colonies were patched in parallel onto ATGN Km and ATSN plates. KmS SucR recombinants were then tested for the targeted deletion by diagnostic PCR using primers external to the pdhS2 locus (JEH100 and JEH113) as well as internal primers (JEH85 and JEH87). Candidate colonies were further streak purified and verified a second time by diagnostic PCR before being used in downstream assays.
Site-directed mutagenesis of pdhS2 was achieved using mutagenic primer pairs JEH245/JEH246 (for generating the His271Ala allele, H271A) or JEH261/JEH262 (for generating the Thr275Ala allele, T275A). Plasmid pJEH021 carrying the wild-type pdhS2 sequence was amplified by PCR using the above primer pairs. Following amplification, reaction mixtures were treated with DpnI restriction endonuclease to remove template plasmid and then transformed into TOP10 F’ E. coli competent cells. Purified plasmids from each transformation were sequenced and those containing the desired mutations, pJEH091 for His271Ala and pJEH099 for Thr275Ala, were selected for sub-cloning. pJEH091 and pJEH099 were digested with NdeI and NheI followed by gel electrophoresis and purification of the resulting insert. Inserts were ligated into similarly digested pSRKGm and transformed into competent E. coli TOP10 F’ cells. Purified plasmids from each transformation were sequenced to verify their identity. The resulting plasmids, pJEH092 (H271A) and pJEH102 (T275A), were used to transform A. tumefaciens. To generate a PdhS2 allele carrying both H271A and T275A mutations the same steps were followed as above using plasmid pJEH091 as template with mutagenic primers JEH261/JEH262.
Translational fusions of full-length wild-type PdhS2 and DivJ to GFP were constructed as follows. pdhS2 and divJ, each lacking a stop codon were amplified by PCR using primer pairs JEH65/JEH146 (pdhS2) and JEH147/JEH148 (divJ) with A. tumefaciens strain C58 genomic DNA as template. Primer design for these amplifications included 5’ NdeI and 3’ NheI restriction sites. The gfpmut3 gene including a 5’ NheI site and a 3’ KpnI site was amplified using primer pair JEH149/JEH150 and pJZ383 as template. Amplicons were gel purified, ligated into pGEM-T Easy, transformed into competent TOP10 F’ E. coli, and eventually sequenced. The resulting plasmids, pJEH052 (pdhS2), pJEH053 (gfpmut3), and pJEH054 (divJ) were digested with either NdeI and NheI (pJEH052 and pJEH054) or NheI and KpnI (pJEH053). Inserts were gel purified and used in a three-component ligation with NdeI/KpnI-digested pSRKGm generating pJEH060 (PdhS2-GFP) and pJEH078 (DivJ-GFP). Sequenced plasmids were used to transform A. tumefaciens. Translational fusions of full-length mutant PdhS2 alleles (H271A or T275A) to tdTomato (Table S2) were generated using primers JEH65 and JEH272 with either pJEH091 or pJEH099 as template. The resulting amplicons were gel purified and ligated into pGEM-T Easy to generate plasmids pJEH125 and pJEH126. pJEH125 and pJEH126 were digested with NdeI and SacI and the resulting inserts gel purified and ligated with similarly treated pSRKKm-tdTomato, generating plasmids pJEH127 and pJEH128.
Reporter gene fusion constructs included predicted promoter regions from between 200 bp and 400 bp upstream of the indicated gene through the start codon. Each upsream region was amplified by PCR using the primers listed in Table S3 using A. tumefaciens genomic DNA as template. Amplicons were gel purified, ligated into pGEM-T Easy, transformed into competent TOP10 F’ E. coli, and eventually sequenced. The resulting plasmids, pJEH113 (ccrM, Atu0794, promoter), pJEH115 (ctrA, Atu2434, promoter), and pJEH119 (pdhS1, Atu0614, promoter) were digested with either KpnI and HinDIII (pJEH113 and pJEH119) or KpnI and PstI (pJEH115). Inserts were gel purified and ligated with similarly treated pRA301 containing a promoterless E. coli lacZ gene without its own ribosome binding site. The resulting constructs (pJEH121, pJEH122, and pJEH124) carry lacZ translationally fused to the start codon for each gene with transcription and translation driven by the fused upstream region. pJEH121, pJEH122, and pJEH124 were used to transform A. tumefaciens for subsequent beta-galactosidase assays.
Static biofilm assays
Overnight cultures in ATGN were sub-cultured in fresh ATGN to an optical density at 600 nm (OD600) of 0.1 and grown with aeration at 28°C until an OD600 of 0.25-0.6. Cultures were diluted to OD600 of 0.05 and 3 mL were inoculated into each of four wells in a 12-well plate. A single coverslip was placed vertically into each well to submerge approximately half of each coverslip. Plates were incubated in a humidified chamber at 28° for 48 h. Coverslips were removed from each well, rinsed with water, and adherent biomass stained by 5 min immersion in a 0.1% (w/v) crystal violet solution. Adsorbed crystal violet was solubilized by immersion in 1 mL 33% acetic acid and the absorbance of this solution determined at 600 nm (A600) on a Synergy HT multi-detection microplate reader (Bio-Tek). Culture density for each sample was also determined by measuring the OD600 of each culture. Data are typically presented as A600/OD600 ratios normalized to values obtained for the wild-type strain within each experiment. ATGN was supplemented with antibiotics and IPTG as appropriate. Final inoculations also included supplemental iron (22 μM). Each mutant was evaluated in three independent experiments each of which contained three technical replicates.
Motility assays
Wet mounts of exponentially growing cultures were observed under brightfield optics using a Zeiss Axioskop 40 equipped with an AxioCam MRm monochrome digital camera. Swim plates containing 0.3% agarose in ATGN, supplemented with 1 mM IPTG and antibiotics when appropriate, were inoculated with a single colony of the indicated strain at a central point and incubated for 7 days at 28°C. Swim ring diameters were measured daily for seven days. Each experimental condition was tested in three independent experiments containing three technical replicates.
Microscopy
Cell morphology and localization of PdhS2-GFP and DivJ-GFP was evaluated using a Nikon E800 fluorescence microscope equipped with a Photometrics Cascade cooled CCD camera. Overnight cultures were grown in ATGN with gentamicin and 250 μM IPTG. The following day each strain was sub-cultured to OD600 0.1 and then grown at 28°C with aeration until ~OD600 0.5-0.8. The culture (0.5 μl) was transferred to a 1% ATGN/agarose pad on a clean glass slide and a clean 22 x 22 mm number 1.5 glass coverslip placed on top. Images were acquired using a 100X oil immersion objective and phase contrast optics or epifluorescence with a FITC-HYQ filter set (Nikon; excitation filter = 480/40 nm, dichromatic mirror = 505 nm, absorption filter = 535/50 nm). Time-lapse microscopy utilized a Nikon Ti-E inverted fluorescence microscope with a Plan Apo 60X/1.40 oil Ph3 DM objective, a DAPI/FITC/Cy3/Cy5 filter cube, an Andor iXon3 885 EMCCD camera, and a Lumencor Spectra X solid state light engine at 20% power. For time-lapse imaging agarose pads included 250 μM IPTG and coverslips were attached to the glass slide using a gas-permeable 1:1:1 mixture of Vaseline, lanolin, and paraffin. Phase and fluorescence images were captured every 20 min for 8 h using a 60 ms (phase) or 2 s (fluorescence) exposure. Images were analyzed using ImageJ (Schneider et al., 2012, Schindelin et al., 2012, Schindelin et al., 2015). Localization of all PdhS2-tdTomato constructs was evaluated using a Zeiss AxioObserver.Z1 fluorescence microscope equipped with a Photometrics Cascade cooled CCD camera. Cultures were treated as above and images acquired using a 100x oil immersion objective and Semrock TXRED-4040b-zhe-zero filter cube.
Transcriptional profiling
Whole-genome transcriptional profiling using custom 60-mer oligonucleotide microarrays was performed essentially as previously described (71). Arrays were produced by Agilent Technologies, and consist of 8455 features that represent 5338 predicted protein-encoding open reading frames, tRNA and rRNA encoding genes, and 2,983 duplicate spots. Cultures of wild-type or the ΔpdhS2 mutant strain of A. tumefaciens strain C58 were grown overnight in ATGN to full turbidity and then sub-cultured 1:150 into fresh ATGN for a second overnight growth. The following morning a volume equivalent to 11 ml of OD600 0.6 was prepared for RNA extraction using RNAprotect Bacteria Reagent (QIAGEN, Germantown, MD) following the manufacturer’s protocol. RNA was extracted from these samples using QIAGEN RNA midipreps (QIAGEN, Germantown, MD) following the manufacturer’s protocol. DNA contamination was removed by DNase digestion using the TURBO DNA-free kit (Ambion, Austin, TX) with the incubation time extended to two hours. First strand cDNA synthesis was performed using Invitrogen SuperScript Indirect Labeling Kit, and cDNA was purified on Qiagen QIAQuick columns. cDNA was labeled with AlexaFluor 555 and 647 dyes using Invitrogen SuperScript cDNA Labeling Kit, and repurified on QIAQuick columns. cDNA was quantified on a NanoDrop spectrophotometer. Hybridization reactions were performed using Agilent in situ Hybridization Kit Plus, boiled for 5 min at 95°C, applied to the printed arrays, and hybridized overnight at 65°C. Hybridized arrays were washed with Agilent Wash Solutions 1 and 2, rinsed with acetonitrile, and incubated in Agilent Stabilization and Drying Solution immediately prior to scanning the arrays. Three independent biological replicates were performed, with one dye swap. Hybridized arrays were scanned on a GenePix Scanner 4200 in the Center for Genomics and Bioinformatics (CGB) at Indiana University. GenePix software was used to define the borders of hybridized spots, subtract background, measure dye intensity at each spot, and calculate the ratio of dye intensities for each spot. Analysis of the scanned images was conducted using the LIMMA package in R/Bioconductor. Background correction of the data was performed using the minimum method (72, 73). The data was normalized within arrays with the LOESS method, and between arrays with the quantile method. Statistical analysis was performed using linear model fitting and empirical Bayesian analysis by least squares. Genes with significant P values (≤ 0.05) and with log2 ratios of ≥ 0.50 or ≤ -0.50 (representing a fold-change of ± 1.4) are reported here. Expression data have been deposited in the Gene Expression Omnibus (GEO) database at the National Center for Biotechnology Information (NCBI) under accession number GSE71267 (74).
β-galactosidase activity was measured using a modified protocol of Miller (75). Cultures carrying transcriptional reporter plasmids were grown overnight in ATGN and sub-cultured the following morning to OD600 0.15. Diluted cultures were grown at 28°C with aeration until reaching mid-exponential growth. Between 100 and 300 μL of exponential phase culture was mixed with Z buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgSO4, pH = 7.0) to a final volume of 1 mL (volume of culture = f) plus two drops 0.05% sodium dodecyl sulfate and 3 drops CHCl3. The amount of culture volume used was calibrated to generate reaction times between 15 minutes and two hours for cultures with activity. 0.1 mL of a 4 mg·mL-1 solution in Z buffer of the colorimetric substrate 2-nitrophenyl β-d-galactopyranoside (ONPG) was added and the time (t) required for the solution to turn yellow was recorded. The reaction was stopped by addition of 1 M Na2CO3 and the absorbance at 420 nm (A420) of each solution was measured. Promoter activity is expressed in Miller units (MUs = [1000 x A420nm]/[OD600nm x t x f]). Each mutant was tested in three independent experiments containing five technical replicates.
Protein stability assays
Steady-state levels of CtrA were determined from stationary phase cultures of wild-type, ΔdivK, and ΔpdhS2 strains of A. tumefaciens. Overnight cultures of each strain were grown in TY broth at 28°C with aeration to an OD600 > 1. Two 1 mL aliquots were removed from each culture, pelleted by centrifugation (13,200 x g, 2 min) and supernatants discarded. One of the resulting pellets was resuspended on ice in 50 μL 100 mM Tris·HCl, pH 6.8, followed by 50 μL 2X SDS-PAGE loading buffer (65.8 mM Tris·HCl, pH 6.8, 26.3% (v/v) glycerol, 2.1% (w/v) sodium dodecyl sulfate (SDS), 0.01% (w/v) bromophenol blue), then stored frozen at -20°C. The second pellet was resuspended in 100 μL 1X protein assay buffer (32.9 mM Tris·HCl, pH 6.8, 1%SDS), boiled 10 minutes, and used for protein concentration determination using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific), per manufacturer’s instructions. Frozen resuspended pellets were thawed on ice and β-mercaptoethanol added to a final concentration of 5% prior to electrophoresis. Samples were normalized for protein concentration and separated on a 12.5% SDS-polyacrylamide gel. Following electrophoresis proteins were transferred to Immobilon-FL polyvinyl difluoride membranes (EMD Millipore). Membranes were rinsed in 1X Tris-buffered saline (TBS; 50 mM Tris·HCl, pH 7.5, 150 mM NaCl) solution and air dried. Membranes were wetted with MeOH and incubated in blocking buffer (1X TBS, 5% non-fat dairy milk [NFDM]) for 1 h at room temperature and then incubated overnight at 4°C with primary antibody (1:5000 dilution of rabbit anti-CtrA from C. crescentus, anti-CtrACc, in 1X TBS/5% NFDM/0.2% Tween 20). The following day membranes were rinsed thoroughly with 1X TBS/0.1% Tween 20 and incubated 1 h at room temperature with secondary antibody (1:20,000 dilution of IRDye 800CW-conjugated goat anti-rabbit antibody (LI-COR) in 1X TBS/5% NFDM/0.2% Tween 20/0.01% SDS). Membranes were rinsed thoroughly with 1X TBS/0.1% Tween 20 followed by 1X TBS alone and air dried in the dark. The resulting blot was imaged using a LI-COR Odyssey Classic infrared imaging system. Band intensities were quantified using the Odyssey Classic software.
Proteolytic turnover of CtrA was evaluated using translational shut-off assays. Overnight cultures were grown in TY broth at 28°C with aeration. The following day each strain was sub-cultured in fresh TY broth to an OD600 0.05 and incubated at 28°C with aeration. To inhibit protein synthesis 90 μg·mL-1 chloramphenicol was added to each culture at OD600 0.5. Starting at the time of chloramphenicol addition 5 mL aliquots were removed every 30 min for 3 h. Each aliquot was pelleted by centrifugation (5000 x g, 10 min). Cleared supernatants were discarded and pellets resuspended to an OD600 10.0 in Tris-Cl, (10 mM, pH 8.0). Resuspended pellets were mixed with 2X SDS-PAGE loading buffer and stored frozen at -20°C. Levels of CtrA were determined by SDS-PAGE and Western blotting as described above. Band intensities were quantified using the Odyssey Classic software and normalized to the band intensity of CtrA from the wild-type background at t = 0 min.
Global cdGMP measurement
Measurement of cdGMP levels was performed by liquid chromatography, tandem mass spectrometry (LC-MS/MS) on a Quattro Premier XE mass spectrometer coupled with an Acquity Ultra Performance LC system (Waters Corporation), essentially as previously described (76). Concentrations of cdGMP in cell samples were compared to chemically synthesized cdGMP (Axxora) dissolved in water at concentrations of 250, 125, 62.5, 31.2, 15.6, 7.8, 3.9, and 1.9 nM to generate a calibration curve. A. tumefaciens derivatives were grown in ATGN overnight at 28°C to stationary phase. Culture densities were normalized after collecting cells by centrifugation and then resuspension in the appropriate volume of ATGN. Cultures were then pelleted by centrifugation and resuspended in ice-cold 250 μL extraction buffer (methanol:acetonitrile:water, 40:40:20 + 0.1 N formic acid) and incubated for 30 min at -20°C. Resuspensions were transferred to microcentrifuge tubes and pelleted (13,000 x rpm, 5 min). 200 μL of the resulting supernatant was neutralized with 8 μL 15% NH4HCO3. Neutralized samples were stored at -20°C prior to mass spectrometry analysis.
ACKNOWLEDGEMENTS
This project was supported by National Institutes of Health (NIH) grants GM080546 (C.F.) and GM109259 (C.M.W.). J.E.H. was supported by a Ruth L. Kirschstein National Research Service Award (1 F32 GM100601) from the NIH.
REFERENCES
REFERENCES
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.