Production of 3',3'-cGAMP by a Bdellovibrio bacteriovorus promiscuous GGDEF enzyme, Bd0367, regulates exit from prey by gliding motility

Abstract

Bacterial second messengers are important for regulating diverse bacterial lifestyles. Cyclic di-GMP (c-di-GMP) is produced by diguanylate cyclase enzymes, named GGDEF proteins, which are widespread across bacteria. Recently, hybrid promiscuous (Hypr) GGDEF proteins have been described in some bacteria, which produce both c-di-GMP and a more recently identified bacterial second messenger, 3',3'-cyclic-GMP-AMP (cGAMP). One of these proteins was found in the predatory Bdellovibrio bacteriovorus, Bd0367. The bd0367 GGDEF gene deletion strain was found to enter prey cells, but was incapable of leaving exhausted prey remnants via gliding motility on a solid surface once predator cell division was complete. However, it was unclear which signal regulated this process. We show that cGAMP signalling is active within B. bacteriovorus and that, in addition to producing c-di-GMP and some c-di-AMP, Bd0367 is a primary producer of cGAMP in vivo. Site-directed mutagenesis of serine 214 to an aspartate rendered Bd0367 into primarily a c-di-GMP synthase. B. bacteriovorus strain bd0367S214D phenocopies the bd0367 deletion strain by being unable to glide on a solid surface, leading to an inability of new progeny to exit from prey cells post-replication. Thus, this process is regulated by cGAMP. Deletion of bd0367 was also found to be incompatible with wild-type flagellar biogenesis, as a result of an acquired mutation in flagellin chaperone gene homologue fliS, implicating c-di-GMP in regulation of swimming motility. Thus the single Bd0367 enzyme produces two secondary messengers by action of the same GGDEF domain, the first reported example of a synthase that regulates multiple second messengers in vivo. Unlike roles of these signalling molecules in other bacteria, these signal to two separate motility systems, gliding and flagellar, which are essential for completion of the bacterial predation cycle and prey exit by B. bacteriovorus.

Fig 1. Bioinformatic analysis of Bd0367.

A- Diagrammatic representation of the domains of B. bacteriovorus Bd0367. B- Alignment of the primary amino acid sequence of the GGDEF domains of Caulobacter vibriodes (CB15) PleD, Bdellovibrio bacteriovorus Bd0367, and Geobacter sulfurreducens GacA(GSU1658). The selectivity residue (blue) differs in the Hypr proteins (Bd0367 and GSU1658) where a serine residue is present, compared to the aspartate residue at position 344 in a canonical GGDEF enzyme such as PleD. Residues involved in the GGDEF domain (green) and I-site domains (red) are also highlighted.

Fig 2. Bd0367 produces different cyclic dinucleotides.

Cellulose TLC analysis of radiolabeled products from enzymatic reactions with 1:1 ATP-to-GTP substrates in excess and doped with trace amounts of α-³²P-labeled GTP (left panel) or α-³²P-labeled ATP (right panel). Before loading, reactions were quenched with alkaline phosphatase to digest unreacted nucleotides, resulting in production of inorganic phosphate (Pi). Residue R260 of Bd0367 is located in the putative I-site. Residue S214 is located at the selectivity position identified in previous studies and is necessary for cGAMP, but not c-di-GMP, production as mutants (S214D, SD) showed no bands at the expected position (cGAMP) whereas positive controls (DncV and MBP-Bd0367 R260A) did. WspR D70E is a positive control for c-di-GMP production and negative control for cGAMP production. DisA is a positive control for c-di-AMP production and negative control for cGAMP production.

Fig 3. Cyclic nucleotide levels in B. bacteriovorus. A- cGAMP was extracted from whole B. bacteriovorus cells of wild type HID22 (WT), mutants Δbd0367 and S214D, or complemented mutants (Δbd0367+WTbd0367 and S214D+WTbd0367) and quantified using HPLC-MS/MS.

Means and standard deviations of three biological replicates are presented. Cyclic GAMP levels in the complemented mutants were not significantly different from wild-type (p = 0.9976 for Δbd0367+WTbd0367 and p = 0.9997 for S214D+WTbd0367 by one-way ANOVA and Tukey’s multiple comparison test). B- c-di-GMP was extracted from whole B. bacteriovorus cells of wild type HID22 (WT), mutant S214D, or complemented mutant S214D+WTbd0367 and quantified using HPLC-MS/MS. Means and standard deviations of three biological replicates are presented. Cyclic-di-GMP levels were not significantly different from wild-type (p = 0.0857 for S214D and p = 0.8326 for S214D+WTbd0367 by one-way ANOVA and Tukey’s multiple comparison test).

Fig 4. Testing the ability of B. bacteriovorus strains to exit from E. coli S17-1 prey on a hydrated 0.5% agarose pad.

Stills from time-lapse microscopy are shown: A. The B. bacteriovorus bd0367S214D site-directed mutant strain (S214D) invaded prey, predators elongated and septated, then bdelloplast lysis occurs, seen by rapid loss of prey fluorescence. However, no motile processive exit was observed and progeny remained in the prey remnant, as was also the case for the bd0367 deletion strain (Δbd0367). B. Replacing the mutant bd0367 genes with a wild-type copy gave strains of B. bacteriovorus which exited the bdelloplast by gliding motility as did the host-independent wild type strain HID13 (WT). The arrows highlight the new progeny B. bacteriovorus exiting the lysed bdelloplast. Scale bars are 1 μm. Images are representative of three biological repeats.

Fig 5. Plaque assay.

Spots of host-independent (axenic) B. bacteriovorus cultures OD₆₀₀ 1.0 on YPSC agar E. coli S17-1 prey overlay plates, incubated for 2 weeks. Two wild type host-independent B. bacteriovorus strains were tested (HID13 and HID22), alongside the B. bacteriovorus Δbd0367 deletion strain, the bd0367(S214D) site-directed mutant (two isolates–S214D-1 and S214D-2) and the corresponding wild type complemented strains (Δbd0367 + WTbd0367, S214D-1 + WTbd0367 and S214D-2 + WTbd0367).). Complementation with a GGAAF mutated genotype as a single crossover was also tested (Δbd0367 + bd0367GGAAF SXO) and its positive control complemented strain (Δbd0367 + WTbd0367 SXO). Prey killing can present itself in a number of ways in this assay, such as individual plaques, or large areas of prey clearing (+). Strains unable to kill prey showed a region of HI cell growth with no clearing (-). The numbers of experiments and plaques observed are presented in Fig E in S1 Text.

Fig 6. Gliding phenotype of bd0367 mutant strains.

A. Time-lapse microscopy was carried out over 7 hours, on a 1% agarose surface, to investigate the percentage of each B. bacteriovorus population able to glide after application from a liquid culture onto a solid surface. Two values are shown for each strain, one for the mean percentage of cells in the population that are able to move on a solid surface (any type of solid surface movement observed), and another for the mean percentage of cells that glide and progress further than 2 μm. Error bars show one standard deviation of three biological repeats. Both the B. bacteriovorus bd0367 deletion (Δbd0367) and bd0367(S214D) site-directed mutant strains were unable to glide. There was no significant difference in the gliding abilities of the strains that are able to glide (one-way ANOVA with Tukey’s multiple comparisons test–Cell movement: HID13 vs Δbd0367+WTbd0367, p = 0.64; HID13 vs bd0367(S214D)+WTbd0367, p = 0.94; Δbd0367+WTbd0367 vs bd0367(S214D)+WTbd0367, p = 0.8. Cell progression beyond 2μm: HID13 vs Δbd0367+WTbd0367, p = 0.35; HID13 vs bd0367(S214D)+WTbd0367, p = 0.08; Δbd0367+WTbd0367 vs bd0367(S214D)+WTbd0367, p = 0.62). Gliding of the GGAAF complemented strain (Δbd0367+GGAAF SXO) was significantly less (p = 0.0044 for cell movement, p = 0.0327 for cell progression by unpaired t-test) than the control single crossover strain (Δbd0367+WTbd0367 SXO). B. The time taken for each strain to initiate gliding (onset of gliding) over the 7 hour time-lapse. The mean onset of gliding (± standard deviation, 3 biological repeats) of the complementation strains were found to not significantly differ (Δbd0367+WTbd0367 = 2.47±0.66 hrs; bd0367(S214D)+WTbd0367 = 2.08±0.57 hrs) to the host-independent wild type strain gliding onset (HID13 = 2.62±0.6 hrs) (one-way ANOVA with Tukey’s multiple comparisons test: HID13 vs Δbd0367+WTbd0367, p = 0.95; HID13 vs bd0367(S214D)+WTbd0367, p = 0.53; Δbd0367+WTbd0367 vs bd0367(S214D)+WTbd0367, p = 0.71). Onset of gliding was not significantly different (p = 0.21 by unpaired t-test) for the GGAAF complemented strain (Δbd0367+GGAAF SXO) compared to the control single crossover strain (Δbd0367+WTbd0367 SXO).

Fig 7. RT-PCR investigating gliding gene expression on a surface or in liquid.

A- Expression of gliding genes bd3099 and bd1480 and control housekeeping gene dnaK in wild-type, HD grown, predatory B. bacteriovorus HD100 incubated on a surface for 0 (S0) or 4 hours (S4) and controls incubated in liquid (L0 and L4) show induction of the gliding genes, but not control gene dnaK at 4 hours on surface (red box). Control RNA isolated from the prey E. coli cells (S17-1) did not give products with these primers designed to specifically anneal to the Bdellovibrio genes. B- RT-PCR with RNA isolated from host independent (HI) strains incubated on a surface for 0 (S0) or 4 hours (S4) shows lower expression of predicted gliding genes in mutant strains (Δ367 and S214D) relative to wild type (HID22) and reconstructed wild-type bd0367 strains (+WT367; red boxes), but similar expression of control dnaK gene in all strains. C- HI cultured strains incubated in liquid for 0 (L0) or 4 hours (L4) also show lower expression of predicted gliding genes in mutant strains (Δ367 and S214D) relative to wild type (HID22) and reconstructed wild-type bd0367 strains (+WT367; red boxes), but similar expression of control dnaK gene. Images are representative of two independent repeats.–ve no template negative control, +ve genomic DNA positive control, M- NEB 100bp ladder; lower band 100bp, higher band 200bp. D- Quantitative RT-PCR to confirm expression pattern of bd1480. Quantification of cDNA was by comparison to purified PCR product standard curve and normalized relative to whole signal from each biological repeat.

Fig 8. Swimming phenotype of all strains complemented with fliDS in trans.

A- Cultures were grown in PY broth supplemented with gentamycin and scored for presence or absence of actively swimming cells. No swimming cells of the Δbd0367 strain were ever observed, whilst all other strains did exhibit swimming. Data are from 3 biological repeats with cultures grown above OD₆₀₀ 0.45 per repeat (dependent upon variable growth phenotype of the HI cells and cutoff for this OD₆₀₀ chosen as swimming cells were never observed in cultures of OD₆₀₀ < 0.45; see Table C in S1 Text for n values). Error bars are SEM. B- Transmission electron micrographs of cells grown after complementation with fliS supplied in trans on a plasmid. No flagella were seen on Δbd0367 cells, but flagella were seen on cells of the bd0367(S214D) strain and the strains reconstructed with wild-type bd0367 (when also complemented with fliS). Cells were stained with 0.5% uranyl acetate for 1 minute on carbon formvar grids and imaged on a FEI Technai bio-twin 12 TEM. Scale bars are 2 μm.