The Diguanylate Cyclase HsbD Intersects with the HptB Regulatory Cascade to Control Pseudomonas aeruginosa Biofilm and Motility

Abstract

The molecular basis of second messenger signaling relies on an array of proteins that synthesize, degrade or bind the molecule to produce coherent functional outputs. Cyclic di-GMP (c-di-GMP) has emerged as a eubacterial nucleotide second messenger regulating a plethora of key behaviors, like the transition from planktonic cells to biofilm communities. The striking multiplicity of c-di-GMP control modules and regulated cellular functions raised the question of signaling specificity. Are c-di-GMP signaling routes exclusively dependent on a central hub or can they be locally administrated? In this study, we show an example of how c-di-GMP signaling gains output specificity in Pseudomonas aeruginosa. We observed the occurrence in P. aeruginosa of a c-di-GMP synthase gene, hsbD, in the proximity of the hptB and flagellar genes cluster. We show that the HptB pathway controls biofilm formation and motility by involving both HsbD and the anti-anti-sigma factor HsbA. The rewiring of c-di-GMP signaling into the HptB cascade relies on the original interaction between HsbD and HsbA and on the control of HsbD dynamic localization at the cell poles.

Fig 1. Occurrence of PA3343 (hsbD) in Pseudomonas strains and its relation with the flagellar genes reorganization in P. aeruginosa.

(A) Evolutionary relationships of Pseudomonas taxa carrying hptB orthologs and taxonomic distribution of PA3341-PA3353 genes. The phylogenetic tree of Pseudomonas strains was constructed using MEGA 6 [58]. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. Filled squares indicate the presence of an ortholog gene, while empty squares indicate the absence. Location of the PA3341-PA3353 genes is shown as in P. aeruginosa. When a green or blue background is used, it indicates that each gene set is present, but at different location on the chromosome, as explained in panel B. Non-aeruginosa strains are separated by a dashed line. (B) Location of flagella and hptB related genes in Pseudomonas. In P. aeruginosa (left) flagella genes are located in three regions of the chromosome [68] while in other Pseudomonas species (right) two regions are present and in general physically separated with some exceptions (e.g. Pseudomonas entomophila L48) [26].

Fig 2. HsbD is a diguanylate cyclase which activity intersects with the HptB regulatory pathway.

(A) Top panels show E.coli expressing HsbD and a HsbD variant with a mutated active site (A-site) and detection of DGC activity by Congo red binding. Lower graph shows detection of HsbD binding to ³²P-labeled c-di-GMP using DRaCALA performed with E. coli extracts prepared from strains carrying cloning vector (ev), or recombinant plasmids overexpressing His-HsbD, or His-HsbD variant with a mutated inhibitory site (I-site). In all strains the expression and stability of HsbD and variants was prior verified by SDS-page and Western blot. The chart shows the quantification of the fraction of ³²P-c-di-GMP bound to the protein spot on the nitrocellulose membrane from three independent experiments (Student t-test, ***, p ≤ 0.0001). (B) Graph depicting quantification of c-di-GMP levels in P. aeruginosa PAK wild type (WT) and PAK ΔhsbD, ΔhptB, ΔhptBΔhsbD mutant strains by LC-MS/MS. Data are expressed as picomoles of c-di-GMP per mg of total protein (see Materials and Methods). Each value is the average of three different cultures ± standard deviation (Student t-test, *, p < 0.05). (C-E) Experiments performed in P. aeruginosa PAK wild type (WT) and PAK ΔhsbD, ΔhptB, ΔhptBΔhsbD, ΔretS and ΔretSΔhsbD mutant strains. (C) Expression of a cdrA-gfp reporter fusion (indicative of c-di-GMP levels) measured in P. aeruginosa strains grown to OD₆₀₀ ~2.0 in LB medium. Relative fluorescence units (RFU) are corrected for background (empty vector) and for cell density as described in the Materials and Methods. (D) Biofilm formation of P. aeruginosa strains measured by crystal violet staining. Bacterial strains were grown in LB medium in 24-microtiter plates for 14 hours. (E) β-Galactosidase activity of a rsmY–lacZ transcriptional fusion in P. aeruginosa strains grown in rich LB medium to an OD600~2.0, as described in Bordi et al. (2010). (F) Biofilm formation, measured by crystal violet staining, of PAK WT, ΔhptB, ΔhsbD and ΔhptBΔhsbD strains grown in 24-microtiter plates for 8 hours; carrying either a pMMBRMCS4 empty plasmid (grey columns, -) or the pBBR3347 plasmid overexpressing hsbR (black columns, +). Each value is the average of three different cultures ± standard deviation (Student t-test, **, p < 0.01).

Fig 3. Interaction of HsbD with HsbA is dependent on HsbA phosphorylation.

(A) Reconstitution of adenylate cyclase in the E. coli strain DHM1 using a bacterial two-hybrid approach (see S1 Table) was detected by blue staining due to X-gal hydrolysis when colonies where grown on LB–X-gal agar plates containing 0.5 mM IPTG, 100 μg/ml ampicillin, and 50 μg/ml chloramphenicol agar plates. The interactions were also quantified by β-galactosidase assays using liquid cultures of the same strains. Each value is the average of three different cultures ± standard deviation (*, p < 0.05; **, p < 0.01). (B) Dot blot analysis of HsbD-HsbA interaction in P. aeruginosa PAK. PAK wild type (WT) cell lysates overexpressing HsbD (FLAG-tagged), SadC, HsbR (prey proteins) are spotted on a membrane and incubated with purified HA-tagged HsbA, HsbA_S56D and HsbD_S56A proteins (bait proteins). Detection of HsbA bait proteins bound to HsbD prey on the blot was performed using α-HA antibody. Empty vector was used as negative control. Production of HsbD-FLAG/HsbA-HA variants is shown by Western blot. (C) Biofilm formation of P. aeruginosa wild-type PAK, ΔhsbA, ΔhsbD and ΔhsbAΔhsbD strains measured by crystal violet staining. Bacterial strains were expressing in trans either HsbA, HsbA_S56D or HsbA_S56A and grown in LB medium in 24-microtiter plates for 14 hours in presence of 50 μg/ml gentamycin and 0.5 mM IPTG. ev: empty vector (pME6032, see S1 Table). Each value is the average of three different cultures ± standard deviation. Asterisks indicate statistically significant difference of biofilm formation compared to the ev (*, p< 0.05; **, p< 0.01).

Fig 4. A ΔhptBΔhsbD mutant is hyper-swarming.

Swarming motility of PAK wild type (WT) and PAKΔhptB, ΔretS, ΔrsmA, ΔhsbD, ΔhptBΔhsbD, ΔretSΔhsbD, ΔrhlA mutant strains growing in minimal medium supplemented with glucose and casamino-acids (A) or nutrient agar plus glucose (B). Surface area covered by the swarming cells (± standard deviation) was calculated by averaging data from four individual swarm plates (Student t-test, *, p< 0.01; **, p < 0.005; ****, p < 0.0001). (C) Fold reduction of swarming motility due to HsbA_S56D or HsbA_S56A overexpression relative to strains carrying the empty vector (ev) in ΔhsbA (white columns) and ΔhsbAΔhsbD (grey columns) strains. Bacterial strains were grown in in minimal medium supplemented with glucose and casamino-acids.

Fig 5. A ΔhptBΔhsbD mutant is impaired in twitching motility.

(A) Twitching motility of PAK wild-type (WT) and ΔretS, ΔhptB, ΔhsbD, ΔretSΔhsbD, ΔhptBΔhsbD mutant strains. Twitching zones are visualized by crystal violet staining as indicated in Materials and Methods. At least three independent experiments were performed. Twitching diameters are indicated at the bottom of each panel ± standard deviation. (B) Analysis of PilA production and localization via Western blotting. Sheared-surface type IV pili (SP) and whole-cell extracts (WC) with identical samples were visualized via Western blotting. The following strains were analyzed: PAK wild type and PAKΔhptB, ΔhsbD, ΔhptBΔhsbD, ΔpilA. (C) TEM analysis of PAK wild-type (WT) and ΔhptBΔhsbD mutant strains. Representative cells are shown entirely and insets illustrate higher magnification. Scale bar, 1000 nm and 200 nm, respectively. Closed arrowhead denotes flagellum, open arrowheads indicate representative type IV pili.

Fig 6. HsbD is regulating chemotaxis motility.

(A) Swimming motility of PAK wild type (WT) and ΔretS, ΔhptB, ΔhsbD, ΔretSΔhsbD, ΔhptBΔhsbD mutant strains. White dashed circles correspond to the diameter of the WT strain. At least three independent experiments were performed. (B) Chemotactic response of PAK WT, hptB, hsbD and hptB/hsbD strains using 0.5% of casaminoacids as chemoatractant (black histograms) or PBS as control. The duration of chemotactic incubation time is 20 min. Each value is the average of three separate assay ± standard deviation (*, p < 0.05; **, p < 0.01).

Fig 7. Role of HsbD and SadC in the HptB pathway.

(A) Relative fold change difference in biofilm formation, swarming, twitching and swimming motilities of either the PAKΔhptBΔsadC (black) or ΔhptBΔhsbD (grey) mutant compared to the ΔhptB mutant strain (ns: not significant). (B) Twitching motility of PAK ΔhptB and ΔhptBΔhsbD mutant strains carrying either a pMMBRMCS4 empty plasmid (-) or the pBBR3347 plasmid overexpressing sadC (+). Each value is the average of three different cultures ± standard deviation (Student t-test, **, p < 0.01).

Fig 8. HsbD polar localization in P. aeruginosa.

(A) Localization of HsbD-YFP (green) in cells grown in LB medium and induced with 100μM IPTG for two hours (left panel). Three representative cells (right panel) where chosen for the quantification of the fluorescence intensity across the cell contour. Black line: cell I, red: cell II and green: cell III. A pseudo-colored fluorescence image (green YFP) of each cell is shown on top of the graph. Cartoon: representation of coordinates reported in the graph. Scale bar = 3 μm (B) Localization of the GGDEF-containing domain of HsbD (HsbD_C-ter-YFP) in cells grown in LB medium and induced with 100μM IPTG for two hours. First column shows fluorescence images (YFP) while the second column the overlay of the fluorescence channel (in green) and the phase contrast image. Scale bar = 2 μm. (C) Representative cells where chosen to illustrate the three localization patterns of HsbD_C-ter-YFP. Scale bar = 1 μm (D) Quantification of the distinct HsbD_C-ter-YFP localization patterns. Error bars represent the standard deviation (n = 3 replicates of more than 200 cells each). At least three independent experiments were performed.

Fig 9. Dynamics of HsbD and FhlF cellular (co-)localization.

(A) Localization of HsbD-YFP (green) and FhlF-RFP (red) in a P. aeruginosa wild-type strain. Scale bar = 2 μm. Bottom panel: three representative patterns (1:1, 1:2 foci and 2:2 YFP:RFP foci) are illustrated in a closer view. Scale bar = 1 μm. Cell boundaries are delineated in white. (B) HsbD and FhlF polar localization frequency and their colocalization. More than 200 cells were analyzed from different fields. At least three independent experiments were performed. (C) Dynamics of HsbD-YFP (green) and FhlF-RFP (red). An overlay of the fluorescence channel(s) and the phase contrast image illustrate HsbD-YFP and FlhF-RFP subcellular localization pattern. Fluorescence images are shown in S6 Fig. Scale bar = 1 μm. Bottom panel shows the quantification of the HsbD-YFP (green) and FhlF-RFP (red) fluorescence intensity across the cell length. Cartoon: representation of coordinates reported in the graph.

Fig 10. HsbD polar localization in a P. aeruginosa ΔhptB mutant.

(A) Localization of the GGEEF-containing domain of HsbD (HsbD_C-ter-YFP) in cells grown in LB medium and induced with 100μM IPTG for two hours. Scale bar = 2 μm. (B) Quantification of the distinct HsbD_C-ter localization patterns in cell population. Error bars represent the standard deviation (n = 3 replicates of more than 200 cells each). (C) Localization of HsbD_C-ter-YFP (green) and FhlF-RFP (red) in a ΔhptB mutant strain. Scale bar = 2 μm (D) Three representative patterns are illustrated in a closer view. Scale bar = 1 μm.

Fig 11. Working model for the HptB signaling pathway in P. aeruginosa.

(A) HptB is activated (i.e. phosphorylated) via either one of three orphan sensor kinase hybrids: PA1611, ErcS’ (PA1976) or SagS (PA2824) [–28]. When phosphorylated, HptB transfers a phosphoryl group to the HsbR receiver domain (HsbR-P), thus repressing HsbR kinase activity while activating its phosphatase activity. HsbR-P dephosphorylates the anti-anti-sigma factor HsbA. Dephosphorylated HsbA leads to dissociation of the HsbA-HsbR complex and a subsequent stable sequestration of the anti-sigma factor FlgM by HsbA [20]. The interaction HsbA-FlgM induces flagellar genes expression, by allowing the release of the FliA sigma factor (σ²⁸) [51]. When HptB is inactive (dephosphorylated or in an hptB mutant), swimming and swarming are no longer supported. Instead, the HsbR kinase phosphorylates HsbA (HsbA-P) and HsbA-P interaction with HsbD leads to an increase of c-di-GMP and RsmY levels. The activation of HsbD strengthens swarming repression and results in hyper-biofilm and hyper-twitching phenotypes. This cascade of events is in agreement with the waves of regulatory actions that cause bacteria progression from an early surface attachment and colonization (swimming and swarming) to the development of mature sessile biofilms [15, 26, 56]. The symbol ➔ indicates positive regulation, while -¦ repression and de–P (de)phosphorylation. Dashed lines suggest a probable indirect regulation. Components of the HptB signaling pathway are colored in green. (B) Localization dynamics of HsbD (green square) during cell division. FhlF (red square) dynamics is reported as commented previously by Burrows LL [49]. Green gradient in the cell undergoing division illustrates bimodal distribution of c-di-GMP levels as reported by Christen and colleagues [14].? Question mark represents possible scenarios on the asymmetric partitioning of HsbD during cell division in respect to flagellum biogenesis and DipA (blue square) localization. For details see text.