Analysis of LuxR Regulon Gene Expression during Quorum Sensing in Vibrio fischeri^▿

Bacterial strains and growth media.

The bacterial strains and plasmids used in this study are described in Table 1. Luria-Bertani broth (LB) was used to grow E. coli strains. It was supplemented, where indicated below, with ampicillin (Ap) (100 μg/ml), chloramphenicol (Cm) (20 μg/ml), 200 nM N-(β-ketocaproyl)-l-homoserine lactone (3-oxo-C6-HSL) (Sigma, St. Louis, MO), or 1 mM isopropyl-β-d-1-thiogalatopyranoside (IPTG) (Sigma). Seawater complete (SWC) medium with instant ocean substituted for natural seawater (Aquarium Systems, Mentor OH) (20) or LBS medium (6) was used to grow V. fischeri MJ-100.

Plasmid construction.

The acfA and luxI genes were previously cloned into pSUP102 (2). V. fischeri chromosomal DNA was purified as a template (37), and qsrP and ribB were amplified using the primers indicated in Table 2 and cloned into pSUP102. The promoter regions of a number of genes containing putative lux boxes (luxI, qsrP, ribB, acfA, ainS, qsrQ, qsrRST, and arcA) were PCR amplified using the primers indicated in Table 2, cloned into pGEM-T (Promega, Madison, WI), and sequenced to confirm their integrity (Virginia Bioinformatics Institute Core Laboratory, Virginia Tech, Blacksburg). The promoter sequences were subsequently subcloned into the promoter-probe β-galactosidase reporter pSP417 (23) using EcoRI/BamHI sites to construct pNQ101 to pNQ110 (Table 1). Two versions of the P_qsrQ and P_qsrRST promoters were cloned, one with the 5′ end just upstream of the putative lux box and the other with additional upstream sequences containing the divergently transcribed promoters P_qsrP and P_acfA, respectively. For the RT-PCR analysis, internal fragments of qsrP and nadB were PCR amplified and cloned into pGEM-T, creating pNQqsrP and pNQnadB.

β-Galactosidase assays.

The pNQ101 to pNQ110 constructs were transformed individually into E. coli JM109 competent cells containing pAMS121 (32) containing a luxR gene that could be induced by IPTG. Overnight cultures were initially inoculated from freezer stocks into 5 ml LB medium with appropriate antibiotics and incubated at 30°C. A subculture of each strain was prepared by transferring the correct amount of the overnight culture into 5 ml LB medium plus Ap (100 μg/ml) and Cm (20 μg/ml) with either 200 nM 3-oxo-C6-dl-HSL or 1 mM IPTG, both 3-oxo-C6-dl-HSL and IPTG, or nothing to obtain an optical density at 600 nm (OD₆₀₀) of 0.025. When the OD₆₀₀ reached 0.5, 5 μl of cells was diluted 1:200 in Z buffer (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄) with 400 μM dithiothreitol and lysed with 50 μl of chloroform. Chemiluminescent β-galactosidase assays (Tropix, Bedford, MA) were performed with 5 μl of cell lysate using an LD 400S luminescence detector (Beckman Coulter, Fullerton, CA) and a 20-s integration time. All assays were performed in three trials with triplicate samples in each trial, and the results were expressed in relative light units.

RNA purification.

For primer extension, overnight cultures were initially inoculated from freezer stocks into 5 ml LB medium plus Ap (100 μg/ml) and Cm (20 μg/ml) (for E. coli) or SWC medium with no antibiotics (for V. fischeri) and incubated at 30°C. A subculture of each strain was prepared by transferring an appropriate amount of the overnight culture into 5 ml LB medium plus Ap (100 μg/ml) and Cm (20 μg/ml) with and without 200 nM 3-oxo-C6-HSL or SWC medium with and without 200 nM 3-oxo-C6-HSL to obtain an OD₆₀₀ of 0.025. Then the cells were grown to an OD₆₀₀ of 1.0. Five milliliters of cells was harvested and resuspended in TE buffer (10 mM Tris, 2 mM EDTA) and passed through 19- and 26-gauge needles (Becton Dickinson & Co, Franklin Lakes, NJ) four times each to improve lysis. Subsequent steps were performed using the QIAGEN RNeasy kit protocol (QIAGEN, Valencia, CA). After purification, the RNA concentration was determined using a spectrophotometer. The sample was then dried in a Speedvac instrument and resuspended in RNase-free water.

For real-time RT-PCR, V. fischeri MJ-100 was grown in LBS medium to an OD₆₀₀ of 0.25, 0.5, 1.0, or 2.0. In addition to the QIAGEN RNeasy kit, QIAGEN RNeasy Protect bacterial reagent and QIAGEN RNase-free DNase I were also used to stabilize the RNA and remove the residual DNA. The RNA was sent to the Virginia Bioinformatics Institute Core Laboratory (QIAGEN) to check the quantity and quality of the RNA before use in real-time RT-PCR protocols.

Primer extension.

Each primer was 5′ end labeled using [γ-³²P]dATP (Amersham, Piscataway, NJ) and a Primer Extension System kit (Promega, Madison, WI). Primer extension was performed using this kit according to the manufacturer's instructions, except that an ethanol precipitation step was used to remove unincorporated [γ-³²P]dATP from the primer. For ethanol precipitation 90 μl RNase-free water, 11 μl 3 M RNase-free sodium acetate, and 220 μl ethanol were added, mixed by vortexing, and kept at room temperature for 1 h. DNA sequences were obtained using a plasmid template purified with a QIAGEN Miniprep kit (QIAGEN) and were diluted to obtain a concentration of 100 ng/μl. Sequencing reactions were performed using an fmol DNA cycle sequencing system kit (Promega) according to the manufacturer's instructions.

Real-time RT-PCR.

An Agilent Bioanalyzer 2100 (Agilent, Palo Alto, CA) was used first to make sure that there was no DNA or other contaminants in the RNA samples and to determine the quantities of RNA that had been purified. cDNA was made from 1 μg of total RNA using random hexamer primers and an iScrip cDNA synthesis kit (Bio-Rad, Hercules, CA). The resulting cDNA was diluted 1:10 and used as a template in RT-PCR with SYBR green detection using the Bio-Rad iCycler iQ real-time PCR detection system (Bio-Rad). The starting quantity of template for each sample was determined using a five-point standard curve generated by amplification of PCR products containing portions of the genes of interest from known quantities of plasmid templates. Melting curves were analyzed at the end of each PCR run, and controls included PCRs without total RNA and without the reverse transcriptase. RT-PCR was performed using the cDNA of each set of RNA in triplicate. The mean quantities of luxI, qsrP, ribB, and acfA transcripts were normalized based on the mean quantity of the control gene, nadB.

EMSA.

The EMSA protocol used was based on the method of Urbanowski et al. (34). A DNA probe that served as both positive and negative controls was derived from a 400-bp PCR fragment containing the luxI-luxR regulatory region using template pSH202 (3). This PCR product was labeled at both ends using [γ-³²P]ATP plus T4 nucleotide kinase and was subsequently cleaved with MwoI, resulting in 160- and 240-bp fragments. Other DNA probes were generated via PCR and contained approximately 130 bp of the luxI, qsrP, acfA, or ribB promoter region or the lux box of luxD (Table 2). Protein-DNA binding reaction mixtures contained 1 fmol of each DNA probe in 20 μl (final volume) of DNA binding buffer (20 mM Tris-HCl [pH 7.4 at 22°C], 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 100 μg of bovine serum albumin/ml, 5% glycerol). Purified LuxR and 6 μM 3-oxo-C6-l-HSL were added as indicated below, and the reaction mixtures were incubated at 26°C for 20 min. One microliter of loading dye (0.1% xylene cyanol in 50% glycerol) was added, the reaction mixtures were loaded onto a native 5% Tris-glycine-EDTA gel on ice, and electrophoresis was performed at 10 V/cm for 2 h at 4°C.

LuxR dependence of V. fischeri promoters in recombinant E. coli.

β-Galactosidase assays were performed with recombinant E. coli/pAMS121 cells containing pNQ101 to pNQ110 individually (Table 1) in order to establish the presence of LuxR-dependent gene expression from the constructs. The reporter constructs contained the promoters for luxI, qsrP, ribB, acfA, ainS, qsrQ (two versions), qsrRST (two versions), and arcA from V. fischeri MJ-100 fused to lacZ. Cells were grown to the mid-exponential phase (OD₆₀₀, 0.5), when the LuxR-dependent quorum-sensing response is normally upregulated. Growth media with four different combinations of IPTG and 3-oxo-C6-HSL levels were used to influence the expression and activity of the LuxR encoded on pAMS121. Four promoters controlling luxI, qsrP, ribB, and acfA expression were shown to be activated in a LuxR-dependent manner, but only in the presence of both LuxR and 3-oxo-C6-HSL (Fig. 2). The levels of β-galactosidase expressed from these constructs were not equal; the luxI and qsrP constructs had the highest β-galactosidase levels, the ribB construct had the lowest β-galactosidase levels, and acfA expression was intermediate compared to the expression for the other genes in recombinant E. coli. This range suggests that there are differences in promoter strength and message stability between these four genes. The remaining promoters that were examined for LuxR-dependent expression did not drive β-galactosidase expression at levels above the background level, as shown for P_ainS (Fig. 2 and data not shown). Therefore, we concluded that in recombinant E. coli, ainS, qsrQ, qsrRST, and arcA are not upregulated by LuxR under the conditions tested here. To test if these promoters were repressed by LuxR, additional β-galactosidase assays were performed to compare expression (relative light units/OD₆₀₀) at an OD₆₀₀ of 0.25 and the expression at an OD₆₀₀ of 0.75; no repression was observed (data not shown). Therefore, no direct induction or repression of ainS, qsrQ, qsrRST, and arcA by LuxR was found in these studies or in the previous 2D SDS-PAGE analysis (2). These findings may not exclude the possibility that there is quorum-dependent regulation under circumstances where the environment within the native host, V. fischeri, plays an additional regulatory role. The genes that were definitively determined to be in the LuxR regulon, qsrP, ribB, and acfA, were used for further analysis.

Analysis of the sites of transcription initiation.

To define the promoter structure of qsrP, ribB, and acfA, the transcriptional start sites were identified. TraR, which is a LuxR homolog in Agrobacterium tumefaciens, can activate transcription from either class I or class II promoters (39). In a class I promoter, the activator binding site is located at approximately position −60 and the activator interacts with the C-terminal domain of the alpha subunit of RNA polymerase (13). In a class II promoter, the activator binding site is centered near position −40, overlapping the −35 recognition site for RNA polymerase and placing the activator in a position to potentially have multiple protein-protein interactions with the alpha and sigma subunits of RNA polymerase (1). In V. fischeri, the luxI promoter has a class II organization (10, 16). Based on the initial sequence analysis of the qsrP and acfA promoters, the lux box was predicted to overlap the −35 site (Fig. 1). Therefore, these two promoters were hypothesized to have a class II structure, similar to the luxI promoter. However, the ribB promoter contained two putative −10 sites, two putative −35 sites, and a lux box overlapping one of the −35 regions (Fig. 1A), indicating that it could potentially have either a class I-type structure or a class II-type structure.

Primer extension analysis was performed with RNA extracted from E. coli/pSC300, a strain expressing luxR under control of P_tac, and the vectors encoding qsrP, ribB, or acfA. RNA was also extracted from the V. fischeri luxI mutant strain MJ-211 (Table 1). Strains were grown either in the presence or in the absence of 3-oxo-C6-HSL to ascertain the dependence of the transcript on the quorum-sensing response. Since a higher copy number of the target transcript was present in E. coli, it was anticipated that more primer extension product would be generated than was generated in V. fischeri, and that is what was observed. The primer extension results for qsrP (Fig. 3) show both a major site and a minor site of transcription initiation under three of the four conditions tested; no transcripts were recovered from V. fischeri in the absence of 3-oxo-C6-HSL. The strongest transcripts were produced from either E. coli or V. fischeri strains in the presence of 3-oxo-C6-HSL. Based on these results, the position of the dominant transcriptional start site of qsrP was identified (Fig. 1A and 3), and the promoter was confirmed to have a class II-type structure.

For ribB and acfA, a series of bands were observed upon analysis of the primer extension products from E. coli. However, in V. fischeri, there was only one band for each gene, which corresponded to the dominant band from the E. coli samples. The strongest primer extension products were detected in samples from E. coli that had been exposed to 3-oxo-C6-HSL. Very weak ribB and acfA primer extension products were detected in V. fischeri cells that had been exposed to 3-oxo-C6-HSL. Therefore, for ribB and acfA, the amount of mRNA produced in V. fischeri appeared to be close to the threshold of detection for primer extension analysis. Nevertheless, the transcriptional start sites for ribB and acfA were identified (Fig. 1A and 3), and we concluded that both promoters have a class II-type architecture.

Timing of LuxR-dependent gene expression in V. fischeri.

In order to ascertain the expression pattern of the LuxR regulon in its native host, V. fischeri, a real-time RT-PCR analysis was carried out with RNA samples extracted from cells grown from the early exponential phase to the stationary phase. In P. aeruginosa, the timing of quorum-sensing gene expression is on a continuum, which means that some genes are induced early in growth and others are induced at the transition from the exponential phase to the stationary phase or during the stationary phase (29). In this study, we used two RNA sets, both purified under the same conditions from V. fischeri cells grown to four different OD₆₀₀ (0.25, 0.5, 1.0, and 2.0). The luminescence outputs of these cells were checked when the cells were harvested for RNA purification. At these four OD₆₀₀, the ratios of relative light units to OD were 27, 583, 643, and 829 (average data from the two trials). Thus, the quorum-sensing phenotype of bioluminescence was activated by mid-exponential-phase growth, as has been observed previously.

Using fragments of nadB and luxI as controls, the expression of the three putative LuxR-dependent genes was determined in two independent experiments using real-time RT-PCR. The nadB gene, which encodes l-aspartate oxidase in E. coli, was found to be expressed constitutively based on the work of other researchers (E. P. Greenberg, personal communication), and our initial analysis supported this finding. The starting quantity (SQ) of transcript was used to plot the RT-PCR data because it showed an elevated response due to quorum sensing, which is intuitively easier to understand. However, the threshold cycle, which is inversely proportional to the log SQ, also confirmed our findings. The SQ of luxI/nadB transcripts increased significantly at an OD₆₀₀ of 0.5, and expression was maintained at OD₆₀₀ of 1.0 and 2.0 (Fig. 4). This pattern of expression is what would be expected for a LuxR-dependent gene activated by the quorum-sensing response of the cells.

The SQ of qsrP/nadB, ribB/nadB, and acfA/nadB were much lower than that of luxI/nadB (note the exponential scale in Fig. 4). However, the former three transcripts were expressed in a pattern similar to that of luxI. The amount of the transcripts was significantly greater than the background amount at an OD₆₀₀ of 0.5, and the level of expression was roughly maintained at OD₆₀₀ of 1.0 and 2.0. Hence, similar to the results obtained with the β-galactosidase reporters in recombinant E. coli, the RT-PCR analysis demonstrated that qsrP, ribB, and acfA are temporally expressed in a pattern like that of luxI. However, the expression of these three genes was found to be significantly lower than that of luxI in the RT-PCR analysis. This was not predicted based on the β-galactosidase assays performed with the heterologous host E. coli. The apparent discrepancy may be resolved by considering that in the β-galactosidase assays, it is not only the rate of transcription but also the stability of the transcript, the rate of translation of lacZ mRNA, and the stability of LacZ that contribute to the final output measured. In the case of real-time RT-PCR, the pool of a particular mRNA of interest is measured and is dependent on the rate of transcription and the stability of the mRNA. To help further determine the relative expression levels of the LuxR regulon genes, EMSA were performed to ascertain whether differences in the DNA binding affinity of LuxR to the lux box of these genes were a key parameter in controlling expression levels.

Affinity of LuxR for the lux boxes of the genes in the LuxR regulon.

EMSA were used to determine the relative LuxR binding affinities for DNA fragments containing the lux boxes from the luxI, qsrP, acfA, and ribB promoters (Fig. 1B). In addition, the luxD box, a proposed lux box internal to luxD within the lux operon that has been hypothesized to be involved in negative regulation of the lux operon (30), was also used as a binding site for LuxR. The promoter of luxR itself served as a negative control. In the presence of 6 μM autoinducer, LuxR could bind to the promoter regions of four of the five lux box sequences examined (Fig. 5). A direct interaction between LuxR and the luxI, qsrP, acfA, and ribB promoters was definitively established. However, a direct association of LuxR with the luxD box could not be established under the conditions that were utilized, even when the ratio of protein to probe was increased 10- to 20-fold (data not shown). Therefore, in contrast to a previous proposal (30), this suggests that the luxD box may not play a major role in regulation of the lux genes. The dissociation constant (K_D) of LuxR for the DNA was roughly calculated by determining the protein concentration that shifted 50% of the probe. K_D values of 0.55 nM for P_luxI, 1.96 nM for P_qsrP, 0.60 nM for P_ribB, and 1.22 nM for P_acfA demonstrated that LuxR has higher affinities for P_luxI and P_ribB and lower affinities for P_qsrP and P_acfA. However, all of the K_D values were within fourfold of one another, suggesting that the weaker expression of qsrP, acfA, and ribB in V. fischeri measured via real-time RT-PCR was not due to dramatic differences in the affinity of LuxR for the promoters under study. Thus, the difference in expression is most likely due to the altered efficiencies in the interactions between LuxR and RNA polymerase, the rate of open complex formation, or the stability of the mRNA.

Conclusions.

Direct LuxR-dependent transcription of qsrP, acfA, and ribB from V. fischeri MJ-100 has been demonstrated. Activation of transcription of these genes occurs during mid-exponential growth, similar to temporal expression of the lux operon. However, real-time RT-PCR analysis indicated that the amounts of the qsrP, acfA, and ribB transcripts present in the cell's mRNA pool are significantly lower than the amount of luxI. EMSA demonstrated that LuxR has a slightly higher affinity for the lux boxes in the luxI and ribB promoters than for those in the qsrP and acfA promoters. LuxR binding to the luxD box, however, could not be demonstrated. The establishment here of a LuxR regulon in V. fischeri may provide insights into the optimal binding site for LuxR and lead to the identification of other LuxR-dependent genes.