Burkholderia cenocepacia integrates cis-2-dodecenoic acid and cyclic dimeric guanosine monophosphate signals to control virulence

Significance

Many bacteria sense their population density by a process called quorum sensing (QS). In addition, individual bacterial cells sense their chemical and physical environment by adjusting levels of the intracellular compound cyclic diguanosine monophosphate (c-di-GMP). Here, we show that RpfR, a QS signal receptor protein from the pathogenic bacterium Burkholderia cenocepacia, forms a complex with c-di-GMP and a regulator named GtrR. This complex is not proficient to activate expression of virulence genes. However, upon binding its QS signal, RpfR degrades c-di-CMP, leading to activation of gene expression by a RpfR–GtrR complex. This work describes a system where a pathogen uses a single protein to integrate information about its physical and chemical surroundings and its population density to control virulence.

Abstract

Quorum sensing (QS) signals are used by bacteria to regulate biological functions in response to cell population densities. Cyclic diguanosine monophosphate (c-di-GMP) regulates cell functions in response to diverse environmental chemical and physical signals that bacteria perceive. In Burkholderia cenocepacia, the QS signal receptor RpfR degrades intracellular c-di-GMP when it senses the QS signal cis-2-dodecenoic acid, also called Burkholderia diffusible signal factor (BDSF), as a proxy for high cell density. However, it was unclear how this resulted in control of BDSF-regulated phenotypes. Here, we found that RpfR forms a complex with a regulator named GtrR (BCAL1536) to enhance its binding to target gene promoters under circumstances where the BDSF signal binds to RpfR to stimulate its c-di-GMP phosphodiesterase activity. In the absence of BDSF, c-di-GMP binds to the RpfR-GtrR complex and inhibits its ability to control gene expression. Mutations in rpfR and gtrR had overlapping effects on both the B. cenocepacia transcriptome and BDSF-regulated phenotypes, including motility, biofilm formation, and virulence. These results show that RpfR is a QS signal receptor that also functions as a c-di-GMP sensor. This protein thus allows B. cenocepacia to integrate information about its physical and chemical surroundings as well as its population density to control diverse biological functions including virulence. This type of QS system appears to be widely distributed in beta and gamma proteobacteria.

Quorum sensing (QS) is a cell-to-cell communication system that is widely employed by bacterial cells to coordinate behaviors that are advantageous to populations of cells, including exoenzyme production, biofilm formation, and antibiotic production (1–3). QS involves the production and detection of diffusible signal molecules and the initiation of appropriate responses in a cell density-dependent manner (1, 4). The original concept of QS was developed based on N-acylhomoserine lactone (AHL) signal molecules, which are constitutively produced at basal levels until they reach a critical threshold concentration and then bind to and activate a cognate receptor to control target gene expression (1, 2, 5). Recently, diffusible signal factor (DSF)-type QS signals have been recognized as another important and common type of QS system (3, 6–8). DSF was originally identified in the gamma proteobacterium Xanthomonas campestris pv. campestris (Xcc) and is involved in regulating biofilm dispersal, motility, and virulence (9–11). More recently, a related signal, cis-2-dodecenoic acid, also called Burkholderia diffusible signal factor (BDSF), was identified in the human pathogen and beta proteobacterium Burkholderia cenocepacia, where it controls similar phenotypes (3, 12–15). B. cenocepacia also has an AHL QS system composed of CepI and CepR proteins and N-octanoyl homoserine lactone (C8-HSL). The two QS systems have both distinct and overlapping effects on gene expression (16, 17).

One of the ways in which QS systems can act is by controlling levels of intracellular cyclic diguanosine monophosphate (c-di-GMP) in bacteria (18–21). c-di-GMP regulates various biological functions such as motility, biofilm formation, and virulence by a variety of mechanisms including binding to effector proteins that are parts of flagella or exopolysaccharide secretion systems, by binding to transcription factors, and by binding to riboswitches (19, 21–25). c-di-GMP is synthesized by diguanylate cyclases with GGDEF domains and degraded by phosphodiesterases with EAL or HD-GYP domains (18, 25, 26). In DSF-type QS, perception of DSF by the transmembrane receptor RpfC stimulates its autophosphorylation and phosphotransfer to RpfG, a HD-GYP domain-containing protein, to activate its c-di-GMP phosphodiesterase activity (18, 19, 27–30). Low intracellular levels of c-di-GMP in turn allow the c-di-GMP–binding protein Clp to become an active transcription factor that turns on gene expression (19). The BDSF-type QS system found in B. cenocepacia also controls intracellular levels of c-di-GMP, but by a totally different mechanism. In this system, a soluble receptor protein named RpfR that includes PAS, GGDEF, and EAL domains, degrades c-di-GMP when it binds BDSF (16, 24); however, the downstream events were unclear.

Here we identify a global regulator, which we name GtrR, that binds to RpfR and controls expression of genes in B. cenocepacia when BDSF levels are relatively high. In this circumstance, intracellular c-di-GMP levels are low due to the phosphodiesterase (PDE) activity of RpfR-BDSF. In the absence of BDSF, RpfR has low PDE activity, and a GtrR-RpfR-c-di-GMP complex forms that is not proficient to regulate gene expression. Thus, RpfR serves as a sensor of both BDSF and c-di-GMP. RpfR and GtrR homologs are present in diverse Gram-negative bacteria, suggesting that the BDSF-type QS system is widespread.

Results

GtrR Is a Global Regulator That Controls BDSF-Regulated Phenotypes in B. cenocepacia.

To identify regulatory components of the BDSF QS system, we screened a Tn5 mutant library of B. cenocepacia strain H111 carrying a lectin-encoding bclACB operon-lacZ promoter fusion plasmid for colonies that were light blue on LB plates supplemented with X-gal. The bclACB operon is controlled by both BDSF and C8-HSL (17, 31). We screened ∼40,000 colonies and identified mutants with insertions in the known regulatory genes rpfR and cepR. In addition, we identified a gene annotated as a Fis family transcriptional regulator (BCAL1536; I35_RS07130). We named this regulator global transcriptional regulator downstream RpfR (GtrR). GtrR has an AAA+ATPase σ⁵⁴-interaction domain and a C-terminal helix-turn-helix DNA-binding motif. It has previously been reported to be regulated by the BDSF system (17) and is important for survival of B. cenocepacia K56-2 in a rat lung infection model (32). An in-frame deletion mutant of gtrR that we constructed had reduced motility and biofilm formation, two phenotypes controlled by BDSF (Fig. 1). We then compared the transcriptome profiles of the wild-type strain and the gtrR mutant by RNA-Seq and saw changes in the expression levels of several hundred genes (SI Appendix, Table S1). These genes are associated with a range of biological functions, including motility and cell attachment, stress tolerance, virulence, regulation, transcriptional regulators, membrane components, transport, multidrug resistance, detoxification, and signal transduction (SI Appendix, Table S1).

Fig. 1.

Effects of GtrR on BDSF-regulated phenotypes. The wild-type strain, a gtrR mutant, and a complemented gtrR mutant were tested for the BDSF-regulated phenotypes of swarming motility (A) and biofilm formation in microtiter trays (B). The data shown are the mean of three replicates, and error bars indicate the SD.

GtrR Is a Downstream Component of the BDSF QS System.

The expression of gtrR in trans in the rpfR mutant fully restored the BDSF-regulated phenotypes of motility (Fig. 2A) and biofilm formation (Fig. 2B). Based on this, we compared the transcriptome profiles of the BDSF receptor rpfR mutant and the gtrR mutant and found substantial overlap in the genes that were affected in expression (SI Appendix, Table S1). Quantitative RT-PCR analysis of the altered expression of select genes confirmed the RNA-Seq results (SI Appendix, Table S2). From this we concluded that GtrR is likely a key downstream component of the BDSF-signaling system in B. cenocepacia.

Fig. 2.

Complementation of the rpfR mutant with rpfR and gtrR. In trans expression of RpfR and GtrR complemented swarming motility (A) and biofilm formation in microtiter trays (B) in the RpfR-deficient mutant. The data shown are the mean of three replicates, and error bars indicate the SD.

GtrR Regulates Target Gene Expression by Directly Binding to Promoters.

To study regulation by GtrR, we constructed PbclACB-lacZ and PcepI-lacZ reporter systems in a gtrR mutant (SI Appendix, Table S3). Both of these target operons are positively controlled by BDSF (SI Appendix, Tables S1 and S2) (16, 17). In agreement with the RNA-Seq and qPCR results, deletion of gtrR resulted in reduced expression levels of both bclACB and cepI (Fig. 3 A and B). To test if transcriptional regulation of these target operons is achieved by direct binding of GtrR to their promoters, electrophoretic mobility shift analyses (EMSA) were performed. A PCR-amplified 506-bp DNA fragment from the bclACB promoter and a 210-bp DNA fragment from the cepI promoter were used as probes (SI Appendix, Table S4). GtrR, which has 463 amino acids and a calculated molecular weight of 51 kDa, was purified using affinity chromatography (SI Appendix, Fig. S1A). As shown in Fig. 3 C and D, the bclACB and cepI promoter DNA fragments formed stable DNA-protein complexes with GtrR and migrated at slower rates than unbound probes. The amount of the probe that bound to GtrR increased with increasing amounts of GtrR (Fig. 3 C and D). The amount of labeled probe bound to GtrR decreased in the presence of 100 times unlabeled probe (SI Appendix, Fig. S2).

Fig. 3.

Effects of GtrR at BDSF-controlled promoters. The effects of GtrR on bclACB (A) and cepI (B) gene expression were measured by assessing β-galactosidase activities of the bclACB-lacZ and cepI-lacZ transcriptional fusions in the H111 wild-type, ΔrpfR, and ΔgtrR strains. (C) EMSA analysis of in vitro binding of GtrR to the promoter of bclACB, in which a ³²P-labeled 506-bp bclACB promoter DNA probe was used for protein-binding assay. A protein–DNA complex, represented by a band shift, was formed when different concentrations of GtrR protein were incubated with the probe at room temperature for 20 min. EMSA analysis of the in vitro binding of GtrR to the cepI promoter (D). Identical to C, a ³²P-labeled 210-bp cepI promoter DNA probe was incubated with GtrR.

RpfR Enhances GtrR Binding to Target Promoter DNA.

Although GtrR appeared to be a downstream component of the BDSF system, the relationship between BDSF sensing by RpfR and activation of gene expression by GtrR was unclear. To explore the possibility that these two proteins might interact, we purified each of them (SI Appendix, Fig. S1) and used microscale thermophoresis (MST) to see if they had an affinity for each other. As shown in Fig. 4A, RpfR interacted with GtrR with an estimated dissociation constant (K_D) of 3.98 ± 0.33 μM, and the association of the two proteins with each other was not inhibited by addition of either BDSF or c-di-GMP (SI Appendix, Fig. S3). We then tested if RpfR affects GtrR binding to target promoter DNA by performing EMSA. As shown in Fig. 4 B and C, the binding of the GtrR to the bclACB and cepI promoter probes was enhanced when RpfR was present in the reaction mixtures. The amounts of the probes that bound to GtrR increased with increasing amounts of RpfR.

Fig. 4.

Interactions between RpfR and GtrR. MST analysis indicated that GtrR binds to RpfR with a K_D of 3.98 ± 0.33 µM (A). Effect of RpfR on binding of GtrR to the bclACB promoter (B) and cepI promoter (C). Bacterial two-hybrid analysis of interaction of RpfR and GtrR in vivo (D). “Fnorm (1/1000)” indicates that the fluorescence time traces changes in the MST response.

To confirm our observed interaction between RpfR and GtrR, we used a bacterial two-hybrid system to detect protein–protein interactions (33). A positive interaction between gtrR fused to the alpha subunit RNA polymerase gene and rpfR fused to the lambda repressor protein was revealed by growth on His-deficient medium containing 5 mM 3-amino-1,2,4-triazole (3-AT) and 12.5 μg/mL streptomycin (Fig. 4D).

RpfR Binds c-di-GMP.

Since RpfR has c-di-GMP PDE activity, it can be surmised that it binds c-di-GMP. To quantify this, we performed MST and we also tested if GtrR could bind c-di-GMP. As shown in Fig. 5, there was no detectable binding between c-di-GMP and GtrR, but c-di-GMP bound to RpfR with an estimated dissociation constant (K_D) of 2.92 ± 0.26 μM. In addition, we purified polypeptides encompassing only the PAS-GGDEF or EAL domains of RpfR and tested their binding to c-di-GMP. We found that only the EAL domain was required for c-di-GMP binding (SI Appendix, Fig. S4). This finding agrees with our previous results in which the EAL domain of RpfR was required for c-di-GMP PDE activity (24).

Fig. 5.

MST analysis of the binding of c-di-GMP to RpfR (A) and GtrR (B). The K_d for the binding between c-di-GMP and RpfR was calculated to be 2.92 ± 0.264 μM. No binding of c-di-GMP to GtrR was detected. “Fnorm (1/1000)” indicates the fluorescence time traces changes in the MST response.

c-di-GMP Inhibits Binding of the RpfR–GtrR Complex to Target Promoter DNA.

To determine how the binding of c-di-GMP to RpfR might affect the activity of GtrR, we examined the effects of c-di-GMP on binding of the RpfR–GtrR complex to the bclACB promoter by EMSA. The addition of c-di-GMP at a relatively high concentration (150 µM) resulted in a decrease in the ability of RpfR–GtrR to bind target promoter DNA (Fig. 6A). RpfR did not retain its c-di-GMP PDE activity in the EMSA. In contrast, the addition of BDSF did not affect binding of the RpfR–GtrR complex to target promoter DNA (SI Appendix, Fig. S5). We obtained the same result when we used the cepI promoter probe in gel shift assays (Fig. 6B). c-di-GMP did not affect GtrR binding to target promoters in the absence of RpfR (SI Appendix, Fig. S6). These results are consistent with the idea that binding of c-di-GMP by RpfR lowers the affinity of the RpfR–GtrR complex for target promoter DNA, ultimately decreasing target gene expression. Neither BDSF nor c-di-GMP affected the binding of RpfR to GtrR in vitro (SI Appendix, Fig. S3). To further assess if c-di-GMP could disrupt the RpfR–GtrR interaction, we carried out the bacterial two-hybrid assay described above and in Fig. 4D, but with the additional expression of the diguanylate cyclase gene, wspR. As expected, in trans expression of wspR caused a significant increase in intracellular c-di-GMP levels in the Escherichia coli reporter strain (SI Appendix, Fig. S7). However, this did not cause a significant disruption of the RpfR–GtrR interaction, as indicated by the ability of cells to grow on M9 medium supplemented with streptomycin and 3-AT. Although c-di-GMP does not affect RpfR–GtrR complex formation in this assay, we cannot exclude that binding of c-di-GMP to RpfR promotes conformational changes in the complex.

Fig. 6.

The effects of c-di-GMP on RpfR–GtrR complex binding to the promoter DNA of bclACB (A) and cepI (B) were assessed by performing EMSA.

To characterize the relationship between c-di-GMP PDE activity and c-di-GMP sensing, we created point mutations in the EAL domain of RpfR to see if we could disrupt c-di-GMP PDE activity but not the binding of c-di-GMP to RpfR. Among nine RpfR variant proteins that we tested, six completely lost their ability to degrade c-di-GMP. The three remaining variant proteins exhibited c-di-GMP PDE activity (SI Appendix, Table S5). We identified one variant, RpfR^E466A, that had a similar binding affinity for c-di-GMP as the wild-type protein, but had no c-di-GMP PDE activity (SI Appendix, Fig. S8 and Table S5). Inclusion of c-di-GMP in EMSA assays carried out with the RpfR^E466A variant at similar concentrations, as used in EMSA assays with wild-type RpfR, inhibited binding of the RpfR^E466A–GtrR complex to target promoter DNA (SI Appendix, Fig. S9).

GtrR Contributes to B. cenocepacia Pathogenicity.

Previous studies characterized the attenuated virulence of BDSF synthase- and receptor-negative mutants of B. cenocepacia (13, 24, 31). To investigate whether GtrR was required for virulence, we tested both rpfR and gtrR mutants in cell line and mouse models. As expected, deletion of either rpfR or gtrR led to a reduction in bacterial virulence in both models (Fig. 7). This result is consistent with previous findings (32). Cytotoxicity was measured by quantifying the release of lactate dehydrogenase (LDH) into the supernatants of cultured cells. When cells were incubated with rpfR and gtrR mutant strains, cytotoxicity levels were 46 and 51%, respectively, of the levels achieved when cells were incubated with the wild-type B. cenocepacia H111 strain at 8 h postinoculation (Fig. 7A). Mouse model assays demonstrated similar results. The mortality of mice infected with the wild-type strain or the rpfR or the gtrR mutant strains was 80, 20, and 30%, respectively, at 5 d postinfection (Fig. 7B). Complementation of the gtrR mutant with a cloned gtrR gene restored the pathogenicity of the mutant to wild-type levels in both the cell line and mouse models. Further investigation was conducted to analyze the symptoms of mouse lungs infected by the mutant strains because the lung is the most important niche for B. cenocepacia. At 2 d postinfection, the macroscopic pathological findings indicated severe pulmonary inflammation when the mice were infected with the wild-type or complemented mutant strains, but there was no or only mild inflammation with decreased inflammatory cell infiltration and serosanguinous exudation when the mice were infected with the rpfR and gtrR mutants (SI Appendix, Fig. S10).

Fig. 7.

The effects of gtrR and rpfR on the pathogenicity of B. cenocepacia were determined using cell line (A) and mouse infection (B) models. Cell cytotoxicity was detected and measured as LDH release. LDH release by B. cenocepacia wild-type strain H111 was arbitrarily defined as 100% and used to normalize the LDH release ratios of the gtrR and rpfR mutants and the complemented strains. The data shown are the mean of three replicates, and error bars indicate the SDs. For the mouse model, mortality was calculated after BALB/c mice were infected using 1 × 10⁹ cfu of B. cenocepacia strains over a 5-d period.

Discussion

From the results of this study, we can put together a model for how BDSF-mediated signaling controls gene expression in B. cenocepacia. At low population densities when the amount of BDSF is low and c-di-GMP is present at basal levels in cells, c-di-GMP binds to RpfR–GtrR complexes, and this prevents GtrR from activating transcription. At high population densities, when the concentration of BDSF that cells are exposed to is relatively high, BDSF binds to RpfR to stimulate its c-di-GMP phosphodiesterase activity. As a result, the concentration of intracellular c-di-GMP drops to a level sufficiently below its binding constant for RpfR as to allow RpfR and GtrR to form a transcriptionally active complex (Fig. 8). Several QS systems modulate c-di-GMP levels in response to exposure to a QS signal, including the RpfC–RpfG system in Xcc (19) and the LuxPQ and CpqS systems in Vibrio cholerae (34). However, the B. cenocepacia BDSF QS system is distinctive in that it regulates c-di-GMP levels in the cells by degrading this molecule in response to BDSF and its activity is negatively controlled by c-di-GMP. Moreover, a single protein, RpfR, integrates sensory input from c-di-GMP and BDSF. One can imagine that this circuitry endows the B. cenocepacia BDSF QS system with the ability to fine-tune gene expression depending on the relative concentrations of BSDF and c-di-GMP that RpfR encounters in cells. B. cenocepacia H111 encodes six c-di-GMP phosphodiesterases, 12 diguanylate cyclases, and, in addition to RpfR, four hybrid diguanylate cyclase/phosphodiesterase proteins and two HD-GYP domain proteins (potential c-di-GMP phosphodiesterases). As has been demonstrated in many other bacteria, a subset of these proteins likely senses the physical and chemical characteristics of the extracellular environment of B. cenocepacia to control the total intracellular concentration of c-di-GMP. Thus, one would expect that there would be environmental circumstances where low intracellular c-di-GMP levels would allow for RpfR-GtrR–directed gene expression to occur in the absence of BDSF. In fact, it has been shown that QS phenotypes of the BDSF synthase mutant rpfF_Bc can be rescued by in trans expression of a c-di-GMP phosphodiesterase (24).

Fig. 8.

Model for BDSF-mediated QS regulation of virulence at low (A) and high (B) cell population densities. RpfF_Bc is involved in synthesis of BDSF signals. At low cell density, a pool of intracellular c-di-GMP is determined by the activities of various diguanylate cyclases and phosphodiesterases in response to physical and chemical cues from the environment. When c-di-GMP is sufficiently high, it binds to RpfR–GtrR and prevents this complex from controlling gene expression. At high cell density, perception of BDSF by RpfR substantially enhances its c-di-GMP phosphodiesterases activity, causing a reduction of the intracellular c-di-GMP to a level sufficient to allow a RpfR–GtrR complex to bind to the target promoter DNA and enhance target gene expression and virulence in B. cenocepacia.

In contrast to the regulatory model that we have just presented for BDSF QS, the DSF-type QS system in Xcc and related bacteria operates with a different set of signal perception and regulatory proteins to regulate intracellular c-di-GMP levels, and its activity is not modulated by c-di-GMP as the BDSF-type QS system is (16, 19, 24, 27–29). The remarkable differences between the QS perception and c-di-GMP sensing mechanisms suggest that the DSF system in Xcc and the BDSF system in B. cenocepacia have evolved independently. The rpfF/rpfC/rpfG gene cluster of the DSF-type QS system is conserved in bacterial species belonging to the genera Xanthomonas, Xylella, Stenotrophomonas, Methylobacillus, Thiobacillus, and Leptospirillum (6). A BLAST search with the B. cenocepacia rpfF_Bc/rpfR gene cluster and gtrR revealed that the BDSF-type QS system likely operates in a different set of bacteria, including members of the genera Burkholderia, Paraburkholderia, Achromobacter, Yersinia, Serratia, Enterobacter, Pantoea, and Cronobacter (SI Appendix, Table S6).

In conclusion, our working model of the RpfR–GtrR complex responds to both BDSF signal and c-di-GMP to control gene expression. Our findings will inform further studies seeking to develop new strategies to target the multifunctional protein RpfR, which senses both QS signals and an intracellular second messenger to control diseases caused by bacterial pathogens.

Materials and Methods

Bacterial Growth Conditions and Virulence Assays.

The bacterial strains used in this work are listed in SI Appendix, Table S3. B. cenocepacia H111 strains were cultured at 37 °C in medium comprised of 5 g peptone (Difco), 3 g yeast extract (Difco), and 20 g glycerol per liter (35) or LB Lennox. The following antibiotics were added when necessary: tetracycline, 100 µg mL⁻¹; trimethoprim, 25 µg mL⁻¹; and gentamicin, 10 µg mL⁻¹. Cell line and mouse infection models were used for virulence assays following the methods indicated in SI Appendix.

Mutagenesis and Phenotype Analysis.

B. cenocepacia H111 was used as the parental strain to generate an in-frame deletion mutant of gtrR, following the methods described previously (24). Swarming motility and biofilm formation assays were performed using previously described methods (36). Detailed descriptions of the above methods are provided in SI Appendix.

Protein Purification and Analysis.

Detailed descriptions are provided in SI Appendix. Briefly, rpfR and gtrR were amplified with the primers listed in SI Appendix, Table S4, and cloned into the expression vector pET-28a. The fusion gene constructs were transformed into E. coli strain BL21. Affinity purification of HIS-GtrR and HIS-RpfR fusion proteins and binding assays were performed following previously described methods (24).