Analysis of Brevundimonas subvibrioides Developmental Signaling Systems Reveals Inconsistencies between Phenotypes and c-di-GMP Levels

Abstract

The DivJ-DivK-PleC signaling system of Caulobacter crescentus is a signaling network that regulates polar development and the cell cycle. This system is conserved in related bacteria, including the sister genus Brevundimonas Previous studies had shown unexpected phenotypic differences between the C. crescentusdivK mutant and the analogous mutant of Brevundimonas subvibrioides, but further characterization was not performed. Here, phenotypic assays analyzing motility, adhesion, and pilus production (the latter characterized by a newly discovered bacteriophage) revealed that divJ and pleC mutants have phenotypes mostly similar to their C. crescentus homologs, but divK mutants maintain largely opposite phenotypes than expected. Suppressor mutations of the B. subvibrioides divK motility defect were involved in cyclic di-GMP (c-di-GMP) signaling, including the diguanylate cyclase dgcB, and cleD which is hypothesized to affect flagellar function in a c-di-GMP dependent fashion. However, the screen did not identify the diguanylate cyclase pleD Disruption of pleD in B. subvibrioides caused no change in divK or pleC phenotypes, but did reduce adhesion and increase motility of the divJ strain. Analysis of c-di-GMP levels in these strains revealed incongruities between c-di-GMP levels and displayed phenotypes with a notable result that suppressor mutations altered phenotypes but had little impact on c-di-GMP levels in the divK background. Conversely, when c-di-GMP levels were artificially manipulated, alterations of c-di-GMP levels in the divK strain had minimal impact on phenotypes. These results suggest that DivK performs a critical function in the integration of c-di-GMP signaling into the B. subvibrioides cell cycle.IMPORTANCE Cyclic di-GMP and associated signaling proteins are widespread in bacteria, but their role in physiology is often complex and difficult to predict through genomic level analyses. In C. crescentus, c-di-GMP has been integrated into the developmental cell cycle, but there is increasing evidence that environmental factors can impact this system as well. The research presented here suggests that the integration of these signaling networks could be more complex than previously hypothesized, which could have a bearing on the larger field of c-di-GMP signaling. In addition, this work further reveals similarities and differences in a conserved regulatory network between organisms in the same taxonomic family, and the results show that gene conservation does not necessarily imply close functional conservation in genetic pathways.

Keywords: Brevundimonas; Caulobacter; c-di-GMP; development; divK; signaling.

FIG 1

Deletions in B. subvibrioides developmental signaling genes results in various physiological phenotypes. (A) Wild-type, divJ, divK, divJ divK, and pleC B. subvibrioides strains were analyzed for swarm expansion and adhesion defects using a soft agar swarm assay and a short-term adhesion assay. Mutant strains were normalized to wild-type results for both assays. Deletion of divJ gives motility defects but minimal adhesion defects, similar to C. crescentus divJ results. B. subvibrioides divK and divJ divK strains yielded opposite results, with severe motility and adhesion defects. The B. subvibrioides pleC strain has reduced motility and moderately reduced adhesion, which is similar but not identical to the C. crescentus pleC strain. (B) Lectin staining of holdfast material of wild-type, divJ, divK, divJ divK, and pleC strains. The pleC strain, despite having reduced adhesion in the short-term adhesion assay, still has detectable holdfast material. (C) GFP-tagged DivJ localizes to the holdfast-producing pole, while PleC-GFP localizes to the pole opposite the holdfast. DivK-GFP displays bipolar localization. These localization patterns are identical to those of C. crescentus homologs.

FIG 2

Bacteriophage Delta serves as a tool to investigate B. subvibrioides pilus production. (A) Phage Delta was tested for infection in 18 different Brevundimonas species. Control assays used PYE medium instead of phage stock. Delta caused a significant reduction in B. subvibrioides and B. aveniformis viability, with some reduction in B. basaltis and B. halotolerans as well. (B) Phage Delta was tested against wild-type and cpaF::pCR B. subvibrioides strains using a soft agar phage assay. The wild type displayed zones of clearing with phage dilutions up to 10⁻⁷, whereas the cpaF strain showed resistance to all phage dilutions. (C) B. subvibrioides developmental signaling mutants were tested with phage Delta in soft agar phage assays. The wild type shows clear susceptibility to Delta, as does the divJ strain, suggesting that, like C. crescentus divJ, it produces pili. The pleC strain shows a 2- to 3-order-of magnitude-reduced susceptibility to the phage, indicating reduced pilus production, which is consistent with the C. crescentus phenotype. The divK and divJ divK strains display a resistance similar to that of the pleC strain. Here again, divK disruption causes the opposite phenotype to divJ disruption, unlike the C. crescentus results.

FIG 3

CleD displays a conserved glutamate residue in place of an aspartate typical of response regulators. CleD orthologs from various Caulobacter and Brevundimonas species were aligned by ClustalW, along with B. subvibrioides DivK. The shaded box indicates B. subvibrioides DivK D53, which is analogous to C. crescentus DivK D53 and is the known phosphoryl-accepting residue. This alignment demonstrates that CleD orthologs all contain a glutamate substitution at that site, which has been found to mimic the phosphorylated state and lock the protein in an active conformation in other response regulators.

FIG 4

Phenotypes exhibited by divK suppressors do not coincide with intracellular c-di-GMP levels. (A) Swarm expansion and surface adhesion of suppressor mutations tested in both wild-type and divK backgrounds. Disruption of cleD and dgcB increased motility in the wild-type and divK backgrounds. Disruptions in the wild-type background lead to various levels of adhesion reduction, but the same disruptions had no effect on adhesion in the divK background. *, motility is statistically insignificant compared to the divK parent; **, motility is statistically significant compared to the divK parent (P < 0.05). (B) c-di-GMP levels were measured using mass spectrometry and then normalized to the amount of biomass from each sample. Despite disruptions causing increased motility in the wild-type background, these strains had different c-di-GMP levels. No disruption changed c-di-GMP levels in the divK background, even though some strains suppressed the motility defect while others did not. These results show a discrepancy between phenotypic effects and intracellular c-di-GMP levels.

FIG 5

Artificial manipulation of c-di-GMP levels do not significantly affect phenotypes in the divK mutant. (A) Swarm expansion and surface adhesion of strains that have altered c-di-GMP levels caused by expression of nonnative enzymes in wild-type and divK mutant backgrounds. Constructs, including the E. coli diguanylate cyclase gene ydeH expressed from a medium-copy-number plasmid (med DGC) and a low-copy-number plasmid (low DGC) and the P. aeruginosa phosphodiesterase gene pchP (PDE), as well as a catalytically inactive variant (inactive PDE), were utilized. Bars below the x axis outline inducer used for plasmids in each strain. In the wild-type background, the medium-copy-number DGC increased motility and decreased adhesion, which is opposite the expected outcome, whereas the PDE reduced motility and severely reduced adhesion. In the divK background, no expression construct significantly altered the phenotypes. *, motility and adhesion were statistically insignificant from the control strain (P > 0.05). (B) c-di-GMP levels were measured using mass spectrometry and then normalized to the amount of biomass from each sample. In the wild-type background, the medium-copy-number DGC significantly increased c-di-GMP levels, whereas the PDE reduced c-di-GMP levels. In the divK background, both DGC constructs increased c-di-GMP levels, although PDE expression has no effect, despite the fact that neither DGC construct has an effect on motility and adhesion phenotypes.

FIG 6

Disruption of pleD does not alter pleC or divK motility or adhesion but does alter divJ motility and adhesion. Wild-type, pleD, divJ, divJ pleD, divK, divK pleD, pleC, and pleC pleD B. subvibrioides strains were analyzed for swarm expansion (A) and adhesion (B) defects using a soft agar swarm assay and a short-term adhesion assay, respectively. Mutant strains were normalized to wild-type results for both assays. The pleD mutation has no effect on the adhesion or motility of the pleC or divK strains but does reduce adhesion and increase motility of the divJ strain. In panel A, the bar with an asterisk indicates that divJ pleD::pCR has statistically significantly more swarm expansion than does the divJ strain (P < 0.05). In panel B, results with the same number of asterisks are not statistically significantly different from each other but are statistically significant compared to results with a different number of asterisks (P < 0.05).

FIG 7

Disruption of pleD does not alter pleC or divK holdfast production or phage sensitivity but does alter divJ holdfast production. (A) Lectin staining of holdfast material of wild-type, pleD, divJ, divJ pleD, divK, divK pleD, pleC, and pleC pleD B. subvibrioides strains. Wild-type and divJ strains have easily detectable holdfast material. pleC and pleC pleD strains have greatly reduced but still detectable holdfast material, while all remaining strains have no detectable holdfast. This includes the divJ pleD strain, which still displays obvious cell filamentation though no longer producing holdfast. (B) B. subvibrioides wild-type, pleD, divJ, divJ pleD, divK, divK pleD, pleC, and pleC pleD strains were tested with phage Delta in soft agar phage assays. While the pleD disruption alters the adhesion and holdfast phenotypes of the divJ strain, this mutation does not alter the phage sensitivity of the parent, since both divJ and divJ pleD strains have sensitivities to the phage similar to that of the wild type.