CosR Is a Global Regulator of the Osmotic Stress Response with Widespread Distribution among Bacteria

Vibrio parahaemolyticus can accumulate compatible solutes via biosynthesis and transport, which allow the cell to survive in high salinity conditions. There is little need for compatible solutes under low salinity conditions, and biosynthesis and transporter systems need to be repressed. However, the mechanism(s) of this repression is not known. In this study, we showed that CosR played a major role in the regulation of multiple compatible solute systems. Phylogenetic analysis showed that CosR is present in all members of the Vibrionaceae family as well as numerous Gammaproteobacteria. Collectively, these data establish CosR as a global regulator of the osmotic stress response that is widespread in bacteria, controlling many more systems than previously demonstrated.

ABSTRACT

Bacteria accumulate small, organic compounds called compatible solutes via uptake from the environment or biosynthesis from available precursors to maintain the turgor pressure of the cell in response to osmotic stress. The halophile Vibrio parahaemolyticus has biosynthesis pathways for the compatible solutes ectoine (encoded by ectABC-asp_ect) and glycine betaine (encoded by betIBA-proXWV), four betaine-carnitine-choline transporters (encoded by bccT1 to bccT4), and a second ProU transporter (encoded by proVWX). All of these systems are osmotically inducible with the exception of bccT2. Previously, it was shown that CosR, a MarR-type regulator, was a direct repressor of ectABC-asp_ect in Vibrio species. In this study, we investigated whether CosR has a broader role in the osmotic stress response. Expression analyses demonstrated that betIBA-proXWV, bccT1, bccT3, bccT4, and proVWX are repressed in low salinity. Examination of an in-frame cosR deletion mutant showed that expression of these systems is derepressed in the mutant at low salinity compared with the wild type. DNA binding assays demonstrated that purified CosR binds directly to the regulatory region of both biosynthesis systems and four transporters. In Escherichia coli green fluorescent protein (GFP) reporter assays, we demonstrated that CosR directly represses transcription of betIBA-proXWV, bccT3, and proVWX. Similar to Vibrio harveyi, we showed betIBA-proXWV was directly activated by the quorum-sensing LuxR homolog OpaR, suggesting a conserved mechanism of regulation among Vibrio species. Phylogenetic analysis demonstrated that CosR is ancestral to the Vibrionaceae family, and bioinformatics analysis showed widespread distribution among Gammaproteobacteria in general. Incidentally, in Aliivibrio fischeri, Aliivibrio finisterrensis, Aliivibrio sifiae, and Aliivibrio wodanis, an unrelated MarR-type regulator gene named ectR was clustered with ectABC-asp, which suggests the presence of another novel ectoine biosynthesis regulator. Overall, these data show that CosR is a global regulator of osmotic stress response that is widespread among bacteria.

IMPORTANCE Vibrio parahaemolyticus can accumulate compatible solutes via biosynthesis and transport, which allow the cell to survive in high salinity conditions. There is little need for compatible solutes under low salinity conditions, and biosynthesis and transporter systems need to be repressed. However, the mechanism(s) of this repression is not known. In this study, we showed that CosR played a major role in the regulation of multiple compatible solute systems. Phylogenetic analysis showed that CosR is present in all members of the Vibrionaceae family as well as numerous Gammaproteobacteria. Collectively, these data establish CosR as a global regulator of the osmotic stress response that is widespread in bacteria, controlling many more systems than previously demonstrated.

INTRODUCTION

Halophilic bacteria, such as Vibrio parahaemolyticus, encounter a range of osmolarities and have an absolute requirement for salt. To combat the loss of turgor pressure due to the efflux of water in high osmolarity conditions, bacteria have developed a short-term “salt in” strategy requiring the uptake of K⁺ and a long-term “salt out” strategy that involves the accumulation of compatible solutes in the cell (1–3). Compatible solutes, as the name suggests, are organic compounds that are compatible with the molecular machinery and processes of the cell and include compounds such as ectoine, glycine betaine, trehalose, glycerol, proline, glutamate, and carnitine, among others (1, 4–9). Compatible solutes are taken up from the environment or biosynthesized from various precursors in response to osmotic stress, which allows cells to continue to grow and divide even in unfavorable environments (2, 4, 10, 11).

Searches of the genome database demonstrated that ectoine biosynthesis genes are present in over 500 bacterial species (12). Most of the species that contain ectoine biosynthesis genes are halotolerant or halophiles. Previously, it was shown that ectoine biosynthesis is present in all halophilic Vibrio species, including Vibrio parahaemolyticus, and this species also possesses the genes for glycine betaine biosynthesis and multiple compatible solute transporters (13). De novo biosynthesis of ectoine requires aspartic acid as the precursor, which can be supplied by the cell (14). Aspartic acid is converted to ectoine by four enzymes, namely, EctA, EctB, EctC, and Asp_Ect, encoded by the operon ectABC-asp_ect (15). Ectoine biosynthesis begins with l-aspartate-β-semialdehyde, which is also pivotal to bacterial amino acid and cell wall synthesis (15). Asp_Ect is a specialized aspartokinase dedicated to the ectoine pathway that, among Proteobacteria, is present only in Alphaproteobacteria, Gammaproteobacteria, and Deltaproteobacteria species (16). Our recent study showed that the quorum-sensing response regulator OpaR was a negative regulator and AphA was a positive regulator of ectABC-asp_ect gene expression (17). In addition, we showed that OpaR and AphA are positive regulators of cosR, which encodes a MarR-type regulator CosR (17). We showed that, similar to the case in Vibrio cholerae, CosR is a repressor of ectABC-asp_ect, indicating that control of ectoine biosynthesis is multilayered and stringent (17, 18).

The production of glycine betaine is a two-step oxidation process from the precursor choline, which is acquired exogenously. De novo biosynthesis of glycine betaine has been identified in only a few species of halophilic bacteria (19–24). Choline is converted to glycine betaine by the products of two genes, namely, betB and betA (25, 26). In Escherichia coli, these genes are located within the operon betIBA, with the regulator BetI shown to repress its own operon (27, 28). In all Vibrio species that biosynthesize glycine betaine, the betIBA genes are in an operon with the proXWV genes, which encode an ATP-binding cassette (ABC)-type transporter named ProU2 (13, 14, 29). The regulation of glycine betaine biosynthesis has been studied in several species, but few direct mechanisms of regulation have been shown beyond BetI (27–32). Recently, in Vibrio harveyi, a close relative of V. parahaemolyticus, betIBA-proXWV was shown to be positively regulated by the quorum-sensing master regulator LuxR (31, 32).

It is energetically favorable to the cell to take up compatible solutes from the environment rather than to biosynthesize them, and Bacteria and Archaea encode multiple osmoregulated transporters (9, 33–38). ABC-type transporters are used to import exogenous compatible solutes into the cell and include ProU (encoded by proVWX) in E. coli and Pseudomonas syringae, OpuA in Lactococcus lactis and Bacillus subtilis, and OpuC in P. syringae (38–43). Vibrio parahaemolyticus encodes two ProU transporters, one on each chromosome. ProU1 is encoded by proVWX (VP1726 to VP1728) and ProU2 is encoded by the betIBA-proXWV operon (VPA1109 to VPA1114) (13). ProU1 is a homolog of the E. coli K-12 ProU, which in this species was shown to bind glycine betaine with high affinity (40, 44, 45). ProU2 is a homolog of the P. syringae proVXW (13).

The betaine-carnitine-choline transporters (BCCTs) are single-component sodium- or proton-coupled transporters, the first of which, BetT, discovered in E. coli, was shown to transport choline with high affinity and is divergently transcribed from betIBA (46, 47). Vibrio parahaemolyticus encodes four BCCTs, namely, BCCT1 to BCCT3 (VP1456, VP1723, and VP1905), and BCCT4 (VPA0356) (13). The bccT2 (VP1723) gene is the only bccT that is not induced by salinity in V. parahaemolyticus (14). All four BCCT transporters were shown to transport glycine betaine among other compatible solutes (48). A study in V. cholerae demonstrated that a bccT3 homolog is repressed by the regulator CosR in low salt conditions (18).

To date, there has been no single regulator identified that controls multiple compatible solute systems in bacteria. In this study, we examined whether CosR could have a broader role in the osmotic stress response. First, we examined the expression of genes encoding osmotic stress response systems in low salinity and used quantitative real-time PCR to quantify the expression of these genes in a ΔcosR deletion mutant. This analysis showed that CosR was a negative regulator of both ectoine and glycine betaine biosynthesis systems and two different transporter systems, namely, the ABC-type transporters ProU1 and ProU2 and the sodium-coupled transporters BCCT1 and BCCT3. These data indicate that the CosR regulon is larger than appreciated and expands the role of CosR to that of a global regulator of the osmotic stress response. We determined whether CosR was a direct regulator using DNA binding assays and an E. coli plasmid-based reporter assay. We also examined whether betIBA-proXWV was under the control of the quorum-sensing regulator OpaR, which also regulates cosR. We showed that OpaR is an activator of betIBA-proXWV in contrast to its repression of ectABC-asp_ect. Phylogenetic analysis of CosR showed it is ancestral to the Vibrionaceae family and present in all members of the group. Bioinformatics analysis indicated that CosR homologs are also prevalent among Gammaproteobacteria in general. Overall, the data show that CosR is a previously unrecognized global regulator of the osmotic stress response that is widespread among bacteria.

RESULTS

Compatible solute biosynthesis and transport genes are downregulated in low salinity.

We have previously shown that V. parahaemolyticus does not produce the compatible solutes ectoine and glycine betaine during growth in minimal medium (M9G) supplemented with 1% NaCl (M9G1%) (13, 14). Here, we quantified expression levels of both biosynthesis operons in M9G1% or M9G3%. RNA was isolated from exponentially growing wild-type V. parahaemolyticus RIMD2210633 cells, at an optical density at 595 nm (OD₅₉₅) of 0.45, after growth in M9G1% or M9G3%. Real-time quantitative PCR (qPCR) showed that ectoine biosynthesis genes ectA and asp_ect are differentially expressed in M9G1% compared with expression in M9G3%. ectA is significantly downregulated 794.6-fold and asp_ect is significantly downregulated 204.9-fold in M9G1% (Fig. 1A). The betIBA-proXWV operon is also significantly repressed in M9G1%, with fold changes of 25.8-fold, 22-fold, 33.7-fold, and 52.8-fold for betI, betB, proX, and proW, respectively (Fig. 1B).

FIG 1.

RNA was isolated from RIMD2210633 after growth in M9G 1% NaCl and M9G 3% NaCl at an OD₅₉₅ of 0.45. Expression analysis of ectA and asp_ect (A); betI, betB, proX2, and proW2 (B); and bccT1, bccT2, bccT3, bccT4, and proV1 (C) by quantitative real-time PCR (qPCR). 16S was used for normalization. Expression levels shown are levels in M9G1% relative to M9G3%. The mean and standard error of two biological replicates are shown. Statistics were calculated using a Student’s t test (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Similarly, the expressions of bccT1, bccT3, bccT4, and proV1 are significantly repressed in M9G1% by 500-fold, 71.4-fold, 11.6-fold, and 2,786-fold, respectively, compared with expression in M9G3% (Fig. 1C). The bccT2 gene remained unchanged. Previously, we reported that bccT2 is not induced by salinity (48), and our data indicated that it has a basal level of transcription in the cell based on similar cycle threshold (C_T) values in both salinities tested (data not shown). Overall, the data demonstrate osmoregulation of ectABC-asp_ect, betIBA-proXWV, bccT1, bccT3, bccT4, and proV1.

CosR represses compatible solute biosynthesis and transport genes in low salinity.

We know CosR is a repressor of ectoine biosynthesis genes, and we wondered whether it played a broader role in the regulation of other osmotic stress response genes. Therefore, we examined expression in the wild type and an in-frame cosR deletion mutant. RNA was isolated from the ΔcosR mutant strain at mid-exponential phase (OD₅₉₅, 0.45) after growth in M9G1% and compared with the wild type grown under identical conditions. Using qPCR analysis, we determined the expression levels of ectA and asp_ect and showed they are significantly upregulated, 818.5-fold and 308.2-fold, respectively, in a ΔcosR mutant compared with the wild type in M9G1% (Fig. 2A), indicating derepression in the absence of CosR. Next, we examined expression levels of betIBA-proXWV and showed these genes are significantly derepressed in the ΔcosR mutant (Fig. 2B). Similarly, relative expression levels of bccT1, bccT3, and proV1 were significantly higher in the ΔcosR mutant than the wild type, while levels of bccT2 and bccT4 were unchanged (Fig. 2C). In summary, these data demonstrated that CosR is a repressor of ectABC-asp_ect, betIBA-proXWV, bccT1, bccT3, and proVWX1 under low salinity conditions. Thus, CosR is a unique example of a regulator that controls multiple compatible solute systems.

FIG 2.

RNA was isolated from RIMD2210633 and ΔcosR strains after growth in M9G 1% NaCl at an OD₅₉₅ of 0.45. Expression analysis of ectA and asp_ect (A); betI, betB, proX2, and proW2 (B); and bccT1, bccT2, bccT3, bccT4, and proV1 (C) by qPCR. 16S was used for normalization. Expression levels shown are ΔcosR levels relative to wild type. The mean and standard error of two biological replicates are shown. Statistics were calculated using a Student’s t test (*, P < 0.05; **, P < 0.01).

CosR binds directly to the promoter of the betIBA-proXWV operon and represses transcription.

To determine whether CosR regulation of betIBA-proXWV is direct, we performed DNA binding assays with purified CosR protein and DNA probes of the regulatory region of this operon. The regulatory region was split into five overlapping probes, namely, PbetI probes A to E (Fig. 3A). CosR bound to probe A, which is directly upstream of the start codon for betI, and it also bound to probes B and D (Fig. 3B). CosR did not bind to probes C and E, which demonstrated specificity of CosR binding (Fig. 3B).

FIG 3.

(A) The regulatory region of betIBA-proXWV was divided into five probes for EMSAs, namely, PbetI A to E, that were 125 bp, 112 bp, 142 bp, 202 bp, and 158 bp long, respectively. The regulatory region used for the GFP reporter assay is indicated with a bracket. (B) An EMSA was performed with purified CosR-His (0 to 0.62 μM) and 30 ng of each PbetI probe, with DNA:protein molar ratios of 1:0, 1:1, 1:5, and 1:10. (C) A P_betI-gfp reporter assay was performed in E. coli strain MKH13 containing an expression plasmid with full-length cosR (pcosR). Specific fluorescence of the CosR-expressing strain was compared with a strain harboring an empty expression vector. The mean and standard deviation of two biological replicates are shown. Statistics were calculated using a Student’s t test (***, P < 0.001).

To demonstrate that direct binding by CosR results in transcriptional repression of the betIBA-proXWV operon, we performed a green fluorescent protein (GFP) reporter assay in E. coli strain MKH13. Full-length cosR was expressed from a plasmid (pBBRcosR) in the presence of a gfp-expressing reporter plasmid under the control of the glycine betaine biosynthesis system regulatory region (P_betI-gfp). Relative fluorescence and OD₅₉₅ were measured after overnight growth in M9G1%. Specific fluorescence was calculated by normalizing to OD and compared with specific fluorescence in a strain with an empty expression vector (pBBR1MCS) that also contained the P_betI-gfp reporter plasmid. The activity of the P_betI-gfp reporter was significantly repressed 4.84-fold compared with the empty vector strain (Fig. 3C). This indicates that CosR directly represses transcription of the betIBA-proXWV genes.

CosR binds directly to the promoter of bccT1 and bccT3 and is a direct repressor of bccT3.

Next, we wanted to investigate whether CosR repression of bccT1 and bccT3 was direct. We designed probes upstream of the translational start for bccT1 and bccT3. The 291-bp regulatory region of PbccT1, which includes 15 bp of bccT1 and 276 bp of the intergenic region, was split into three overlapping probes, PbccT1 probes A, B, and C (Fig. 4A). DNA binding assays were performed with increasing concentrations of CosR. CosR bound directly to the PbccT1 probe B but did not bind to the other probes tested, which indicated direct and specific binding by CosR (Fig. 4B). Next, we performed reporter assays in E. coli using a GFP expression plasmid under the control of the regulatory region of bccT1 (P_bccT1-gfp) and a CosR expression plasmid (pBBRcosR). Specific fluorescence in the presence of CosR was compared with a strain with an empty expression vector (pBBR1MCS). The activity of the P_bccT1-gfp reporter was not significantly different than the strain harboring the empty expression vector, which indicates that CosR does not directly repress bccT1 (Fig. 4C). We speculate that CosR may still directly repress bccT1, but in our reporter assay, the low level of activation of the bccT1 regulatory region in E. coli may have affected the significance of the results. In the E. coli heterologous background, additional proteins, which are present in the native species, may be necessary for full repression of bccT1 by CosR.

FIG 4.

(A) The regulatory region of bccT1 was divided into three similarly sized probes for EMSAs, namely, PbccT1 A to C, which were 120 bp, 110 bp, and 101 bp long, respectively. The regulatory region used for the GFP reporter assay is indicated with a bracket. (B) An EMSA was performed with purified CosR-His (0 to 0.69 μM) and 30 ng of PbccT1 probe with DNA:protein molar ratios of 1:0, 1:1, 1:5, and 1:10. (C) A P_bccT1-gfp reporter assay was performed in E. coli strain MKH13 containing an expression plasmid with full-length cosR (pcosR). Specific fluorescence of the CosR-expressing strain was compared with a strain harboring an empty expression vector (pBBR1MCS). The mean and standard deviation of two biological replicates are shown. Statistics were calculated using a Student’s t test.

Two overlapping probes designated PbccT3 probe A and B were designed encompassing 196 bp of the regulatory region of bccT3 (Fig. 5A). Because bccT3 is divergently transcribed from cosR, we used approximately half of the regulatory region for the PbccT3 electrophoretic mobility shift assay (EMSA). An EMSA showed that CosR bound directly to the PbccT3 probe A, which is proximal to the start of the gene, but not probe B (Fig. 5B). We then performed reporter assays in E. coli using a GFP expression plasmid under the control of the regulatory region of bccT3, utilizing the entire 397-bp intergenic region between bccT3 and cosR. Transcriptional activity of the P_bccT3-gfp reporter is repressed in a CosR-expressing strain (Fig. 5C), although not to the same extent that we saw in expression analyses in V. parahaemolyticus. This is not surprising, given that it appears the regulatory region of bccT3 is not very active in an E. coli background, which made detection of repression more difficult. Additionally, other proteins are likely necessary for full repression of the regulatory region of bccT3 that are not present in an E. coli background. The E. coli GFP assay did show a direct interaction between CosR and the bccT3 regulatory region that resulted in repression of transcription. This result, in combination with expression analyses in a cosR mutant (Fig. 2C) and binding assays which demonstrated direct binding (Fig. 5B), indicate that CosR directly represses bccT3. In addition, we showed that CosR does not bind to the regulatory region of bccT2 and bccT4 (Fig. 5D and E), which is in agreement with the cosR mutant expression data (Fig. 2C). These data suggest that bccT2 and bccT4 are under the control of a yet-to-be-described regulator.

FIG 5.

(A) A 196-bp portion of the regulatory region of bccT3 was split into two probes for EMSAs, namely, PbccT3 A and B, which are 108 bp and 107 bp long, respectively. The regulatory region used for the GFP reporter assay is indicated with a bracket. (B) An EMSA was performed with purified CosR-His (0 to 0.65 μM) and 30 ng of PbccT3 probe with DNA:protein molar ratios of 1:0, 1:1, 1:5, and 1:10. (C) The P_bccT3-gfp reporter assay was performed in E. coli strain MKH13 containing an expression plasmid with full-length cosR (pcosR). Specific fluorescence of the CosR-expressing strain was compared with a strain harboring an empty expression vector (pBBR1MCS). The mean and standard deviation of two biological replicates are shown. Statistics were calculated using a Student’s t test (**, P < 0.01). (D) Diagrams indicating the regulatory regions of bccT2 and bccT4 that were used as probes in a CosR EMSA. (E) An EMSA was performed with CosR-His (0 to 0.18 μM) and probes of the regulatory regions of bccT2 and bccT4. Each lane contains 30 ng of DNA and DNA:protein molar ratios of 1:0, 1:1, 1:5, and 1:10.

CosR is a direct repressor of proVWX1.

The regulatory region upstream of the proV1 gene was divided into four probes (Fig. 6A). A DNA binding assay was performed with increasing concentrations of CosR and 30 ng of each probe. A shift in the DNA bands of probe D, which is proximal to the start codon of proV1, indicated that CosR binds directly to this region (Fig. 6B). CosR did not bind to the other probes tested, which indicated that CosR binding is specific.

FIG 6.

(A) The 447-bp regulatory region of the proV1 gene was divided into four probes for EMSAs, namely, PproV1 A to D, which were 160 bp, 134 bp, 108 bp, and 109 bp long, respectively. The regulatory region used for the GFP reporter assay is indicated with a bracket. (B) An EMSA was performed with purified CosR-His (0 to 0.64 μM) and 30 ng of each PproV1 probe with DNA:protein molar ratios of 1:0, 1:1, 1:5, and 1:10. (C) A reporter assay was conducted in E. coli MKH13 harboring the P_proV1-gfp reporter plasmid and the expression plasmid pcosR. Specific fluorescence of the CosR-expressing strain was compared with an empty vector strain. The mean and standard deviation of two biological replicates are shown. Statistics were calculated using a Student’s t test (*, P < 0.05).

We also performed a reporter assay in E. coli utilizing the cosR expression plasmid (pBBRcosR) and a GFP reporter plasmid (P_proV1-gfp). In a CosR-expressing strain, expression of the P_proV1-gfp reporter was repressed compared with an empty expression vector strain (Fig. 6C). This repression was to a lesser extent than is seen in V. parahaemolyticus, but recapitulation of the same magnitude of repression in the heterologous background is not to be expected given the potential absence of additional factors present in the native background. Overall, the results of the E. coli reporter assay, taken together with expression analyses in the native background (Fig. 2C) and the DNA binding assay (Fig. 6B), indicate that CosR is a direct repressor of the proVWX1 operon.

CosR is not autoregulated.

In V. cholerae, expression levels of cosR were upregulated in 0.5 M NaCl compared with levels in 0.2 M NaCl (18). It was suggested that one reason for the upregulation of cosR in higher salinity could be that it is involved in an autoregulatory feedback loop (18). In V. parahaemolyticus, we found that levels of cosR were not significantly upregulated in 3% NaCl compared with 1% NaCl (data not shown). We have already shown that CosR binds to the intergenic region between bccT3 and cosR, but the binding site location is proximal to the start codon of bccT3, more than 300 bp upstream of the cosR gene (Fig. 5A and B). Therefore, to investigate CosR autoregulation, we designed two probes of 105 bp and 142 bp, which comprise a 220-bp portion of the regulatory region upstream of cosR (VP1906) (Fig. 7A), and used them in a DNA binding assay with various concentrations of purified CosR (Fig. 7B). There were no shifts observed in the binding assay, which indicated that CosR does not bind (Fig. 7B). We then performed a GFP reporter assay in E. coli, utilizing the entire 397-bp intergenic region between bccT3 and cosR, to determine if CosR directly represses transcription of its own gene. The transcriptional activity of P_cosR-gfp in the presence of CosR was not significantly different from the empty-vector strain (P = 0.09) (Fig. 7C). Because we cannot assess the expression of cosR in a ΔcosR mutant, we examined it in a GFP reporter assay in the wild type and a ΔcosR mutant after growth in M9G1%. We found that the activity of a P_cosR-gfp reporter was not different between the wild type and the cosR mutant (Fig. 7D). Taken together, a lack of CosR binding in the EMSA and both in vivo and E. coli reporter assays lead us to conclude that under these conditions CosR does not autoregulate and that the CosR binding site proximal to the bccT3 gene does not affect transcription of the cosR gene.

FIG 7.

(A) A 220-bp section of the regulatory region of cosR was split into two similarly sized probes for EMSAs, namely, PcosR A and B, which were 105 bp and 142 bp long, respectively. The regulatory region used for the GFP reporter assay is indicated with a bracket. (B) An EMSA was performed with increasing concentrations of purified CosR-His (0 to 0.66 μM) and 30 ng of each probe with DNA:protein molar ratios of 1:0, 1:1, 1:5, and 1:10. (C) A P_cosR-gfp reporter assay was performed in E. coli strain MKH13 the pcosR expression plasmid. Specific fluorescence of the CosR-expressing strain was compared with a strain harboring an empty expression vector. The mean and standard deviation of two biological replicates are shown. (D) A P_cosR-gfp reporter assay was performed in V. parahaemolyticus wild-type (WT) and ΔcosR mutant strains. The mean and standard deviation of three biological replicates are shown.

BetI represses its own operon, betIBA.

Previously, it was shown that BetI represses its own operon in several bacterial species and this repression is relieved in the presence of choline (27, 30, 31). To demonstrate that BetI regulates its own operon in V. parahaemolyticus, we performed a reporter assay utilizing the P_betI-gfp reporter in the wild type and a ΔbetI mutant strain. Strains were grown overnight in M9G3%, with and without choline, and specific fluorescence was calculated. Expression of the reporter was derepressed in the ΔbetI mutant when no choline was present, indicating that BetI is a negative regulator of its own operon (Fig. 8A). In the presence of choline, there was no longer a difference in reporter activity between the wild-type strain and the ΔbetI mutant strain, indicating that repression by BetI was relieved (Fig. 8B). To confirm that the regulation of betIBA-proXWV by BetI is direct, we performed a GFP reporter assay in the E. coli MKH13 strain. The P_betI-gfp reporter was transformed into E. coli MKH13 (which lacks the betIBA operon) along with an expression vector harboring full-length betI under the control of an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter. In the BetI-expressing strain, P_betI-gfp expression was significantly repressed, which indicated that BetI is a direct repressor of its own operon in V. parahaemolyticus (Fig. 8C).

FIG 8.

(A) Expression of a P_betI-gfp transcriptional fusion reporter in the wild type and a ΔbetI mutant. Relative fluorescence intensity (in RFUs) and OD₅₉₅ were measured after growth in M9G3% (A) or M9G3% (B) with the addition of choline. Specific fluorescence was calculated by dividing RFUs by OD. The mean and standard deviation of two biological replicates are shown. Statistics were calculated using a Student’s t test (*, P < 0.05). (C) A reporter assay was conducted in E. coli MKH13 using the P_betI-gfp reporter plasmid and an expression plasmid with full-length betI (pbetI). The specific fluorescence was calculated and compared with a strain with an empty expression vector (pBBR1MCS). The mean and standard deviation of two biological replicates are shown. Statistics were calculated using a Student’s t test (***, P < 0.001).

The quorum-sensing LuxR homolog OpaR is a positive regulator of betIBA-proXWV in V. parahaemolyticus.

We examined expression of the P_betI-gfp reporter in the wild type and the ΔopaR mutant in V. parahaemolyticus. Expression of the reporter was significantly downregulated in ΔopaR, indicating that OpaR is a positive regulator of the glycine betaine biosynthesis operon (Fig. 9A). We also examined whether regulation of PbetI by OpaR was direct utilizing an EMSA with purified OpaR protein. OpaR bound to PbetI probes A, B, C, and E and very weakly to probe D, which indicated that regulation of betIBA-proXWV by OpaR is direct (Fig. 9B). These results are in agreement with a previous study, which also showed direct positive regulation of betIBA-proVWX by LuxR in V. harveyi (31). Thus, it appears that the quorum-sensing master regulator may be a conserved regulatory mechanism of glycine betaine biosynthesis among Vibrio species.

FIG 9.

(A) Expression of a P_betI-gfp transcriptional fusion reporter in wild-type and ΔopaR mutant strains. Relative fluorescence intensity (in RFUs) and OD₅₉₅ were measured after growth in M9G3%. Specific fluorescence was calculated by dividing RFUs by OD. The mean and standard deviation of two biological replicates are shown. Statistics were calculated using a one-way analysis of variance (ANOVA) with a Tukey-Kramer post hoc test (**, P < 0.01). (B) An EMSA was performed with 30 ng of each PbetI probe A to E used previously in the CosR EMSA and purified OpaR protein (0 to 0.41 μM) in various DNA:protein molar ratios (1:0, 1:1, and 1:5 for probe A; 1:0, 1:1, and 1:10 for all other probes).

Motif identification and phylogenetic distribution of CosR.

CosR bound to eight of the probes tested in our DNA binding analyses, including two probes of the ectABC-asp_ect regulatory region, as shown previously (17). We utilized these sequences in multiple EM for motif elicitation (MEME) analysis (49), and identified a 24-bp pseudopalindromic motif present in each of the eight sequences (Fig. 10A). The motif is an imperfect inverted repeat separated by 2 bp (TTTGA-NN-TCTAA). The alignment of the motifs found within each sequence is shown in Fig. 10B.

FIG 10.

(A) A CosR DNA binding motif was created using MEME analysis with sequences of the EMSA probes that were bound by CosR. (B) An alignment of the motif sites found in each sequence used for MEME analysis with corresponding P values.

CosR, a MarR-type regulator, is a 158-amino-acid protein that is divergently transcribed from bccT3 on chromosome 1 in V. cholerae and V. parahaemolyticus, two distantly related species. Bioinformatics analysis showed that a CosR homolog is present in over 50 Vibrio species, and in all cases, the cosR homolog was divergently transcribed from a bccT transporter (Fig. 11). Within these Vibrio species, similarity ranged from 98% to 73% amino acid identity and showed that CosR was present in phylogenetically divergent Vibrio species (Fig. 12). We found that in Vibrio splendidus, Vibrio crassostreae, Vibrio cyclitrophicus, Vibrio celticus, Vibrio lentus, and Aliivibrio wodanis, the CosR homolog is present directly downstream of the betIBA-proXWV operon, and in Vibrio tasmaniensis strains and Vibrio sp. MED222, the ectoine biosynthesis operon clustered in the same genome location (Fig. 11). Collectively, these data indicated that CosR function is conserved among this divergent group of species and that CosR is an important regulator of the osmotic stress response. In all strains of Aliivibrio fischeri, the cosR homolog (which shares 73% amino acid identity with CosR from V. parahaemolyticus) clusters with two uncharacterized transporters. A recent phylogenomics study of the distribution of ectoine biosynthesis genes and a homolog of CosR showed the presence of this regulator in species of the Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria (50). This, again, suggests that the role of CosR in the osmotic stress response is conserved and phylogenetically widespread.

FIG 11.

Schematic of the genomic context of CosR homologs from select Vibrionaceae species. Open reading frames are designated by arrows.

FIG 12.

Phylogenetic and distribution analysis of CosR. The phylogeny of CosR was inferred using the neighbor-joining method. The optimal tree with the sum of branch length of 5.19649696 is shown. The percentages of replicate trees in which the associated taxa clustered together in the bootstrap test (1,000 replicates) are 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. Brackets represent groups based on genus.

Incidentally, a second MarR-type regulator, a 141-amino-acid protein, which we name ectR, clusters with the ectoine biosynthesis genes in Aliivibrio fischeri (Fig. 11). EctR shares only 31% identity with less than 60% query coverage to CosR from V. parahaemolyticus and a similar level of low amino acid identity to EctR1 from Methylomicrobium alcaliphilum. EctR was clustered with the ectABC-asp_ect genes in all strains of Aliivibrio finisterrensis and Aliivibrio sifiae and most Aliivibrio wodanis strains. Thus, in Aliivibrio species, it appears that the ectoine biosynthesis gene cluster has a new uncharacterized MarR-type regulator.

DISCUSSION

Here, we have shown that the compatible solute biosynthesis and transport genes are downregulated in V. parahaemolyticus in low salinity and these genes are derepressed in a cosR mutant. Our genetic analyses, binding analyses, and reporter assays demonstrated that CosR is a regulator of betIBA-proXWV, bccT1, bccT3, and proVWX (Fig. 13). Additionally, we showed that under the conditions tested, CosR is not autoregulated. To date, it has now been demonstrated that CosR is a regulator of six different compatible solute systems, including the two biosynthesis systems ectoine and glycine betaine and four transporters (Fig. 13). Our phylogenetic and bioinformatics analyses indicated that CosR is universal within Vibrionaceae and widespread in Gammaproteobacteria, suggesting a conserved previously unrecognized global regulator of the osmotic stress response in bacteria.

FIG 13.

A model of CosR regulation of the osmotic stress response and its known regulators. Solid arrows indicate direct positive regulation, dashed arrows indicate indirect positive regulation, solid hammers represent direct repression, and dashed hammers indicate indirect repression. Transporters colored purple are osmotic responsive. The quorum-sensing regulators OpaR and AphA were shown in previous studies to directly and indirectly positively regulate CosR, respectively, and in addition, directly regulate ectoine and glycine betaine biosynthesis operons.

The physiological importance of CosR repression of compatible solute biosynthesis in low salinity is to protect levels of key intracellular metabolites. Ectoine biosynthesis requires the precursor aspartate, and this affects the level of glutamate, acetyl-coenzyme A (CoA), and oxaloacetate (51, 52). Thus, tight regulation of ectoine biosynthesis is essential for cellular fitness. CosR characterized from Vibrio species shows ∼50% amino acid identity to EctR1, which was first described in the halotolerant methanotroph Methylmicrobium alcaliphilum that repressed ectoine biosynthesis (53). In this species, ectR1 is divergently transcribed from the same promoter as ectABC-ask. Mustakhimov and colleagues showed that EctR1 directly repressed expression of the ectABC-ask operon in response to low salinity (53). EctR repression of the ectoine biosynthesis genes was also shown in both Methylophaga alcalica and Methylophaga thalassica, two moderately halophilic methylotrophs (54, 55). Czech and colleagues showed that CosR/EctR1 was phylogenetically widespread and clustered with ect genes in some species (50). In V. cholerae, CosR was also identified as a repressor of ectoine biosynthesis genes, although it does not cluster with ectABC-asp_ect (18). The cosR gene in V. cholerae is divergently transcribed from the opuD gene (a bccT3 homolog), which was also repressed by CosR (18). Similarly, in V. parahaemolyticus, the cosR (VP1906) homolog is divergently transcribed from bccT3 (VP1905) and is a direct negative regulator of both bccT3 and ectABC-asp_ect (17). Our phylogenetic analysis found that the CosR homolog was present in all members of Vibrionaceae and among many Vibrio species was clustered with a bccT3 homolog (Fig. 12). The phylogeny of CosR mirrored the branching pattern of the relationships of members of the group for other housekeeping genes. These data indicate that CosR is ancestral to the group, and the conservation of genomic context suggests functional conservation (Fig. 11 and 12). We used CosR from Aeromonas species to root the tree and, similar to Vibrionaceae species, CosR is present in all members of this group (Fig. 12). Indeed, bioinformatics analysis indicated that a CosR homolog is present in many other Gammaproteobacteria, suggesting that it is an underappreciated player in the osmotic stress response in bacteria (data not shown). In several Vibrio species, the CosR homolog was clustered with the betIBA-proXWV operon, which is further suggestive of its role in the regulation of compatible solute biosynthesis among Vibrio species.

The MarR family of transcriptional regulators, first characterized in E. coli, are important regulators of a number of cellular responses, typically responding to a change in the external environment (56–58). The literature suggests that MarR-type regulators form dimers and bind to a 20- to 45-bp pseudopalindromic site in the intergenic region of genes they control (56, 59–61). We utilized the regulatory regions of each of the osmotic stress response genes that CosR regulates and identified a pseudopalindromic CosR DNA binding motif (Fig. 10). This motif is similar to the binding sequence of the CosR homolog EctR1 identified previously in M. alcaliphilum, which was also pseudopalindromic with 2-bp separating inverted repeats (53). The activity of MarR-type regulators can be modulated by the presence of a chemical signal, either a ligand, metal ion, or reactive oxygen species. Binding of these signals causes the protein to undergo a conformational change, thereby affecting the DNA binding capability (56, 62, 63). We modeled a CosR homodimer using the SWISS-MODEL and did not identify a ligand-binding pocket (data not shown). In V. cholerae, CosR activity was not affected by the presence of exogenous compatible solutes, including ectoine, glycine betaine and proline, and opuD (bccT homolog) transcripts were unchanged in a cosR mutant (50). Hence, the environmental or cellular signals that modulate the activity of CosR remain unidentified, as was noted by Czech and colleagues (50). Interestingly, our modeling of the EctR regulator clustered with ectoine genes identified in Aliivibrio species indicated it also does not have a ligand-binding pocket (data not shown). Autoregulation was shown for several MarR family regulators, including ectR1 in M. alcaliphilum (53, 56). In V. parahaemolyticus, we showed CosR does not bind to its own regulatory region, and our reporter assays suggested that CosR does not autoregulate. It is interesting to note that EctR1 does not participate in an autoregulatory feedback loop in M. thalassica (55, 64).

Similar to ectoine biosynthesis gene expression, few direct regulators of glycine betaine biosynthesis genes have been identified. In E. coli, expression of betIBA was repressed by BetI and repression was relieved in the presence of choline (27). BetI was shown to directly regulate transcription at this locus via DNA binding assays (28). ArcA was shown to repress the betIBA operon under anaerobic conditions in E. coli, although direct binding was not shown (27). In V. harveyi, it was shown that betIBA-proXWV were repressed 2- to 3-fold when betI was overexpressed from a plasmid. Purified BetI bound directly to the regulatory region of the betIBA-proXWV operon in DNA binding assays (31, 32). In these studies, it was also shown that the quorum-sensing response regulator LuxR, along with the global regulator integration host factor (IHF), activated expression of betIBA-proXWV (31, 32). Here, we have shown that BetI represses its own operon in V. parahaemolyticus, as expected, and we identified a novel regulator of glycine betaine biosynthesis genes, CosR, which directly represses under low salinity conditions (Fig. 13). We also confirm that, similar to V. harveyi, the quorum-sensing master regulator OpaR directly induced betIBA-proXWV expression in V. parahaemolyticus, indicating this mechanism is likely conserved in Vibrio species (Fig. 13).

Biosynthesis of compatible solutes is an energetically costly process for bacteria (34). Vibrio parahaemolyticus does not accumulate compatible solutes in low salinity (13, 14, 48), and therefore, the transcription of biosynthesis and transport genes is unnecessary. CosR represses these genes involved in the osmotic stress response in V. parahaemolyticus. The high conservation of the CosR protein across Vibrionaceae and Gammaproteobacteria and its genomic context indicate that regulation by CosR of compatible solute systems is widespread in bacteria.

MATERIALS AND METHODS

Bacterial strains, medium, and culture conditions.

Listed in Table 1 are all strains and plasmids used in this study. A previously described streptomycin-resistant clinical isolate of V. parahaemolyticus, RIMD2210633, was used as the wild-type strain (65, 66). Vibrio parahaemolyticus strains were grown either in lysogeny broth (LB) (Fisher Scientific, Fair Lawn, NJ) supplemented with 3% NaCl (wt/vol) (LBS) or in M9 minimal medium (47.8 mM Na₂HPO₄, 22 mM KH₂PO₄, 18.7 mM NH₄Cl, and 8.6 mM NaCl) (Sigma-Aldrich, USA) supplemented with 2 mM MgSO₄, 0.1 mM CaCl₂, and 20 mM glucose as the sole carbon source (M9G) and 1% or 3% NaCl (wt/vol) (M9G1% and M9G3%, respectively). E. coli strains were grown in LB supplemented with 1% NaCl (wt/vol) or M9G1% where indicated. E. coli β2155, a diaminopimelic acid (DAP) auxotroph, was supplemented with 0.3 mM DAP and grown in LB 1% NaCl. All strains were grown at 37°C with aeration. Antibiotics were used at the following concentrations (wt/vol) as necessary: ampicillin (Amp), 50 μg/ml; chloramphenicol (Cm), 12.5 μg/ml; tetracycline (Tet), 1 μg/ml; and streptomycin (Str), 200 μg/ml. Choline was added to media at a final concentration of 1 mM, when indicated.

TABLE 1.

Strains and plasmids

Bacterial strain or plasmid	Genotype or description	Reference or source
Strains
Vibrio parahaemolyticus
RIMD2210633	O3:K6 clinical isolate, Strr	65, 66
ΔcosR	RIMD2210633 ΔcosR (VP1906), Strr	17
ΔbetI	RIMD2210633 ΔbetI (VPA1114), Strr	This study
SSK2516 (ΔopaR)	RIMD2210633 ΔopaR (VP2516), Strr	70
Escherichia coli
DH5α λpir	Δlac pir	ThermoFisher Scientific
β2155 λpir	ΔdapA::erm pir for bacterial conjugation	82
BL21(DE3)	Expression strain	ThermoFisher Scientific
MKH13	MC4100 (ΔbetTIBA) Δ(putPA)101 Δ(proP)2 Δ(proU); Spr	83
Plasmids
pDS132	Suicide plasmid; Cmr; SacB	84
pBBR1MCS	Expression vector; lacZ promoter; Cmr	85
pBBRcosR	pBBR1MCS harboring full-length cosR (VP1906)	17
pRU1064	Promoterless gfpUV, Ampr, Tetr, IncP origin	86
pRUPectA	pRU1064 with PectA-gfp, Ampr, Tetr	17
pRUPbetI	pRU1064 with PbetI-gfp, Ampr, Tetr	This study
pRUPbccT1	pRU1064 with PbccT1-gfp, Ampr, Tetr	This study
pRUPbccT3	pRU1064 with PbccT3-gfp, Ampr, Tetr	This study
pRUPproV1	pRU1064 with PproV-gfp, Ampr, Tetr	This study
pRUPcosR	pRU1064 with PcosR-gfp, Ampr, Tetr	17
pET28a+	Expression vector, 6×His; Kanr	Novagen
pETcosR	pET28a+ harboring cosR, Kanr	17

Construction of the betI deletion mutant.

An in-frame betI (VPA1114) deletion mutant was constructed as described previously (17). Briefly, the Gibson assembly protocol, using NEBuilder high-fidelity (HiFi) DNA assembly master mix (New England BioLabs, Ipswich, MA), followed by allelic exchange, was used to generate an in-frame 63-bp truncated, nonfunctional betI gene (67, 68). Two fragments, AB and CD, were amplified from the RIMD2210633 genome using primers listed in Table 2. These fragments were ligated with pDS132, which had been digested with SphI via Gibson assembly to produce suicide vector pDS132 with a truncated betI allele (pDSΔbetI). pDSΔbetI was transformed into E. coli strain β2155 λpir, followed by conjugation with V. parahaemolyticus. The suicide vector pDS132 must be incorporated into the V. parahaemolyticus genome via homologous recombination, as V. parahaemolyticus lacks the pir gene required for replication of the vector. Growth without chloramphenicol induces a second recombination event which leaves behind either the truncated mutant allele or the wild-type allele. Colonies were plated on sucrose for selection, as pDS132 harbors a sacB gene, which makes sucrose toxic to cells still carrying the plasmid and colonies appear soapy. Healthy colonies were screened via PCR and sequenced to confirm an in-frame deletion of the betI gene.

TABLE 2.

Primers used in this study

Primer name by category	Sequence (5′-3′)a	Length (bp)
Mutant
VPbetIA	gcttcttctagaggtaccgcatgcGCCAGTTTTATGTGCTCACC	580
VPbetIB	atattttatgagaCATCCCCACCTTTGGCATTTTG
VPbetIC	gatgcctgaaCTCGACAAGCAGCTAACG	688
VPbetID	ggagagctcgatatcgcatgcTCTGCCCTACCCGGTAATC
VPbetIFLFwd	AGCATAGCACAATAAGAGTCG	1,895
VPbetIFLRev	CCTGATTCGCCAGTGAACGA
EMSA
VPbetIFwdA	CGGTTTTCTGATTTCAGGC	125
VPbetIRevA	CTTTTAATGATAAATCGTTTGAGTTCG
VPbetIFwdB	ATGCCAAAAATTTAGTTCGAAC	112
VPbetIRevB	GGTCTTTGAATGGATGGTAGGG
VPbetIFwdC	CCCTACCATCCATTCAAAGACC	142
VPbetIRevC	CTAAGGCTTCTACATTGCTTTC
VPbetIFwdD	GAAAGCAATGTAGAAGCCTTAG	202
VPbetIRevD	GAACTTGGATATGCGTCCATT
VPbetIFwdE	AATGGACGCATATCCAAGTTC	158
VPbetIRevE	AGCATAGCACAATAAGAGTCG
VPbccT1FwdA	ACCGCAAACTTCCCGATC	120
VPbccT1RevA	CGGTATTCAGTACAAAAGAA
VPbccT1FwdB	TTCTTTTGTACTGAATACCG	110
VPbccT1RevB	TGTCTTCAACTCACAAGAAT
VPbccT1FwdC	ATTCTTGTGAGTTGAAGACA	101
VPbccT1RevC	AGCGAATTTTATCACCAATCACA
VPbccT3FwdA	CGCTTTTTGTAATGCAAATTACC	107
VPbccT3RevA	CCCGTGAAAGCGGAAGATC
VPbccT3FwdB	GATCTTCCGCTTTCACGGG	108
VPbccT3RevB	TCTATACCCTTTGTCATCGTTCCTC
VPcosRFwdA	CAAATCTCCACACCATTAATTAG	105
VPcosRRevA	CGTCTTTGGTGATTTCTTTTTATTCG
VPcosRFwdB	GCGAATAAAAAGAAATCACCAAAGACG	142
VPcosRRevB	CCAATTTTTTCATCCAGTCTGTAGGG
VPproU1FwdA	TCTTTATTCCATGCGTTG	160
VPproU1RevA	AGAGGCAGAAAGAACAGTGAA
VPproU1FwdB	TTCACTGTTCTTTCTGCCTCT	134
VPproU1RevB	GGTTATGAATGTGTTCGTTTGT
VPproU1FwdC	ACAAACGAACACATTCATAACC	108
VPproU1RevC	TGGCTTGGCTTATTGGTGTTC
VPproU1FwdD	GAACACCAATAAGCCAAGCCA	109
VPproU1RevD	GGGATCCATGTTAATTGTCCTTTG
VPbccT2Fwd	ACCGAGACATGCCAATTTCTG	233
VPbccT2Rev	CGGTGCTCACGAATAATCTCC
VPbccT4Fwd	AGAACAGGTTGGCTCAATGT	244
VPbccT4Rev	TTCCCCTCACATCAAGTCG
Expression
PbetIFwd	TCTAAGCTTGCATAGCACAATAAGAGTCGC	594
PbetIRev	TATACTAGTTTTGCGTCCTTGTTATTTTTAATTG
PbccT1Fwd	tagatagagagagagagagaAAACCGCAAACTTCCCGATC	278
PbccT1Rev	actcattttttcttcctccaCAATCACAAATTTATGCAAAAATGAC
PbccT3Fwd	tagatagagagagagagagaAATTTTTTCATCCAGTCTGTAGG	397
PbccT3Rev	actcattttttcttcctccaCGTTCCTCTCTATTTTTGTATTATTTTTTC
PproU1Fwd	tagatagagagagagagagaTCTTTATTCCATGCGTTG	438
PproU1Rev	actcattttttcttcctccaGTTAATTGTCCTTTGTTATGTG
PcosRFwd	tagatagagagagagagagaCGTTCCTCTCTATTTTTGTATTATTTTTTC	397
PcosRRev	cggccgctctagaactagtgTTATTCTGGTTTGGTGATG
qPCR primers
VPbccT1Fwd	GTTCGGTCTTGCGACTTCTC
VPbccT1Rev	CCCATCGCAGTATCAAAGGT
VPbccT2Fwd	AACAAAGGGTTGCCACTGAC	167
VPbccT2Rev	TTCAAACCTGTTGCTGCTTG
VPbccT3Fwd	TGGACGGTATTCTACTGGGC	202
VPbccT3Rev	CGCCTAACTCGCCTACTTTG
VPectAFwd	TCGAAAGGGAAGCGCTGAG	125
VPectARev	AGTGCTGACTTGGCCATGAT
VPasp_ectFwd	CGATGATTCCATTCGCGACG	126
VPasp_ectRev	GTCATCTCACTGTAGCCCCG
VPproV1Fwd	GCATCGTTTCTCTCGACTCC	163
VPproV1Rev	TGCTCATCGACTACTGGCAC
VPAbccT4Fwd	CAAGGCGTAGGCCGCATGGT	234
VPAbccT4Rev	ACCGCCCACGATGCTGAACC
VPAbetIFwd	ACTTCGGTGGTAAGCATGGG	138
VPAbetIRev	TGCCGTCAATAATGGCGTTG
VPAbetBFwd	TGGAAATCAGCACCAGCACT	160
VPAbetBRev	TCTGCCCTACCCGGTAATCA
VPAproXFwd	TTCCTTGGTAACTGGATGCC	216
VPAproXRev	ATCGTTACCTGGTTCGATGC
VPAproWFwd	ATCACAGCGGCACTGGCTTGG	190
VPAproWRev	GGCGATGCGCTGCCATGATC
16SFwd	ACCGCCTGGGGAGTACGGTC	234
16SRev	TTGCGCTCGTTGCGGGACTT

^aLowercase letters denote complementary regions for Gibson assembly.

RNA isolation and qPCR.

Vibrio parahaemolyticus RIMD2210633 and ΔcosR were grown with aeration at 37°C overnight in LBS. Cells were pelleted, washed twice with 1× phosphate-buffered saline (PBS), diluted 1:50 into M9G3% or M9G1%, and grown with aeration to mid-exponential phase (OD₅₉₅, 0.45). RNA was extracted from 1 ml of culture using TRIzol, following the manufacturer’s protocol (Invitrogen, Carlsbad, CA). The samples were treated with Turbo DNase (Invitrogen), followed by heat inactivation of the enzyme as per the manufacturer’s protocol. The final RNA concentration was quantified using a Nanodrop spectrophotometer (Thermo Scientific, Waltham, MA). A total of 500 ng of RNA was used for cDNA synthesis by priming with random hexamers using SuperScript IV (SSIV) reverse transcriptase (Invitrogen). Synthesized cDNA was diluted 1:25 and used for quantitative real-time PCR (qPCR). qPCR experiments were performed using PowerUp SYBR master mix (Life Technologies, Carlsbad, CA) on a QuantStudio6 fast real-time PCR system (Applied Biosystems, Foster City, CA). Reaction mixtures were set up with the following primer pairs listed in Table 2: VPbccT1Fwd/Rev, VPbccT2Fwd/Rev, VPbccT3Fwd/Rev, VPbccT4Fwd/Rev, VPectAFwd/Rev, VPasp_ectFwd/Rev, VPproV1Fwd/Rev, VPAbetIFwd/Rev, VPAbetBFwd/Rev, VPAproXFwd/Rev, VPAproWFwd/Rev, and 16SFwd/Rev for normalization. Expression levels were quantified using cycle threshold (C_T) and were normalized to 16S rRNA. Differences in gene expression were determined using the ΔΔC_T method (69).

Protein purification of CosR.

CosR was purified as described previously (17). Briefly, full-length cosR (VP1906) was cloned into the protein expression vector pET28a (+) containing an IPTG-inducible promoter and a C-terminal 6×-His tag (Novagen). Expression of CosR-His was then induced in E. coli BL21(DE3) with 0.5 mM IPTG at OD₅₉₅ of 0.4 and grown overnight at room temperature. Cells were harvested, resuspended in lysis buffer (50 mM NaPO4, 200 mM NaCl, and 20 mM imidazole buffer [pH 7.4]), and lysed using a microfluidizer. CosR-His was bound to an Ni-nitrilotriacetic acid (NTA) column and eluted with 50 mM NaPO4, 200 mM NaCl, and 500 mM imidazole buffer [pH 7.4] after a series of washes to remove loosely bound protein. Protein purity was determined via SDS-PAGE. OpaR was purified as described previously (70).

Electrophoretic mobility shift assay.

Five overlapping DNA fragments, designated PbetI probe A (125 -bp), probe B (112 bp), probe C (142 bp), probe D (202 bp), and probe E (158 bp), were generated from the betIBA-proXWV regulatory region (includes 36 bp of the coding region and the 594-bp upstream intergenic region) using primer sets listed in Table 2. Three overlapping DNA fragments, designated PbccT1 probe A (120 bp), probe B (110 bp), and probe C (101 bp), were generated from the bccT1 regulatory region (includes 15 bp of the coding region and the 276-bp upstream intergenic region) using primer sets listed in Table 2. Two overlapping DNA fragments, designated PbccT3 probe A (108 bp) and probe B (107 bp), were generated from the bccT3 regulatory region (includes 17 bp of the coding region and 179 bp of the upstream intergenic region) using primer sets listed in Table 2. Four overlapping DNA fragments, designated PproV1 probe A (160 bp), probe B (134 bp), probe C (108 bp), and probe D (109 bp), were generated from the proV1 regulatory region (includes 9 bp of the coding region and the 438-bp upstream intergenic region) using primer sets listed in Table 2. Fragments designated PbccT2 (233 bp) and PbccT4 (244 bp) were generated from the bccT2 and bccT4 regulatory regions, respectively, using primers listed in Table 2. Two overlapping DNA fragments, designated PcosR probe A (105 bp) and probe B (142 bp), were generated from the cosR regulatory region (includes 4 bp of the coding region and 216 bp of the upstream intergenic region) using primer sets listed in Table 2. The concentration of purified CosR-His and OpaR was determined using a Bradford assay. CosR or OpaR was incubated for 20 min with 30 ng of each DNA fragment in a defined binding buffer (10 mM Tris, 150 mM KCl, 0.5 mM dithiothreitol, 0.1 mM EDTA, and 5% polyethylene glycol [PEG] [pH 7.9 at 4°C]). A 6% native acrylamide gel was prerun for 2 h at 4°C (200 V) in 1× Tris-acetate-EDTA (TAE) buffer. Gels were loaded with the DNA:protein mixtures (10 μl) and run for 2 h at 4°C (200 V). Finally, gels were stained in an ethidium bromide bath for 15 min and imaged.

Reporter assays.

A GFP reporter assay was conducted using the E. coli strain MKH13 (71). GFP reporter plasmids were constructed as previously described (17). Briefly, each regulatory region of interest was amplified using primers listed in Table 2 and ligated via the Gibson assembly protocol with the promoterless parent vector pRU1064, which had been digested with SpeI, to generate reporter plasmids with GFP under the control of the regulatory region of interest. Complementary regions for Gibson assembly are indicated in lowercase letters in the primer sequences (Table 2). Reporter plasmid P_betI-gfp encompasses 594 bp upstream of the betIBA-proXWV operon. Reporter plasmid P_bccT1-gfp encompasses 278 bp upstream of the PbccT1 regulatory region. Reporter plasmid P_bccT3-gfp encompasses 397 bp upstream of the PbccT3 regulatory region. Reporter plasmid P_proV1-gfp encompasses 438 bp upstream of the PproV1 regulatory region. Reporter plasmid P_cosR-gfp encompasses 397 bp upstream of the PcosR regulatory region. Full-length cosR was then expressed from an IPTG-inducible promoter in the pBBR1MCS expression vector. Relative fluorescence units (RFUs) and OD₅₉₅ were measured; specific fluorescence was calculated by dividing RFU by OD₅₉₅. Strains were grown overnight with aeration at 37°C in LB1% with ampicillin (50 μg/ml) and chloramphenicol (12.5 μg/ml), washed twice with 1× PBS, and then diluted 1:1000 in M9G1%. Expression of cosR was induced with 0.25 mM IPTG, and strains were grown for 20 h at 37°C with aeration under antibiotic selection. GFP fluorescence was measured with excitation at 385 and emission at 509 nm in black, clear-bottom 96-well plates on a Spark microplate reader with Magellan software (Tecan Systems Inc., San Jose, CA). Specific fluorescence was calculated for each sample by normalizing fluorescence intensity to OD₅₉₅. Two biological replicates were performed for each assay.

A GFP reporter assay was conducted in RIMD2210633 wild-type, ΔbetI, ΔopaR, or ΔcosR mutant strains. The P_betI-gfp or P_cosR-gfp reporter plasmid was transformed into E. coli β2155 λpir and conjugated into wild-type, ΔbetI, ΔopaR, or ΔcosR mutant strains. Strains were grown overnight with aeration at 37°C in LB3% with tetracycline (1 μg/ml). Cells were then pelleted, washed two times with 1× PBS, diluted 1:100 into M9G3%, and grown for 20 h with antibiotic selection. Choline was added to a final concentration of 1 mM, where indicated. GFP fluorescence was measured with excitation at 385 and emission at 509 nm in black, clear-bottom 96-well plates on a Spark microplate reader with Magellan software (Tecan Systems Inc.). Specific fluorescence was calculated for each sample by normalizing fluorescence intensity to OD₅₉₅. Two biological replicates were performed for each assay.

Bioinformatics and phylogenetic analyses.

Sequences of EMSA probes PectA A and B; PbetI A, B, and D; PbccT1 B; PbccT3 A; and PproV1 D to which CosR bound were input into the multiple EM for motif elicitation (MEME) tool (meme-suite.org/tools/meme) (49). We set the parameters to search for one occurrence of one motif per sequence, with a minimum width of 18 bp and a maximum width of 35 bp. The V. parahaemolyticus protein CosR (BAC60169) was used as a seed for BLASTp to identify homologs in the Vibrionaceae family in the NCBI database. Sequences of representative strains were downloaded from NCBI and used in a Python-based program Easyfig to visualize gene arrangements (71). GenBank accession numbers for select strains were the following: BA000031 (V. parahaemolyticus RIMD), CP016229 (Vibrio crassostreae 9CS106), CP031056 (V. splendidus BST398), NZ_FLQZ01000088 (V. celticus CECT7224), NZ_MCUE01000044 (V. lentus 10N.286.51.B9), FM954973 (V. tasmaniensis LGP32), CP039701 (V. cyclitrophicus ECSMB14105), CP000021 (Aliivibrio fischeri ES114), CP001133 (A. fischeri MJ11), LN554847 (A. wodanis AWOD1), and LR721750 (A. wodanis 06/09/160). V. parahaemolyticus RIMD2201633 CosR and A. fischeri ES114 EctR protein sequences were retrieved from NCBI using accession numbers BAC60169 and AAW88191.1, respectively, and were input into the SWISS-MODEL workspace, which generated a three-dimensional (3D) model of a homodimer to identify putative ligand-binding pockets (72–76). Evolutionary analysis was performed on the CosR protein from all species within the family Vibrionaceae with completed genome sequence, and as an outgroup, we used CosR from members of the genus Aeromonas. Protein sequences were obtained from the NCBI database and aligned using the Clustal W algorithm (77). Aligned protein sequences were used to generate a neighbor-joining tree with a bootstrap value of 1,000 (78, 79). The evolutionary distances were computed using the Jones-Taylor-Thornton (JTT) matrix-based method and are in the units of the number of amino acid substitutions per site (80). The rate variation among sites was modeled with a gamma distribution (shape parameter of 5). This analysis involved 96 amino acid sequences. All ambiguous positions were removed for each sequence pair using the pairwise deletion option. There were a total of 173 positions in the final data set. Evolutionary analyses were conducted in MEGA X (81).