Regulatory Small RNA Qrr2 Is Expressed Independently of Sigma Factor-54 and Can Function as the Sole Qrr Small RNA To Control Quorum Sensing in Vibrio parahaemolyticus

ABSTRACT

Bacterial cells alter gene expression in response to changes in population density in a process called quorum sensing (QS). In Vibrio harveyi, LuxO, a low-cell-density activator of sigma factor-54 (RpoN), is required for transcription of five noncoding regulatory small RNAs (sRNAs), Qrr1 to Qrr5, which each repress translation of the master QS regulator, LuxR. Vibrio parahaemolyticus, the leading cause of bacterial seafoodborne gastroenteritis, also contains five Qrr sRNAs that control OpaR (the LuxR homolog), controlling capsule polysaccharide (CPS), motility, and metabolism. We show that in a ΔluxO deletion mutant, opaR was derepressed and CPS and biofilm were produced. However, in a ΔrpoN mutant, opaR was repressed, no CPS was produced, and less biofilm production was observed than in the wild type. To determine why opaR was repressed, expression analysis in ΔluxO showed that all five qrr genes were repressed, while in ΔrpoN the qrr2 gene was significantly derepressed. Reporter assays and mutant analysis showed that Qrr2 sRNA can act alone to control OpaR. Bioinformatics analysis identified a sigma-70 (RpoD) −35 −10 promoter overlapping the canonical sigma-54 (RpoN) –24 –12 promoter in the qrr2 regulatory region. The qrr2 sigma-70 promoter element was also present in additional Vibrio species, indicating that it is widespread. Mutagenesis of the sigma-70 –10 promoter site in the ΔrpoN mutant background resulted in repression of qrr2. Analysis of qrr quadruple deletion mutants, in which only a single qrr gene is present, showed that only Qrr2 sRNA can act independently to regulate opaR. Mutant and expression data also demonstrated that RpoN and the global regulator, Fis, act additively to repress qrr2. Our data have uncovered a new mechanism of qrr expression and show that Qrr2 sRNA is sufficient for OpaR regulation.

IMPORTANCE The quorum sensing noncoding small RNAs (sRNAs) are present in all Vibrio species but vary in number and regulatory roles among species. In the Harveyi clade, all species contain five qrr genes, and in Vibrio harveyi these are transcribed by sigma-54 and are additive in function. In the Cholerae clade, four qrr genes are present, and in Vibrio cholerae the qrr genes are redundant in function. In Vibrio parahaemolyticus, qrr2 is controlled by two overlapping promoters. In an rpoN mutant, qrr2 is transcribed from a sigma-70 promoter that is present in all V. parahaemolyticus strains and in other species of the Harveyi clade, suggesting a conserved mechanism of regulation. Qrr2 sRNA can function as the sole Qrr sRNA to control OpaR.

INTRODUCTION

Bacteria monitor changes in cell density using a process termed quorum sensing (QS) (1, 2). QS is a regulatory mechanism used to alter global gene expression in response to cell density changes (1–6). In many Gram-negative bacteria, N-acylhomoserine lactone (AHL) is a common QS autoinducer synthesized intracellularly and secreted out of the cell (2, 7). By surveying AHL levels in its environment, a bacterium can regulate gene expression in response to growth phase. Quorum sensing has been characterized in several marine species in the genus Vibrio, including V. anguillarum, V. cholerae, V. harveyi, and V. parahaemolyticus, and shown to modulate expression of bioluminescence, capsule formation, biofilm, natural competence, swarming motility, and virulence (5, 7–20). Many of the original QS studies were performed in Vibrio harveyi ATCC BAA-1116, also known as BB120; however, this strain has been reclassified as V. campbellii, but to avoid confusion with published literature we will continue to use the name V. harveyi (21). In V. harveyi and V. anguillarum, it was shown that LuxO, the QS response regulator, is an activator of sigma factor-54, encoded by rpoN, which along with RNA polymerase, initiates transcription of the noncoding quorum regulatory small RNAs (Qrr) (8, 22, 23).

Noncoding sRNAs are a group of regulators present in prokaryotes that together with the RNA chaperone Hfq control gene expression in a range of phenotypes (24–26). The Qrr sRNAs are classified as trans-acting sRNAs that, along with Hfq, target mRNA via base pairing to the 5′ untranscribed region (UTR) to stabilize or destabilize translation (27–29). In V. harveyi, the nucleoid structuring protein Fis was shown to be a positive regulator of qrr gene expression (30). The Qrr sRNAs are posttranscriptional regulators that, in V. harveyi, enhanced translation of the QS low-cell-density (LCD) master regulator AphA and inhibited translation of the QS high-cell-density (HCD) master regulator, LuxR (28, 31–33). At HCD in V. harveyi, LuxO is not phosphorylated and therefore cannot activate sigma-54 (RpoN), the five Qrr sRNA genes qrr1 to qrr5 are repressed, and LuxR translation is derepressed. In addition, AphA and LuxR repress each other transcriptionally, providing a further level of regulation (32, 34–36). Studies have shown that in V. harveyi, Qrr1 has a 9-bp deletion in the 5′ region of the sRNA and therefore cannot activate aphA translation but can still repress luxR translation. The deletion in qrr1 is also present in V. cholerae, V. parahaemolyticus, and several other Vibrio species (32, 37). In V. harveyi, Qrr2, Qrr3, Qrr4, and Qrr5 sRNAs were additive in function and controlled the same target sites (27, 28, 38). However, the qrr genes showed distinct expression patterns and controlled the QS output signal at different levels coordinated with highest to lowest expression, Qrr4 > Qrr2 > Qrr3 > Qrr1 > Qrr5 (28). V. cholerae carries four Qrr sRNAs, Qrr1 to Qrr4, that were redundant in function, with any one of the four Qrr sRNAs sufficient to repress HapR (the LuxR homolog) (27).

Vibrio parahaemolyticus (VP) is a halophile, residing in marine environments as a free-living organism or in association with marine flora and fauna (39–41). This species is the leading cause of seafoodborne bacterial gastroenteritis worldwide, causing increasing infections each year, and is also a serious pathogen in the aquaculture industry (42, 43). Vibrio parahaemolyticus has dual flagellar systems, with the lateral flagellum system required for swarming motility, an important multicellular behavior (44). Vibrio parahaemolyticus has the same QS components and pathway as V. harveyi, containing five Qrr sRNAs that are predicted to control aphA and opaR (Fig. 1). In this species, the LuxR homolog is named OpaR (opacity regulator), for its role as an activator of capsule polysaccharide (CPS) production that results in an opaque, rugose colony morphology (45). The McCarter group first showed that an opaR deletion mutant in strain BB22OP had a translucent, smooth colony morphology, reduced CPS, and biofilm (46). A further study by the same group also showed that ΔopaR produced less biofilm when grown statically at 30°C for 16 h but showed similar biofilm levels at 24 h and more biofilm than the wild type after 48 h of growth (20). In V. parahaemolyticus RIMD2210633, ΔopaR also showed a defect in CPS and biofilm, but the biofilm defect disappeared after 24 h of growth (14). A recent study in this same strain proposed that opaR was a negative regulator of biofilm through modulation of c-di-GMP, although the method and time point examined differed from the previous study (47). A third strain, HZ, with a deletion in opaR also showed a defect in biofilm compared to the wild type at both 24 h and 48 h (48). These data indicate that the role of CPS production in biofilm formation is complex, as has been shown for other Vibrio species, and is influenced by the methods and time points used (49–52). In addition to CPS and biofilm formation, OpaR was shown to regulate swimming and swarming motility, surface sensing, metabolism, and the osmotic stress response in this species (14, 15, 20, 53–56). A V. parahaemolyticus ΔluxO deletion mutant, in which the qrr sRNAs were not expressed, showed that opaR was highly induced and produced both CPS and biofilm, output signals of the QS pathway (14). Interestingly, an earlier study examining an ΔrpoN deletion mutant in V. parahaemolyticus showed that it did not produce CPS and generated less biofilm than the wild type (57). This is unexpected because previous studies of V. harveyi suggest that in a ΔrpoN mutant, the qrr sRNA were repressed and the luxR (opaR homolog) was derepressed, inducing bioluminescence (8). In V. parahaemolyticus, OpaR is a positive regulator of the CPS biosynthesis operon, and therefore we expect that in a ΔrpoN mutant CPS should be produced.

FIG 1.

Vibrio parahaemolyticus quorum sensing pathway. Autoinducers (AIs) are synthesized internally by three synthases and then excreted outside the cell. At low cell density, three histidine-kinase receptors are free of AIs and therefore act as kinases, phosphorylating LuxU and ultimately LuxO. LuxO-P activates RpoN and, along with Fis, positively regulates transcription of five small quorum regulatory RNAs (Qrr sRNAs). The Qrr sRNAs, along with Hfq, stabilize aphA transcripts and destabilize opaR transcripts. In addition, AphA is a negative regulator of opaR expression. At high cell density, LuxO is unphosphorylated and inactivated, no qrrs are transcribed, opaR is expressed, and aphA is repressed. OpaR positively regulates capsule polysaccharide production (CPS), biofilm formation, type 6 secretion system 1, and the type IV pilin MSHA, among other genes. OpaR negatively regulates swarming motility, surface sensing, and two contact-dependent secretion systems, T3SS-1 and T6SS-1.

Here, we examined mutants of the QS pathway in V. parahaemolyticus to determine why the QS pathway output phenotypes differ between the ΔluxO and ΔrpoN mutants. We examined single and double mutants of ΔluxO and ΔrpoN for CPS and biofilm formation. Then, we determined the expression patterns of opaR, aphA, and the five qrr genes in these mutants. Data showed that qrr2 was derepressed in the ΔrpoN mutant and opaR was repressed. Bioinformatics analysis of the qrr2 regulatory region identified an RpoD −35 −10 promoter region that overlapped with the RpoN –24 –12 promoter, suggesting a mechanism by which qrr2 is expressed in the ΔrpoN mutant. We performed mutagenesis analysis of the RpoD promoter to examine this further. To determine whether the other Qrr sRNAs can also function independently, we constructed quadruple qrr mutants, in which only one qrr is present, and examined CPS and motility phenotypes. Using a DNA affinity pulldown assay, we identified several potential novel regulators of qrr2. One of these, Fis, was examined further to determine its role in qrr2 expression. We preformed sequence comparative analysis of the promoter regions of qrr2 among several species within the Harveyi clade to determine if the presence of an RpoD promoter is prevalent. Overall, our data showed that Qrr2 can function solely to control OpaR and has a novel mechanism of expression in V. parahaemolyticus.

RESULTS

Differential expression of opaR and aphA in ΔluxO versus ΔrpoN mutants.

We used CPS production as a readout of the presence of OpaR in the V. parahaemolyticus cell. Based on the quorum sensing pathway in V. harveyi, we would expect both a ΔluxO and ΔrpoN deletion mutant to produce CPS, as the Qrr sRNAs should be repressed, and therefore OpaR should be derepressed (Fig. 1). In a CPS assay, the ΔluxO mutant produced CPS forming opaque, rugose colonies. However, the ΔrpoN mutant did not produce CPS and, instead, formed a translucent, smooth colony morphology, similar to that of the ΔopaR strain (Fig. 2A). Complementation of ΔrpoN with a functional copy of rpoN from an expression plasmid restored the CPS phenotype, and similarly, complementation of ΔopaR with a functional copy of opaR restored CPS production (see Fig. S1A and B in the supplemental material). In addition, a ΔrpoN ΔluxO double mutant also lacked CPS and produced a translucent, smooth colony morphology. Similarly, when we examined biofilm formation, both the ΔrpoN and ΔrpoN ΔluxO double mutant strains produced less biofilm than the wild type and ΔluxO strains (Fig. 2B). These data suggest that in the ΔrpoN mutant, opaR is repressed. To test this, we complemented the ΔrpoN mutant with the opaR gene (pBADopaR), and in these cells CPS production was restored, indicating that the absence of opaR in the ΔrpoN mutant led to the CPS defect (Fig. S1C).

FIG 2.

Quorum sensing phenotypes. (A) Wild-type (WT) and QS mutant strain production of capsule polysaccharide (CPS) and colony morphology on Congo red plates. (B) Biofilm assay from cultures grown for 24 h, stagnant and stained with crystal violet. Images are representatives from three bio-reps. Biofilm quantification of three bio-reps in duplicate. Statistics were calculated using Student’s t test. ***, P < 0.001.

Next, we investigated the expression profiles of the QS master regulators in the ΔluxO and ΔrpoN deletion mutants. RNA isolation and quantitative real-time PCR (qPCR) assays were performed from cells grown in LB 3% NaCl to optical densities (OD) of 0.1 and 0.5. At OD 0.1, expression of opaR in ΔluxO relative to the wild type was significantly upregulated; however, opaR expression was unchanged in the ΔrpoN mutant (Fig. 3A). At OD 0.5, expression of opaR in ΔluxO matched that of the wild type; however, in the ΔrpoN mutant expression of opaR was significantly downregulated relative to that of the wild type (Fig. 3B). Expression of aphA, the low-cell-density QS master regulator, was repressed in the ΔluxO mutant compared to that of the wild type and unchanged in the ΔrpoN mutant at OD 0.1 (Fig. 3C). At OD 0.5, aphA expression was upregulated compared to that of the wild type in ΔrpoN (Fig. 3D). These data indicate that opaR is repressed and aphA is induced in the ΔrpoN deletion mutant.

FIG 3.

Quantitative real-time PCR expression analysis of cells grown to 0.1 (A, C) or 0.5 OD (B, D) in LB medium supplemented with 3% NaCl. Bars represent the expression of opaR and aphA in ΔluxO and ΔrpoN mutants normalized to the expression of the 16S rRNA housekeeping gene, relative to expression in the wild type. The means and standard error of at least two biological replicates are shown. Statistics were calculated using Student’s t test. *, P < 0.05; **, P < 0.01.

Expression analysis of qrr1 to qrr5 in ΔluxO and ΔrpoN mutants.

Since opaR showed different levels of expression in the ΔluxO and ΔrpoN deletion mutants, we wanted to determine whether this was due to differences in qrr expression levels. We examined the expression of all five qrr genes in cells grown to OD 0.1 and OD 0.5 and showed that the expression of qrr1, qrr2, qrr3, and qrr5 was higher at OD 0.1 than OD 0.5 (Fig. S2). Expression of qrr4 was detected at OD 0.1 but was not detected at OD 0.5 in the wild type. In addition, qrr4 expression was not detected in either ΔluxO or ΔrpoN at OD 0.1 or 0.5 indicating that it has a strict requirement for LuxO and RpoN, but expression of qrr4 in the wild type at OD 0.1 matched expression of the other qrr genes. This suggests that the main role of qrr4 could be regulation at low cell density. Next, we examined expression of the qrr genes at OD 0.1 in the ΔluxO mutant relative to the wild type and showed that qrr1 expression was unchanged, but there was significant downregulation of qrr2, qrr3, and qrr5 (Fig. 4A), whereas at OD 0.5, their expression matched that of the wild type (Fig. 4B). In the ΔrpoN mutant, expression of qrr1, qrr3, and qrr5 matched that of the ΔluxO mutant (Fig. 4C); however, qrr2 was upregulated at both OD 0.1 and OD 0.5 (Fig. 4D). To confirm that qrr2 was differentially regulated between ΔluxO and ΔrpoN, the qrr2 regulatory region was cloned into the pRU1064 reporter vector upstream of a promoter-less gfp cassette (Pqrr2-gfp). The specific fluorescence of Pqrr2-gfp was examined in the wild type, ΔluxO, and ΔrpoN mutants and measured as a cumulative readout of qrr2 transcription (Fig. 5A). The level of specific fluorescence of Pqrr2-gfp was reduced in the ΔluxO mutant relative to that of the wild type, whereas in the ΔrpoN mutant, fluorescence was significantly increased (Fig. 5A). Next, we examined the opaR regulatory region cloned into the pRU1064 reporter vector upstream of a promoter-less gfp cassette (PopaR-gfp) in the wild type, a Δqrr2 single mutant, and a Δqrr3,1,4,5 quadruple mutant with only qrr2 present (Fig. 5B). In Δqrr2 compared to the wild type, PopaR-gfp showed significantly increased fluorescence, whereas the quadruple qrr deletion mutant, with qrr2 present, was similar to the wild type (Fig. 5B). We predicted that deletion of qrr2 in the ΔrpoN mutant background should restore opaR expression and CPS production. We constructed a ΔrpoN Δqrr2 double mutant and examined opaR and aphA expression levels (Fig. S3). Quantitative real-time PCR assays showed that opaR was highly expressed in a ΔrpoN Δqrr2 double mutant compared to the wild type (Fig. S3). Examination of CPS formation showed that the ΔrpoN Δqrr2 double mutant produced a rough colony morphology (Fig. S4A). Similarly, in biofilm assays, the ΔrpoN mutant produced a significantly reduced biofilm, whereas the ΔrpoN Δqrr2 double mutant produced a biofilm similar to that of the wild type (Fig. S4B). Overall, these data demonstrate that Qrr2 sRNA is present in the ΔrpoN deletion mutant and that Qrr2 sRNA can function alone to control OpaR and QS phenotypes.

FIG 4.

Quantitative real-time PCR (qPCR) analysis of cells grown to OD 0.1 (A, C) or OD 0.5 (B, D) in LB medium supplemented with 3% NaCl. Expression of qrr1 to qrr5 in the ΔluxO and ΔrpoN mutants, relative to the wild type, RIMD2210633, and normalized to the 16S rRNA housekeeping gene, relative to wild-type expression of each gene. Expression of qrr4 was not detected in mutant strains. The means and standard error of at least two biological replicates are shown. Statistics were calculated using Student’s t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FIG 5.

Expression analysis of qrr2. (A) Pqrr2-gfp reporter assay of qrr2 in ΔluxO and ΔrpoN mutants. (B) PopaR-gfp reporter assays in a single qrr2 deletion mutant and a quadruple mutant with only qrr2 present. Cultures were grown for 20 h in LB 3% NaCl. The means and standard error of at least three biological replicates are shown. Statistics were calculated using a one-way analysis of variance (ANOVA) and Tukey-Kramer post hoc test. **, P < 0.01.

Overlapping sigma-70 and sigma-54 promoters.

The expression of qrr2 in the ΔrpoN mutant background indicates that an additional sigma factor can initiate qrr2 transcription. To examine this further, the regulatory regions of qrr1 to qrr5 in V. parahaemolyticus RIMD2210633 were aligned and surveyed for the presence of additional promoter regions using the Virtual Footprint promoter analysis program and manual scanning of sequences. Although the five Qrr sRNAs share homology, their regulatory regions are divergent, with the exception of the sigma-54 canonical −24 (TTGGCA) and −12 (AATGCA) promoter sites, with the nucleotides in bold conserved among all five qrr regulatory regions (Fig. S5). In the regulatory region of qrr2, promoter analysis identified a housekeeping sigma-70 (RpoD) –35 (TTGAAA) and –10 (ATAATA) promoter (Fig. 6A). The putative sigma-70 promoter overlapped with the sigma-54 –24 and –12 promoter (Fig. 6A) and was absent from the regulatory regions of qrr1, qrr3, qrr4, and qrr5 (Fig. S5). This suggested that qrr2 can be transcribed by either sigma-54 or sigma-70 and could explain its expression in the absence of rpoN. To examine this further, we mutated three base pairs of the putative sigma-70 –10 ATAATA site to ATACCC in the pRUPqrr2 reporter vector (Fig. 6A). The mutagenized vector, pRUPqrr2-10CCC, was conjugated into wild type and ΔrpoN, and specific fluorescence was determined. The ΔrpoN pRUPqrr2-10CCC strain showed significantly reduced fluorescence relative to ΔrpoN pRUPqrr2, indicating that this site is required for qrr2 transcription in the absence of RpoN (Fig. 6B). The data suggest that qrr2 can be transcribed by two sigma factors using dual overlapping promoters, suggesting a unique mode of regulation for Qrr2 sRNA. Comparisons of the qrr2 regulatory region among Harveyi clade species V. alginolyticus, V. campbellii, V. harveyi, and V. parahaemolyticus showed that the sigma-70 promoter –10 region was highly conserved among these species (Fig. S6). Each of the five Qrr sRNAs also shared homology among these species (Fig. 7). The qrr1 gene among all four species showed high homology, clustering closely together on the phylogenetic tree, but were distantly related to the other four qrr genes. The qrr3 and qrr4 genes each clustered tightly together on the tree, whereas qrr2 and qrr5 each showed divergence among the species (Fig. 7). Overall divergence in regulatory regions and gene sequence among the qrr genes likely suggests differences in how each qrr gene is regulated and differences in the target genes of each Qrr sRNA.

FIG 6.

qrr2 promoter analysis. (A) Analysis of the qrr2 regulatory region indicates overlapping sigma-54 and sigma-70 promoters. (B) Pqrr2 GFP reporter assay of qrr2 in ΔrpoN relative to the wild-type and mutated putative −10 RpoD binding site are indicated by asterisks. The means and standard error of three biological replicates shown. Statistics were calculated using a one-way ANOVA and Tukey-Kramer post hoc test. ***, P < 0.001.

FIG 7.

Phylogenetic tree of the qrr genes from V. harveyi ATCC 33843, V. campbellii ATCC BAA-1116 (formerly V. harveyi), V. parahaemolyticus RIMD2210633, and V. alginolyticus FDAARGOS_114. The numbers along the branches indicate bootstrap values. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. This analysis involved 20 nucleotide sequences.

Qrr2 sRNA can function independently of the other Qrr sRNAs.

Next, we determined whether Qrr2 sRNA has a distinct role in this species and whether any of the four other qrr genes can act alone. Using a qrr1 to qrr5 quintuple deletion mutant (Δqrr-null) and five quadruple qrr deletion mutants, each containing a single qrr, we examined several QS phenotypes (Fig. 8). In swarming motility assays, the Δqrr-null strain was swarming deficient, as swarming is negatively regulated by OpaR (Fig. 8A). In addition, four quadruple mutants, Δqrr3 Δqrr2 Δqrr4 Δqrr5, Δqrr2 qrr1 Δqrr4 Δqrr5, Δqrr3 Δqrr2 Δqrr1 Δqrr5, and Δqrr3 Δqrr2 Δqrr1 Δqrr4, were all swarming deficient, indicating that Qrr1, Qrr3, Qrr4, and Qrr5 sRNAs cannot function independently to control OpaR (Fig. 8A). In swarming motility assays, the Δqrr3 Δqrr1 Δqrr4 Δqrr5 mutant that contained only qrr2 behaved similarly to the wild type and was swarming proficient, indicating repression of opaR (Fig. 8A). In swimming assays, the quadruple mutants that lacked qrr2 produced similar results to the null mutant with defects in swimming (Fig. 8B), whereas only the mutant that contained only qrr2 showed swimming motility similar to that of the wild type (Fig. 8B). Additionally, in CPS assays, the qrr2-positive strain also showed a colony morphology similar to that of the wild type (Fig. 8C). Analysis of a single qrr2 deletion mutant indicates that it is not essential for CPS production or swarming and that the other qrr genes can function to control opaR in the absence of qrr2 (Fig. S7). In summary, these data demonstrate that only Qrr2 sRNA can function independently in V. parahaemolyticus.

FIG 8.

Phenotypes of qrr deletion mutants. (A) Swarming assay conducted on heart-infusion medium incubated at 30°C for 48 h. (B) Swimming motility assay conduced on semisolid agar plates grown at 37°C for 24 h. Swimming plate quantification of three biological replicates. Statistics were calculated using Student’s t test relative to the wild type. ***, P < 0.001. (C) CPS assays conducted of strains of interest. Colonies were grown on Congo red plates for 48 h at 30°C prior to imaging.

RpoN and Fis are not required for qrr2 expression.

In order to identify additional regulators of qrr2 transcription, a DNA-affinity pulldown was performed. We used ΔrpoN cell lysate grown to OD 0.5 and Pqrr2 bait DNA. We identified a number of candidate regulators previously shown to bind to the qrr sRNA regulatory regions in V. harveyi (30, 36, 58) (Fig. S8 and S9). We decided to examine the nucleoid-associated protein Fis further since it is known to be a positive regulator of qrr sRNA expression and binds to the qrr sRNA regulatory regions in both V. harveyi and V. parahaemolyticus (30, 59). The qrr2 regulatory region shows at least three Fis binding sites containing the conserved Fis binding motif previously described in V. parahaemolyticus (59). A Fis binding site was located adjacent to the −35 promoter site, as well as two additional Fis binding sites, at 193 bp and 229 bp upstream of the qrr2 transcriptional start site (Fig. 9A). To confirm these Fis binding sites, we constructed three DNA probes of the qrr2 regulatory region to use in electrophoretic shift mobility assays (EMSAs) with purified Fis protein. DNA probe 1A, encompassing a single binding site, showed binding and was similarly Fis bound to probe 1C, which contained two Fis binding sites, both probes bound in a concentration-dependent manner (Fig. 9B). Probe 1B, which did not have a putative Fis binding site, showed weak likely nonspecific binding. Next, we examined expression of the Pqrr2-gfp reporter in the wild type, ΔrpoN mutant, and a ΔrpoN Δfis double mutant and confirmed that expression of Pqrr2-gfp was upregulated in the ΔrpoN mutant but was even more highly upregulated in the ΔrpoN Δfis double mutant (Fig. 9C). These data indicate that both Fis and RpoN can act as repressors of qrr2 in V. parahaemolyticus, and Fis likely plays a role in enhancing sigma-54 binding.

FIG 9.

Fis binding in the qrr2 regulatory region. (A) The regulatory region of qrr2 is depicted. Lines represent EMSA probes, and triangles represent putative Fis binding sites using Virtual Footprint prediction software. Numbers indicate Fis binding site distance from qrr2 transcriptional start site. (B) Electrophoretic mobility shift assays of Pqrr2 with purified Fis protein using three qrr2 regulatory region DNA probes (C) pRUPqrr2 reporter assays in ΔrpoN and ΔrpoN Δfis deletion mutants relative to the WT. Cultures were grown for 20 h in LB 3% NaCl. The means and standard error of at least three biological replicates are shown. Statistics were calculated using a one-way ANOVA and Tukey-Kramer post hoc test. ***, P < 0.001.

DISCUSSION

In this study, we investigated the role of sigma-54, LuxO, and the five Qrr sRNAs in the V. parahaemolyticus QS pathway and showed that the QS pathway can function in the absence of sigma-54, qrr1, qrr3, qrr4, or qrr5. This observation reflects the idea that strains and species have different expression patterns of the qrr sRNAs under different growth conditions and each qrr gene is likely controlled by different factors. Our data demonstrated that in a ΔrpoN mutant, cells had a defect in CPS and biofilm formation, QS phenotypes that differed from the ΔluxO mutant. The data showed that Qrr2 is highly expressed in a ΔrpoN mutant and that Qrr2 can act independently of the other Qrr sRNAs to repress OpaR and QS phenotypes. In a ΔrpoN/Δqrr2 double mutant, opaR was derepressed and CPS and biofilm formation were restored. Bioinformatics analysis identified a putative −35 −10 promoter region within the qrr2 regulatory region, and mutagenesis of the –10 promoter sites resulted in repression of qrr2. Overall, the data indicate that qrr2 can be expressed from two promoters, and this ability is likely present in other related species. There have been other accounts of sigma-54-dependent genes showing increased transcription in the absence of rpoN (60, 61). In these cases, a putative sigma-70 promoter was present, suggesting a potential competition for promoter sites (60, 61). For example, in Escherichia coli, glmY, a coding sRNA, contained overlapping sigma-54 and sigma-70 promoters, which were shown to allow for precise control of glmY expression within the cell (62). In our study, we identified a sigma-70 promoter that overlaps with the sigma-54 consensus promoter sequence of qrr2, suggesting that RpoN under differ growth conditions may block RpoD access. We propose that in the wild-type background, qrr2 is transcribed via LuxO-activated RpoN, and in the ΔluxO mutant, qrr2 is not transcribed because sigma-54 is in an inactive state bound to the qrr2 promoter. RpoN physically blocking additional sigma factors from binding was previously proposed based on studies in other bacteria (63, 64). However, in the absence of sigma-54, sigma-70 binds to the qrr2 regulatory region at a conserved –35 and –10 region to initiate transcription (Fig. 10). Fis is a global regulator that is known to enhance and inhibit transcription from promoter regions in many bacterial species (65–68). In V. parahaemolyticus, Fis was shown to positively regulate qrr sRNA expression. The fis gene was shown to be highly expressed in exponential growth only, and Fis bound to the regulatory region of all five qrr sRNA genes (59). Here, we show in DNA protein binding assays in V. parahaemolyticus that Fis binds adjacent to the –35 promoter site. We speculate that Fis functions to enhance RpoN promoter binding to maximize qrr expression. The data showed that in the absence of both RpoN and Fis, however, qrr2 expression is significantly increased compared to that of the ΔrpoN mutant alone. Under these conditions additional binding sites within the qrr2 regulatory region may be fully exposed, allowing sigma-70 full access for increased qrr2 expression (Fig. 10). Our previous study has shown that in a Δfis deletion mutant qrr2 expression is repressed, suggesting that Fis may also block sigma-70 binding in exponential-phase cells (59). A study of V. alginolyticus MVP01, a species closely related to V. parahaemolyticus, also showed differences between the ΔluxO and ΔrpoN mutant strains in their control of cell density-dependent siderophore production. The ΔluxO mutant showed reduced siderophore production, which is negatively regulated by LuxR, and the ΔrpoN mutant showed increased production (69). These data showed RpoN-dependent and -independent siderophore production. We speculate that this could be the result of expression by RpoD since V. alginolyticus has an RpoD –35 –10 promoter in the Qrr2 regulatory region (see Fig. S6 in the supplemental material).

FIG 10.

Model for qrr2 transcription in the ΔluxO and ΔrpoN mutants. In the ΔluxO mutant, under certain conditions, RpoN will be bound to the qrr2 RpoN −24 −12 promoter region. RpoN bound at the promoter will be aided by Fis. This will prevent sigma-70 from binding. In the absence of RpoN (sigma-54), RpoD (sigma-70) binds to the –35 –10 promoter region to initiate transcription. In the absence of Fis in the ΔrpoN mutant, transcription by RpoD is increased further, as in the ΔrpoN Δfis mutant, which suggests that Fis may block RpoD binding in exponential-phase cells.

In V. cholerae, four Qrr sRNAs (Qrr1 to Qrr4) are present that were shown to act redundantly to control bioluminescence; that is, any one of the Qrr sRNAs is sufficient to control HapR (LuxR homolog) (27). In their study, Lenz and colleagues showed that it was not until all four Qrrs were deleted in V. cholerae that there was a difference in density-dependent bioluminescence (27). In V. harveyi, the five Qrrs were shown to act additively to control LuxR expression. Using bioluminescence assays and quadruple qrr mutants, it was determined that each Qrr has a different level of strength in repressing luxR translation (28). In V. parahaemolyticus, it appears that expression of qrr4 is restricted to low-cell-density cells, since we only observed expression at OD 0.1, and has an absolute requirement for LuxO and RpoN since we did not observe expression in either mutant. It is of interest to note that qrr1 expression was unchanged in both the luxO and rpoN mutants at both ODs compared to that of the wild type, which means this qrr uses an alternative sigma factor for transcription. We demonstrated that Qrr2 sRNA is the only Qrr that can act alone to control QS gene expression but is not essential under the conditions examined, since a Δqrr2 mutant behaves like the wild type (Fig. S7). Given that qrr2 can be transcribed independently of RpoN, this suggests that Qrr2 may have unique functions and/or targets in this species. We propose that V. parahaemolyticus can activate the transcription of qrr2 via RpoN or RpoD to alter gene expression in a timely manner, likely under different growth conditions. Overall, our data suggest that the expression of each qrr gene is controlled differently, and likely by a different set of regulators.

MATERIALS AND METHODS

Bacterial strains and media.

In this study, the wild-type (WT) strain is a streptomycin-resistant clinical isolate of Vibrio parahaemolyticus RIMD2210633, and all strains used are described in Table S1 in the supplemental material (70, 71). All V. parahaemolyticus strains were grown in lysogeny broth (LB; Fisher Scientific, Fair Lawn, NJ) supplemented with 3% NaCl (LBS) (weight/volume). E. coli strains were grown in LB 1% NaCl. A diaminopimelic acid (DAP) auxotroph of E. coli β2155 λpir was grown with 0.3 mM DAP in LB 1% NaCl. All strains were grown aerobically at 37°C. Antibiotics were used in the following concentrations: chloramphenicol (Cm), 12.5 μg/ml; streptomycin (Str), 200 μg/ml; and tetracycline (Tet), 1 μg/ml.

Construction of V. parahaemolyticus mutants.

We created the double deletion mutants ΔrpoN ΔluxO and ΔrpoN Δfis using mutant vectors pDSΔluxO and pDSΔfis, conjugated into the V. parahaemolyticus ΔrpoN mutant background. The Δqrr-null mutant was constructed by creating truncated, nonfunctional copies of each qrr using splicing by overlap extension (SOE) primer design with the primers listed in Table S2. All truncated qrr products were cloned into pDS132 suicide vector and transformed into the E. coli β2155 λpir, followed by conjugation and homologous recombination into the V. parahaemolyticus genome. Positive single-crossover colonies were selected using Cm. To induce a double-crossover event, a positive single-cross strain was grown overnight in the absence of Cm, leaving behind either the truncated qrr allele or the wild-type allele in the genome. The overnight culture was plated on sucrose plates for selection of normal versus soupy colony morphology, as the colonies still harboring the pDS132Δqrr vector appear irregular due to the sacB gene. Colonies were screened via PCR for the truncated allele and sequenced to confirm deletion. The qrr null mutant was constructed by deleting qrr genes in the following order: qrr3, qrr2, qrr1, qrr4, qrr5. The quadruple Δqrr3 Δqrr2 Δqrr4 Δqrr5 mutant was constructed by reintroducing qrr1 into the Δqrr-null mutant, and similarly, qrr2 and qrr3 were each separately cloned into the Δqrr-null mutant to create their corresponding quad mutants. The Δqrr3 Δqrr2 Δqrr1 Δqrr5 mutant was constructed by deleting qrr5 in the Δqrr3 Δqrr2 Δqrr1 mutant background, and Δqrr3 Δqrr2 Δqrr1 Δqrr4 was constructed by knocking out qrr4 in the Δqrr3 Δqrr2 Δqrr1 background. The Δqrr2 single mutant was constructed using the pDSΔqrr2 construct conjugated into the wild-type background. The ΔrpoN Δqrr2 mutant was constructed by conjugating the pDSΔqrr2 vector into the ΔrpoN background. All mutants were sequenced to confirm deletions or insertions, ensuring in-frame mutant strains.

Complementation analysis of QS mutants.

To confirm that the phenotypes observed in the QS mutants were not due to secondary mutations within the genome, the ΔrpoN and ΔopaR strains were complemented with functional copies of the gene. The opaR coding region plus 33 bp upstream to include the ribosomal binding site were amplified from V. parahaemolyticus RIMD2210633 genome via the Phusion high-fidelity (HF) polymerase PCR (New England Biolabs). The amplified 670-bp opaR coding region and pBAD33 empty vector (pBADEV) were digested with XbaI and HindIII restriction enzymes prior to ligation and transformation into E. coli β2155. pBADopaR was conjugated into the ΔopaR and ΔrpoN strains. For complementation of the ΔrpoN strain with a functional copy of rpoN, a similar procedure was followed, but using Gibson assembly. Briefly, a full-length copy of the rpoN gene was PCR amplified and cloned into the expression vector pBAD using Gibson assembly primers, transformed into E. coli, and conjugated into the ΔrpoN strain, designated ΔrpoNprpoN. The complementation primers can be found in Table S2.

RNA isolation and real-time PCR.

Vibrio parahaemolyticus wild type and mutants were grown overnight in LBS. Cells were washed twice with 1× phosphate-buffered saline (PBS) and diluted 1:50 into a fresh 5-ml culture of LBS. Cells were harvested at 0.1 OD and 0.5 OD and pelleted at 4°C. RNA was isolated from 4 ml of culture using the miRNAeasy minikit (Qiagen, Hilden, Germany) and Qiazol lysis reagent. The concentration and purity of RNA were determined using a NanoDrop spectrophotometer (Thermo Scientific, Waltham, MA). RNA was treated with Turbo DNase (Invitrogen), and cDNA was synthesized using Superscript IV reverse transcriptase (Invitrogen) from 500 ng of RNA by priming with random hexamers. cDNA was diluted 1:10 for quantitative real-time PCR (qPCR) run on an Applied Biosystems QuantStudio 6 fast real-time PCR system (Applied Biosystems, Foster City, CA) using PowerUp SYBR green master mix (Life Technologies). The qPCR primers used to amplify opaR, aphA, qrr1, qrr2, qrr3, qrr4, qrr5, and 16S rRNA are listed in Table S2 for reference. Cycle threshold (C_T) values were used to determine expression levels normalized to 16S rRNA levels. Expression was calculated relative to wild-type 16S rRNA using the ΔΔC_T method (72). The efficiency of the qPCR primers used in this study was determined using a standard curve and calculated to be between 90 and 110% with melting temperatures between 55 and 60°C. In the WT and at both 0.1 and 0.5 OD, all five qrrs came off between four C_T values of one another. Expression of qrr4 was undetermined at 0.5 OD in the WT and at 0.1 OD and 0.5 OD in the ΔluxO and ΔrproN mutants.

Transcriptional GFP-reporter assay.

The Pqrr2 reporter construct was created using the pRU1064 vector, which contains a promoter-less gfp cassette, as well as Tet and Amp resistance genes (73). Primers, listed in Table S2, were designed using NEBuilder online software to amplify the 337-bp regulatory region of qrr2 from V. parahaemolyticus RIMD2210633 genomic DNA. The pRU1064 vector was purified, digested with SpeI, and ligated with the Pqrr2 fragment via the Gibson assembly protocol (74). The plasmid was then transformed into β2155 λpir and subsequently conjugated into the wild type and the ΔluxO, ΔrpoN, and ΔrpoN Δfis mutants. Cultures were grown overnight in LBS with 1 μg/ml Tet, washed twice with 1× PBS, and then diluted 1:1,000 into fresh LBS plus Tet and grown for 20 h at 37°C. Cultures were washed twice with 1× PBS and loaded into a black, clear-bottom 96-well plate. Final OD and green fluorescent protein (GFP) relative fluoresces were determined using a Tecan Spark microplate reader with Magellan software with excitation at 385 nm and emission at 509 nm (Tecan Systems, Inc., San Jose, CA). Specific fluorescence was calculated by dividing the relative fluorescence by the final OD. This experiment was performed for three biological replicates.

Splicing by overlap extension (SOE) primer design was used to construct a mutated (ATA-10CCC) RpoD promoter. We used the same SOEqrr2A and SOEqrr2D primers used to construct the Δqrr2 mutant in order to create a mutated qrr2 regulatory region. In addition, SOE primers Pqrr2SDMB and Pqrr2SDMC (Table S2) have complementary overlapping sequences that amplify a mutated promoter, indicated in bold. Fragments AB and CD were then used as a template to amplify the AD fragment, containing a mutated RpoD −10 promoter. The AD fragment was then used as the template to create a fragment containing only the qrr2 regulatory region (337 bp) using Gibson assembly primers Pqrr2SDM_GAfwd and Pqrr2SDM_GArev. This mutated regulatory region was then ligated with SpeI-digested pRU1064 using Gibson assembly and confirmed via sequencing.

Capsule polysaccharide (CPS) formation assay.

Capsule polysaccharide (CPS) formation assays were conducted as previously described (14, 46). In brief, single colonies of the wild type and QS mutants were grown on heart infusion (HI) (Remel, Lenexa, KS) plates containing 1.5% agar, 2.5 mM CaCl₂, and 0.25% Congo red dye for 48 h at 30°C. Each image is an example from three biological replicates. The pBAD33 expression vector was used to overexpress opaR in the wild-type and ΔrpoN backgrounds as described in the complementation analysis section. pBADopaR and pBADEV were conjugated into ΔrpoN and ΔopaR and plated on Congo red plates to observe CPS formation. For strains containing pBAD, 0.1% (wt/vol) l-arabinose and 5 μg/ml of Cm were added to the medium after autoclaving, to induce and maintain the plasmid, respectively.

Biofilm assay.

Vibrio parahaemolyticus cultures were grown overnight in LBS at 37°C with shaking. The overnight cultures were then used to inoculate a 96-well microtiter plate in a 1:50 dilution with LBS. After static incubation at 37°C for 24 h, the culture liquid was removed, and the wells were washed with 1× PBS. Crystal violet (Electron Microscopy Sciences), at 0.1% wt/vol, was added to the wells and incubated for 30 min at room temperature. The crystal violet was removed, and wells were washed twice with 1× PBS. The adhered crystal violet was solubilized in dimethyl sulfoxide (DMSO) for an optical density reading at 595 nm (OD₅₉₅).

Motility assays.

Swimming and swarming assays were performed as previously described (14, 57). To assess swimming, a pipette tip was used to pick a single colony and stab into the center of an LB plate containing 2% NaCl and 0.3% agar. Plates were incubated for 24 h at 37°C. Three biological replicates were performed, and the diameter of growth was measured for quantification. Swarming assays were conducted on HI plates containing 2% NaCl and 1.5% agar and incubated at 30°C for 48 h before imaging.

Phylogenetic analysis.

The five qrr genes from four species were downloaded from GenBank and aligned using ClustalW (75). An evolutionary history was inferred by using the maximum likelihood method and Jukes-Cantor model in MEGA X (76, 77). The tree with the highest log likelihood (–467.92) was used. The percentage of trees in which the associated taxa clustered together is shown next to the branches (78). Initial trees for the heuristic search were obtained automatically by applying the neighbor-join and BioNJ algorithms to a matrix of pairwise distances estimated using the maximum composite likelihood (MCL) approach and then selecting the topology with the superior log likelihood value. A discrete gamma distribution was used to model evolutionary rate differences among sites (3 categories [+G, parameter = 0.2492]). The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. This analysis involved 20 nucleotide sequences.

DNA-affinity pulldown.

A DNA-affinity pulldown was performed using previously described methods, with modifications as needed (79–81). Bait DNA primers were designed to amplify the regulatory region of qrr2 (346 bp) with a biotin moiety added to the 5′ end. In addition, a negative-control bait DNA (VPA1624 coding region, 342 bp) was amplified. Both bait DNA probes were amplified using Phusion HF polymerase (New England Biolabs) PCR and purified using ethanol extraction techniques (82). A 5-ml overnight culture of ΔrpoN grown in LB 3% NaCl was used to inoculate a fresh 100-ml culture of LB 3% NaCl grown at 37°C with aeration. The culture was pelleted at 0.5 OD at 4°C for 30 min and stored overnight at 80°C. The cell pellet was suspended in 1.5 ml of FastBreak lysis buffer (Promega, Madison, WI) and sonicated to shear genomic DNA. The ΔrpoN lysate was precleared with streptavidin DynaBeads (Thermo Scientific, Waltham, MA) to remove nonspecific protein-bead interactions. Beads were incubated with 200 μl of probe DNA for 20 min, twice. The ΔrpoN lysate and sheared salmon sperm DNA (10 μg/ml), as competitive DNA, were incubated with the beads twice, and washed. Protein candidates were eluted from the bait DNA-bead complex using elution buffers containing increasing concentrations of NaCl (100 mM, 200 mM, 300 mM, 500 mM, 750 mM, and 1 M). Next, 6× SDS was added to samples along with 1 mM β-mercaptoethanol (BME) and then boiled at 95°C for 5 min. A total of 25 μl of each elution was run on 2 stain-free, 12% gels and visualized using the Pierce silver stain for mass spectrometry kit (Thermo Scientific, Waltham, MA). Pqrr2 bait and negative-control bait were loaded next to each other in order of increasing NaCl concentrations. Bands present in the Pqrr2 bait lanes, but not in the negative-control lanes, were selected and cut from the gel. Each fragment was digested separately with trypsin using standard procedures and prepared for mass spectrophotometry 18C ZipTips (Fisher Scientific, Fair Lawn, NJ). Candidates were eluted in 10 μl twice, pooled, and dried again using a SpeedVac instrument. Dried samples were analyzed using the Thermo Q Exactive Orbitrap mass spectrometer and analyzed using Proteome Discoverer 1.4.

Fis protein purification.

Fis was purified using the method previously described (59). Briefly, the primer pair FisFWDpMAL and FisREVpMAL was used to amplify fis (VP2885) from V. parahaemolyticus RIMD2210633. The fis gene was cloned into the pMAL-c5x expression vector fused to a 6× His tag maltose binding protein (MBP) separated by a Tobacco Etch Virus (TEV) protease cleavage site. Expression of pMALfis in E. coli BL21(DE3) was induced with 0.5 mM IPTG (isopropyl-β-d-thiogalactopyranoside) once the culture reached 0.4 OD₅₉₅ and was grown overnight at room temperature. Cells were harvested, suspended in lysis buffer (50 mM NaPO₄, 200 mM NaCl, and 20 mM imidazole buffer [pH 7.4]) and lysed using a microfluidizer. The lysed culture was subjected to immobilized metal affinity chromatography (IMAC) using HisPur Ni-NTA resin, followed by additional washing steps. After purification, the MBP tag was cleaved overnight at 4°C with a hexahistidine-tagged TEV protease in a 1:10 molar ratio. Mass spectrometry was performed to confirm the Fis protein molecular weight, and SDS-PAGE was conducted to determine its purity along with A260/280 ratio analysis using a NanoDrop spectrophotometer.

Electrophoretic mobility shift assay for Fis.

Purified Fis was used to conduct EMSAs using the conditions previously described (55, 59). Briefly, 30 ng of DNA probe was incubated with various concentrations of Fis (0 to 1.94 μM) in binding buffer (10 mM Tris, 150 mM KCl, 0.1 mM dithiothreitol, 0.1 mM EDTA, 5% polyethylene glycol, pH 7.4) for 20 min. The concentration of Fis was determined using Bradford reagent. A 6% native polyacrylamide gel was prerun for 2 h at 4°C (200 V) with 1× Tris-acetate-EDTA (TAE) buffer. The incubated DNA-protein samples were then loaded onto the gel (10 μl) and run for 2 h under the same conditions. The gel was stained in an ethidium bromide bath (0.5 μg/ml) for 15 min before imaging. Pqrr2 was further divided into smaller probes to determine the specificity of Fis binding to Pqrr2.