Hierarchical Transcriptional Control of the LuxR Quorum-Sensing Regulon of Vibrio harveyi

Quorum sensing is important for survival of bacteria in nature and influences the actions of bacterial groups. In the relatively few studied examples of quorum-sensing-controlled genes, these genes are associated with competition or cooperation in complex microbial communities and/or virulence in a host. However, quorum sensing in vibrios controls the expression of hundreds of genes, and their functions are mostly unknown or uncharacterized. In this study, we identify the regulators of the second tier of gene expression in the quorum-sensing system of the aquaculture pathogen Vibrio harveyi. Our identification of regulatory networks and metabolic pathways controlled by quorum sensing can be extended and compared to other Vibrio species to understand the physiology, ecology, and pathogenesis of these organisms.

ABSTRACT

In vibrios, quorum sensing controls hundreds of genes that are required for cell density-specific behaviors including bioluminescence, biofilm formation, competence, secretion, and swarming motility. The central transcription factor in the quorum-sensing pathway is LuxR/HapR, which directly regulates ∼100 genes in the >400-gene regulon of Vibrio harveyi. Among these directly controlled genes are 15 transcription factors, which we predicted would comprise the second tier in the hierarchy of the LuxR regulon. We confirmed that LuxR binds to the promoters of these genes in vitro and quantified the extent of LuxR activation or repression of transcript levels. Transcriptome sequencing (RNA-seq) indicates that most of these transcriptional regulators control only a few genes, with the exception of MetJ, which is a global regulator. The genes regulated by these transcription factors are predicted to be involved in methionine and thiamine biosynthesis, membrane stability, RNA processing, c-di-GMP degradation, sugar transport, and other cellular processes. These data support a hierarchical model in which LuxR directly regulates 15 transcription factors that drive the second level of the gene expression cascade to influence cell density-dependent metabolic states and behaviors in V. harveyi.

IMPORTANCE Quorum sensing is important for survival of bacteria in nature and influences the actions of bacterial groups. In the relatively few studied examples of quorum-sensing-controlled genes, these genes are associated with competition or cooperation in complex microbial communities and/or virulence in a host. However, quorum sensing in vibrios controls the expression of hundreds of genes, and their functions are mostly unknown or uncharacterized. In this study, we identify the regulators of the second tier of gene expression in the quorum-sensing system of the aquaculture pathogen Vibrio harveyi. Our identification of regulatory networks and metabolic pathways controlled by quorum sensing can be extended and compared to other Vibrio species to understand the physiology, ecology, and pathogenesis of these organisms.

INTRODUCTION

Bacteria coordinate gene expression in response to an array of external stimuli including nutrients, pH, and temperature. Other environmental cues that control gene expression are autoinducers (AIs)—signaling molecules that are produced by bacterial communication systems termed quorum sensing (QS) (1). In nature, bacteria live in complex microbial communities where population-wide synchronization is critical for survival. While QS circuitry differs between organisms, a common function is to enable locally coordinated gene expression. The QS system of Vibrio harveyi, a significant pathogen of marine organisms, has been studied for decades and has yielded a wealth of knowledge about communication in members of the Vibrionaceae family (2–6). In this bacterium, three distinct AIs are produced intracellularly and diffuse into the local surroundings. These three AIs each bind cognate membrane-bound histidine sensor kinase receptors located in the inner membrane. The receptor proteins possess dual functionality as kinases and phosphatases, which controls the flow of phosphate through the quorum-sensing circuit (7). At low cell density (LCD), AI concentration is low and receptor proteins remain predominantly unbound by AIs. In the unbound state, these proteins function as kinases to phosphorylate LuxU, a phosphotransfer protein. In addition to the three membrane-bound AI receptors, a cytoplasmic histidine kinase senses nitric oxide and phosphorylates LuxU (8). Phosphorylated LuxU transfers the phosphate to the response regulator LuxO, which, in its phosphorylated state, activates transcription of the five quorum regulatory RNAs (Qrrs) (9). The Qrrs stabilize aphA mRNA and destabilize luxR mRNA (10). The end result is high-level production of AphA at LCD, the LCD master regulator, and simultaneous suppression of LuxR, the high cell density (HCD) master regulator. At LCD, cells configure gene expression programs via AphA and low levels of LuxR to behave as individuals (3). Conversely, as bacterial populations mature, AI concentrations exceed threshold concentrations in which cognate receptor proteins are highly bound by AIs. AI binding promotes the phosphatase activities of the receptor proteins (11). Under these conditions, LuxO is dephosphorylated, which ultimately results in cessation of AphA production and derepression of luxR. This derepression allows for a 10-fold increase in luxR expression, which increases LuxR concentration from ∼600 dimers per cell at LCD to ∼6,000 at HCD (12, 13). Once at quorum, LuxR is maximally expressed, and gene expression programs are adjusted to produce group behaviors. This circuitry is conserved across vibrios, and LuxR/HapR-type proteins are the core regulators of genes at HCD in all pathogenic vibrios, including the human pathogens Vibrio cholerae (HapR), Vibrio parahaemolyticus (OpaR), and Vibrio vulnificus (SmcR) (4).

Because LuxR is responsible for reconfiguring gene expression as cells transition from LCD to HCD, it is a critical but complex global regulator. Previous studies used microarray, transcriptome sequencing (RNA-seq), and chromatin immunoprecipitation sequencing (ChIP-seq) analyses to show that LuxR binds to 115 promoters to control the expression of >400 genes (13, 14). Genes within the LuxR regulon are involved in a number of processes including bioluminescence, virulence, secretion, and metabolism (13, 15). The roles of some LuxR-regulated genes have been extensively characterized; however, many genes encode proteins that do not have predicted functions and are annotated as “hypothetical” proteins. Here, we examined 16 LuxR-regulated genes that are predicted to be transcription factors. We determined the level of LuxR activation or repression of these genes and confirmed that LuxR directly binds to the promoters of 15 of these genes to directly control transcription regulation. Using RNA-seq, we describe the second tier of LuxR-regulated genes in Vibrio harveyi, which is comprised of 75 genes. These genes are predicted to be associated with many different cellular processes discussed below.

RESULTS AND DISCUSSION

LuxR regulates 16 transcription factors.

Previous studies have determined the genes in the LuxR regulon in V. harveyi using microarray and RNA-seq analyses under near-identical conditions (13, 15). These analyses revealed 625 and 424 genes, respectively, that show significant ≥2-fold regulation (activation or repression) in the presence of LuxR. Among these LuxR-regulated genes are 16 that encode proteins with predicted transcription factor functionality. Based on the microarray and RNA-seq experiments, LuxR represses expression of 13 of these transcription factor genes and activates expression of 3 genes. To verify these results, we performed quantitative real-time PCR (qRT-PCR) to compare RNA levels of these transcription factors between wild-type and ΔluxR strains. As expected, the qRT-PCR results corroborate the microarray and RNA-seq expression data, confirming that 13 genes are repressed and 3 genes are activated (Fig. 1). One gene, VIBHAR_00507, was not significantly different between wild-type and ΔluxR in our qRT-PCR assay. However, in both the previous microarray and RNA-seq data, VIBHAR_00507 expression was found to be significantly different between the two strains (P < 0.0001) (14, 15). Therefore, we conclude that VIBHAR_00507 is activated by LuxR. Importantly, the fold regulation by LuxR of each gene assayed by qRT-PCR closely mirrors the values determined using microarrays and RNA-seq (14, 15).

FIG 1.

LuxR regulates 16 genes encoding putative transcription factors. Relative transcript levels of 16 genes determined by qRT-PCR from RNA isolated from wild-type (BB120) or ΔluxR (KM669) V. harveyi strains. Gene names are indicated under the graph and correspond to the GenBank annotation VIBHAR_XXXXX. Asterisks indicate significant differences between wild-type and ΔluxR strains (*, P < 0.05; ***, P < 0.001; two-way analysis of variance (ANOVA) followed by Sidak’s multiple-comparison test on log-transformed data; n = 3).

LuxR directly binds to the promoter regions of 15 genes encoding transcription factors.

Previous assays have shown that LuxR binds directly to several of the promoters for the 16 genes encoding transcription factors (13, 14, 16). LuxR ChIP-seq peaks were observed in the upstream regions of each of the 16 genes except VIBHAR_00508, VIBHAR_02611, VIBHAR_06912, VIBHAR_02618, and the two type III secretion regulators, exsA and exsB (Fig. 2) (13, 17) However, electrophoretic mobility shift assays (EMSAs) showed that LuxR binds to the promoters of exsA and exsB in vitro (14, 18). It is not clear why LuxR ChIP data (from two studies) do not indicate LuxR binding at this locus, but in vitro assays show LuxR is capable of binding (13, 17). One possibility is that LuxR does bind to this locus in vivo but at earlier time points in the growth curve. The two ChIP-seq data sets were performed at early and late stationary phase. Thus, it is possible that LuxR binds to the promoters of exsA and exsB in log phase and is then outcompeted at stationary phase by another transcription factor. The lack of LuxR ChIP-seq peaks upstream of the VIBHAR_02611 and VIBHAR_00508 genes is not surprising because these genes are likely in operons with VIBHAR_02610 and VIBHAR_00507, respectively. The VIBHAR_02610-VIBHAR_02611 and VIBHAR_00507-VIBHAR_00508 genes in each pair are located in close proximity in the same orientation, and RNA-seq data suggest that these genes are cotranscribed (15). Further, LuxR ChIP-seq peaks exist upstream of VIBHAR_02610 and VIBHAR_00507 (Fig. 2), indicating that LuxR directly regulates these two operons.

FIG 2.

Genetic loci containing 16 putative transcription factor-encoding genes regulated by LuxR. Diagram showing the 16 loci containing genes encoding putative transcription factors (black arrows) and nearby genes regulated by LuxR >2-fold (gray arrows). Genes not regulated by LuxR >2-fold are shown as white arrows. Stars indicate locations of peaks in previously published ChIP-seq assays (13).

Using these previous data as a guide, we hypothesized that LuxR directly binds to the promoters of all of these 16 genes except VIBHAR_06912 and VIBHAR_02618, the only genes for which no direct binding by LuxR has been previously observed. To investigate whether LuxR modulates the expression of these transcription factors directly or indirectly, we performed EMSAs with PCR-amplified DNA corresponding to the 400 bp upstream of the open reading frame (ORF) for each gene. LuxR binding was observed at all of the upstream loci tested except the region upstream of VIBHAR_06912 (Fig. 3). We did observe binding in the promoter of VIBHAR_02618 in vitro and a minor peak in ChIP-seq, suggesting that LuxR has the capability of binding this promoter. Our results also confirmed that LuxR does not directly control VIBHAR_06912. The dynamics of LuxR binding to these loci are variable. For example, some regions possess multiple LuxR binding sites (e.g., VIBHAR_00507), which is apparent from the supershifted band. This is not uncommon, as there is an average of two LuxR binding sites per promoter in the LuxR regulon (13). Additionally, the relative affinity of LuxR binding sites within these promoters is variable. Some loci (e.g., VIBHAR_06181) are completely bound with as little as 10 nM LuxR while others required 100 nM LuxR protein to be shifted (e.g., VIBHAR_01697) (Fig. 3). Again, this is reminiscent of binding sites in other LuxR-regulated promoters (e.g., P_luxCDABE) that show a range of LuxR binding affinities (13). Together, these results indicate that LuxR directly regulates 15 of the 16 genes encoding putative transcription factors.

FIG 3.

LuxR directly binds to the regions upstream of 14 genes encoding putative transcription factors. EMSAs containing PCR-generated, radiolabeled promoter DNA (400 bp) and purified LuxR protein at 0, 10, and 100 nM concentrations (except for P₀₂₆₁₇ and P₀₂₆₁₈ substrates in which 0, 20, and 100 nM were used). Insets show LuxR ChIP-seq peak profiles for each locus using data from a previous study (17).

The second tier of the LuxR regulon contains >300 genes.

Because ChIP-seq data indicate that LuxR directly regulates only ∼115 genes out of >400 genes in its regulon, we predicted that these 15 transcription factors control the remaining genes in the LuxR regulon. To examine this, we performed RNA-seq analyses comparing strains with deletions of a transcription factor gene to the isogenic parent strain (ΔluxR) to determine the regulons of each. We analyzed the regulon of the following genes predicted to encode transcription factors: VIBHAR_00050 (metJ), VIBHAR_02610, VIBHAR_02611, VIBHAR_02617, VIBHAR_02618, VIBHAR_05185, VIBHAR_06912, and VIBHAR_06936. We did not investigate the other transcription factor genes because they have been previously characterized as follows: (i) VIBHAR_06181 encodes BetI, which regulates the betIBA-proXWV operon in V. harveyi (6 genes) (16); (ii) VIBHAR_01697 and VIBHAR_01694 encode ExsA and ExsB, respectively, which regulate the four type III secretion operons in V. harveyi (36 genes) (18); and (iii) VIBHAR_00046 encodes AphA, which controls the LCD regulon in V. harveyi (167 genes) (3, 14). The remaining genes, VIBHAR_00507, VIBHAR_00508, VIBHAR_03493, and VIBHAR_05843, are regulated <3-fold by LuxR; thus, we chose not to pursue them. Because each of the genes that we chose to examine is repressed by LuxR, we used the ΔluxR background for all RNA-seq experiments to determine the regulon of each transcription factor.

The RNA-seq results revealed >75 genes that are regulated by these LuxR-controlled transcription factors. Compounding these results with data from previously published work yields the combined second tier of the LuxR regulon, which contains >300 genes. In the following sections, the constituents of each regulon are discussed in the context of putative biological function.

MetJ regulates 49 genes and controls bioluminescence in V. harveyi.

The VIBHAR_00050 gene encodes the transcriptional regulator MetJ, which shares 80.7% amino acid identity with MetJ from Escherichia coli. Bacterial methionine biosynthesis has been extensively studied, and the role of MetJ is well understood (19). In E. coli, MetJ is a repressor that functionally opposes a transcriptional activator, MetR, to repress many of the methionine biosynthetic enzymes, including metL, metA, metB, metC, metH, and others (20–22). Although LuxR only represses the expression of metJ by 1.5-fold under the single condition we tested, we considered this regulation to be important because MetR regulates one of the most highly activated quorum-sensing phenotypes, bioluminescence, in V. harveyi (23). Among all of the transcription factors we investigated, deletion of metJ affected the largest number of genes; our RNA-seq experiment revealed that metJ controls the expression of 49 genes (±2-fold, P adjusted ≤ 0.05) (Table 1). MetJ activates the expression of 9 genes while repressing the remaining 40. The degree of regulation exhibited by MetJ is heavily shifted toward repression; the most highly activated gene in its regulon is upregulated only ∼4-fold, whereas the met genes (the most highly repressed genes) are downregulated ∼15- to 75-fold.

TABLE 1.

Genes regulated by VIBHAR_00050 (MetJ)

Locus ID	Old locus ID	Fold regulation	P-adjusted value	Predicted function
VIBHAR_RS25170	VIBHAR_06544	4.0	7.5E−07	Hypothetical protein
VIBHAR_RS23080	VIBHAR_06041	3.8	4.6E−02	Hypothetical protein
pspA	VIBHAR_01872	3.1	1.5E−33	Phage shock protein PspA
pspB	VIBHAR_01873	2.9	3.8E−21	Envelope stress response membrane protein PspB
pspC	VIBHAR_01874	2.4	3.2E−28	PspC domain-containing protein
VIBHAR_RS00035	VIBHAR_00008	2.4	9.3E−04	Envelope stress response protein PspG
VIBHAR_RS12360	VIBHAR_02639	2.1	6.4E−10	TetR/AcrR family transcriptional regulator
VIBHAR_RS13045	VIBHAR_02785	2.0	1.3E−07	Hypothetical protein
VIBHAR_RS18605	VIBHAR_05000	2.0	4.0E−04	Glycine C-acetyltransferase
VIBHAR_RS25745	VIBHAR_06685	−2.0	2.8E−03	Hypothetical protein
VIBHAR_RS20725	VIBHAR_05495	−2.1	3.1E−05	Protoheme IX farnesyltransferase 1
VIBHAR_RS07970	VIBHAR_01701	−2.1	2.6E−10	EscD/YscD/HrpQ family type III secretion system inner membrane ring protein
VIBHAR_RS07990	VIBHAR_01705	−2.1	3.8E−10	Type III export protein
VIBHAR_RS06990	VIBHAR_01482	−2.1	3.7E−02	Hypothetical protein
VIBHAR_RS10100	VIBHAR_02153	−2.1	1.9E−04	Hypothetical protein
VIBHAR_RS23730	VIBHAR_06196	−2.2	8.2E−03	Pilus assembly protein CpaC
VIBHAR_RS10700	VIBHAR_02279	−2.2	1.0E−03	Formate dehydrogenase
VIBHAR_RS19920	VIBHAR_05304	−2.2	1.5E−02	1,4-Alpha-glucan-branching protein
VIBHAR_RS21930	VIBHAR_05775	−2.2	2.4E−02	Hypothetical protein
VIBHAR_RS23735	VIBHAR_06197	−2.3	4.0E−02	Pilus assembly protein CpaB
cyoC	VIBHAR_05493	−2.3	1.2E−07	Cytochrome o ubiquinol oxidase subunit III
VIBHAR_RS21180	VIBHAR_05607	−2.4	1.3E−03	Hypothetical protein
VIBHAR_RS10105	VIBHAR_02154	−2.4	1.0E−05	Hypothetical protein
VIBHAR_RS01205	VIBHAR_00256	−2.4	2.3E−10	Glutamate racemase
VIBHAR_RS23745	VIBHAR_06199	−2.5	5.2E−03	Flp family type IVb pilin
VIBHAR_RS22625	VIBHAR_05936	−2.5	4.2E−02	Methyl-accepting chemotaxis protein
VIBHAR_RS10705	VIBHAR_02280	−2.5	1.1E−05	Formate dehydrogenase subunit gamma
VIBHAR_RS16695	VIBHAR_RS16695	−2.7	4.6E−05	Hypothetical protein
VIBHAR_RS18715	VIBHAR_05026	−3.3	4.6E−05	Phage capsid protein
VIBHAR_RS23740	VIBHAR_06198	−3.5	7.1E−03	Prepilin peptidase CpaA/TadV
VIBHAR_RS12030	VIBHAR_02564	−3.5	6.2E−06	Sodium transporter
VIBHAR_RS01210	VIBHAR_00257	−4.9	1.9E−14	Hypothetical protein
VIBHAR_RS17385	VIBHAR_03712	−6.2	6.4E−238	Methionine synthase
VIBHAR_RS01215	VIBHAR_00260	−6.3	7.1E−06	Hypothetical protein
VIBHAR_RS22555	VIBHAR_05920	−6.9	2.3E−04	DUF1097 domain-containing protein
VIBHAR_RS05235	VIBHAR_01104	−8.4	3.3E−23	Sodium/proton antiporter
VIBHAR_RS01220	VIBHAR_00261	−10.2	5.8E−195	Vitamin B12 transporter BtuB
VIBHAR_RS12990	VIBHAR_02773	−15.6	3.1E−23	HTH-type transcriptional regulator MetR
metK	VIBHAR_03568	−17.1	0.0E+00	S-Adenosylmethionine synthase
metN	VIBHAR_01202	−19.0	2.3E−233	d-Methionine ABC transporter, ATP-binding protein
VIBHAR_RS11700	VIBHAR_02491	−26.8	0.0E+00	Homoserine o-succinyltransferase
metQ	VIBHAR_01200	−28.9	0.0E+00	Methionine ABC transporter substrate-binding protein MetQ
VIBHAR_RS05680	VIBHAR_01201	−31.1	5.2E−151	ABC transporter permease
VIBHAR_RS13115	VIBHAR_02800	−44.2	2.2E−20	5-Methyltetrahydropteroyltriglutamate-homocysteine methyltransferase
metF	VIBHAR_00047	−45.2	0.0E+00	Methylenetetrahydrofolate reductase
metL	VIBHAR_00048	−49.6	0.0E+00	Bifunctional aspartate kinase/homoserine dehydrogenase II
metB	VIBHAR_00049	−56.2	0.0E+00	Cystathionine gamma-synthase
VIBHAR_RS13120	VIBHAR_02801	−64.7	2.8E−30	DUF1852 domain-containing protein
VIBHAR_RS12985	VIBHAR_02772	−76.8	1.2E−122	5-Methyltetrahydropteroyltriglutamate-homocysteine methyltransferase (MetE)

MetR is a direct repressor of bioluminescence in V. harveyi, and a deletion of metR increases light ∼10-fold (23). Consistent with data from E. coli, MetJ in V. harveyi represses the expression of MetR (Table 1) (24). To investigate the effects of this regulatory loop, we assayed light production in the ΔmetJ strain. The wild-type strain BB120 was used as a reference and shows increased bioluminescence as a function of cell density. Importantly, light production requires LuxR, and we note that the ΔluxR strains do not bioluminesce. Compared to the wild-type strain, the ΔmetJ strain exhibits an ∼2-fold reduction in bioluminescence at HCD in Luria murine (LM) medium, which is standard lysogeny broth (LB) with 2% NaCl (Fig. 4A). Interestingly, we observed the reciprocal trend when this experiment was performed using an M9-based minimal medium. Under these conditions, the deletion of metJ confers an approximate 2- to 3-fold increase in bioluminescence (Fig. 4A). The concentration of l-methionine in LB is ∼2 to 6 mM; the disparity in l-methionine levels in our media (2 to 6 mM in LM and 0 mM in minimal) could explain the differential bioluminescent phenotypes (25). To verify that MetJ is responsible for this difference, we attempted to complement the ΔmetJ strain using an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible expression vector in LM medium. To our surprise, when MetJ is expressed in trans in the ΔmetJ strain, bioluminescence is not restored to wild-type levels (Fig. 4B). This could indicate the presence of a secondary mutation in the ΔmetJ strain that is affecting bioluminescence. However, we observed that complementation of metJ restored transcript levels of other MetJ-regulated genes (three distinct operons) to wild-type levels as follows: metL (VIBHAR_00048), metK (VIBHAR_03568), and metQ (VIBHAR_01200) (Fig. 4C), indicating that a potential secondary mutation does not affect all of the genes in the MetJ regulon. The secondary mutation could be due to spontaneous mutation during propagation of the strain or a polar effect introduced via deletion of the metJ gene. The metJ gene is not in an operon and rather is flanked by genes in opposing directions (Fig. 2); thus, we cannot rule out effects from the deletion on the transcriptional initiation and/or termination of the adjacent gene(s) in the ΔmetJ mutant strain. Thus, it is not clear how the ΔmetJ allele is affecting bioluminescence expression, but the result is consistent through multiple biological experiments.

FIG 4.

MetJ regulates bioluminescence and methionine synthesis. (A) Bioluminescence (Lux/OD₆₀₀) produced by wild-type (BB120), ΔluxR (KM669), ΔmetJ (RRC176), and ΔluxR ΔmetJ (RRC178) strains grown in LM (left) or M9 (right) medium. Data shown are representative of three independent experiments. (B) Bioluminescence produced by strains as follows: wild-type with empty vector (CAS127), ΔmetJ with empty vector (AB125), and ΔmetJ with pmetJ (AB121). The amount of IPTG used for induction of metJ is indicated. (C) Transcript levels of metL, metK, and metQ (VIBHAR_00048, VIBHAR_03568, VIBHAR_01200, respectively) determined by qRT-PCR of RNA isolated from strains as follows: ΔluxR with empty vector (AB087), ΔmetJ with empty vector (AB125), and ΔmetJ with pmetJ (AB121). A total of 50 μM IPTG was added to all three strains. One-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test was performed (n = 3; P < 0.001).

In addition to the methionine biosynthesis genes, MetJ regulates the expression of several other metabolic enzymes, suggesting that this transcription factor fine-tunes other sectors of cellular metabolism. Although MetJ activates the expression of only 9 genes, some of these are predicted to be involved in interesting behaviors. For example, VIBHAR_02639 encodes a TetR/AcrR family transcriptional regulator, and structural homology analysis identified the closest homologue of VIBHAR_02639 as PsrA from Pseudomonas aeruginosa (39% amino acid identity, 98% coverage). PsrA controls the expression of type III secretion, RpoS, swarming motility, and biofilm formation in P. aeruginosa (26–28). PsrA responds to long-chain fatty acids as a means to control expression of fatty acid degradation genes (fad operon) (29, 30). In V. harveyi, this PsrA-like gene is located directly upstream of a FadE-like gene, which is an acyl coenzyme A (acyl-CoA) dehydrogenase (VIBHAR_02638). FadE from E. coli performs the initial steps of the β-oxidation cycle of fatty acid catabolism. Indeed, VIBHAR_02638 shares significant homology to FadE (50% amino acid identity, 90% coverage). Furthermore, a gene encoding a FadV homologue is situated directly downstream of the FadE homologue (VIBHAR_02637). FadV enzymes are also involved in fatty acid anabolism by catalyzing the final step of fatty acid elongation. Thus, based on genetic organization and protein homology, the VIBHAR_02637 to VIBHAR_02639 locus likely encodes genes involved in fatty acid metabolism. The net effect of LuxR regulation of this operon is negative; LuxR represses MetJ, which activates VIBHAR_02639, ultimately yielding a decrease in the expression of this gene at HCD (observed in reference 15). Thus, LuxR likely plays a role in the fine-tuning of lipid metabolism as nutrient availability changes as cells enter stationary phase.

VIBHAR_02610 and VIBHAR_02611 have distinct regulons.

Genes VIBHAR_02610 and VIBHAR_02611 were first annotated by GenBank as two separate ORFs, with VIBHAR_02611 using a TTG start codon and both having homology to the LysR-type d-serine deaminase transcriptional activator DsdC (Fig. 2). In agreement with this predicted function is the observation that directly upstream is a gene that encodes a homologue of DsdA (d-serine dehydratase, VIBHAR_02609); this mirrors the genetic organization of dsdC and dsdA in various E. coli strains (31). However, the most recent GenBank annotation of the V. harveyi BB120 genome predicts that this region encodes a pseudogene and instead denotes a frameshift. Because GeneMarkS (32) also predicts two open reading frames, we analyzed this region as containing two ORFs (VIBHAR_02610 and VIBHAR_02611) and constructed deletion mutants of each gene individually.

Our RNA-seq experiments revealed that VIBHAR_02610 controls the expression of 13 genes (±2-fold, P adjusted ≤ 0.05) (Table 2). Among these 13 genes, 4 are repressed by VIBHAR_02610 while 9 are activated. VIBHAR_02611 regulates the expression of only 2 genes in V. harveyi (±2-fold, P adjusted ≤ 0.05) (Table 3). Both of these genes are repressed ∼3-fold by VIBHAR_02611.

TABLE 2.

Genes regulated by VIBHAR_02610

Locus ID	Old locus ID	Fold regulation	P-adjusted value	Predicted function
VIBHAR_RS25170	VIBHAR_06544	3.0	1.6E−04	Hypothetical protein
VIBHAR_RS12355	VIBHAR_02638	2.9	3.5E−08	Acyl-CoA dehydrogenase
VIBHAR_RS00035	VIBHAR_00008	2.7	3.0E−04	Envelope stress response protein PspG
VIBHAR_RS12360	VIBHAR_02639	2.5	8.9E−15	TetR/AcrR family transcriptional regulator
pspA	VIBHAR_01872	2.3	1.6E−16	Phage shock protein PspA
VIBHAR_RS03960	VIBHAR_00836	2.2	7.0E−08	Hypothetical protein
pspB	VIBHAR_01873	2.2	1.1E−10	Envelope stress response membrane protein PspB
VIBHAR_RS13960	VIBHAR_02991	2.2	4.7E−02	RNase T
pspC	VIBHAR_01874	2.1	1.7E−18	PspC domain-containing protein
acs	VIBHAR_00169	−3.2	1.7E−03	Acetate-CoA ligase
VIBHAR_RS00780	VIBHAR_00160	−3.6	4.0E−02	Cation acetate symporter
VIBHAR_RS27820	VIBHAR_p08218	−7.2	1.1E−03	TraY domain-containing protein
VIBHAR_RS00775	VIBHAR_00159	−9.7	3.0E−02	DUF4212 domain-containing protein

TABLE 3.

Genes regulated by VIBHAR_02611

Locus ID	Old locus ID	Fold regulation	P-adjusted value	Predicted function
VIBHAR_RS07715	VIBHAR_01640	−2.8	1.41E−02	Hypothetical protein
VIBHAR_RS24415	VIBHAR_06361	−2.9	9.10E−03	Endonuclease/exonuclease/ phosphatase family protein

Although VIBHAR_02610/VIBHAR_02611 display homology to DsdC, which controls d-serine transport in E. coli, they appear to regulate significantly different processes in V. harveyi. The 4 genes repressed by VIBHAR_02610 appear to be involved in acetate transport/processing, conjugal DNA transfer, and metal transport. VIBHAR_02610 represses the expression of an acetate symporter (VIBHAR_00160) and an acetate-CoA ligase (VIBHAR_00169). Presumably, these proteins import acetate and convert it to acetate-CoA (acetyl-CoA), which can be fed directly into the tricarboxylic acid (TCA) cycle. Because these genes are repressed by VIBHAR_02610, which is repressed by LuxR, the expression of these genes is likely elevated at HCD. The increased expression of genes involved in acetate metabolism aligns well with previous findings in Vibrio fischeri that show that QS activates the expression of acetyl-CoA synthase (via the LuxR homologue, LitR) to reduce extracellular acetate that would otherwise accumulate to toxic levels (33). Not surprisingly, VIBHAR_00169 encodes the V. harveyi homologue of this acetyl-CoA synthase (84% amino acid identity, 100% coverage). These observations provide evidence that V. harveyi uses QS to evade acetate toxification comparable to V. fischeri.

The gene most highly repressed by VIBHAR_02610 is VIBHAR_00159, which is annotated as encoding a hypothetical protein containing a DUF4212 domain. Secondary structure prediction tools indicate homology to E. coli protein ZntB, a membrane-embedded protein that forms a membrane channel to transport zinc into the cytoplasm (34). Zinc uptake is critically important for pathogenic bacteria because many metalloproteases require zinc to function (35).

Genes activated by VIBHAR_02610 are associated with membrane stress response, transcription, and RNA processing. The membrane stress response genes correspond to the phage shock proteins pspABCG—these proteins respond to cell envelope stresses through a variety of mechanisms (36, 37). Interestingly, the psp genes and the PsrA-like transcription factor (VIBHAR_02639) are all activated by both VIBHAR_02610 and MetJ. It is possible that VIBHAR_02610 and MetJ function together to coordinate certain gene expression programs.

VIBHAR_02617 and VIBHAR_02618 have small, nearly identical regulons.

VIBHAR_02617 is classified as an xenobiotic response element (XRE)-type transcriptional regulator. These transcriptional regulators are proteins that contain a helix-turn-helix DNA binding motif and are homologous to the CI and Cro proteins from bacteriophage λ. VIBHAR_02618 does not have an obvious classification and instead is annotated generally as a transcriptional regulator. RNA-seq data indicate that the regulons for each of these regulators are very small; VIBHAR_02617 regulates 3 genes while VIBHAR_02618 regulates 2 genes (±2-fold, P adjusted ≤ 0.05) (Table 4 and Table 5). Second, both genes regulated by VIBHAR_02618 are also activated by VIBHAR_02617. The two coregulated genes (VIBHAR_07127 and VIBHAR_05996) encode proteins that are >94% identical in amino acid sequence. Unfortunately, these genes encode proteins with no clear homologues discernible via BLAST/Phyre2; their functions will need to be determined experimentally. We investigated the putative promoter regions upstream of both of these genes and found them to be nearly identical (99% conservation up to bp −300 from the start codon). Because of the nearly identical conservation of regulatory targets, we investigated the homology between VIBHAR_02617 and VIBHAR_02618. Alignment of the amino acid sequences of VIBHAR_02617 and VIBHAR_02618 revealed little, if any, homology to each other (26% identity, 96% coverage).

TABLE 4.

Genes regulated by VIBHAR_02617

Locus ID	Old locus ID	Fold regulation	P-adjusted value	Predicted function
VIBHAR_RS27540	VIBHAR_07127	2.4	4.6E−15	Hypothetical protein
VIBHAR_RS22885	VIBHAR_05996	2.3	3.4E−07	Hypothetical protein
VIBHAR_RS01705	VIBHAR_00366	2.1	3.9E−03	Phosphomethylpyrimidine synthase (ThiC)

TABLE 5.

Genes regulated by VIBHAR_02618

Locus ID	Old locus ID	Fold regulation	P-adjusted value	Predicted function
VIBHAR_RS27540	VIBHAR_07127	3.5	1.2E−27	Hypothetical protein
VIBHAR_RS22885	VIBHAR_05996	3.3	1.3E−13	Hypothetical protein

The gene that is regulated by VIBHAR_02617 but not VIBHAR_02618 is VIBHAR_00366, which is annotated as encoding a phosphomethylpyrimidine synthase. Several BLAST searches revealed this protein as ThiC, a member of the thiamine biosynthetic pathway in bacteria that catalyzes the conversion of 5-aminoimidazole ribotide (AIR) to 5-hydroxymethyl-2-methylpyrimidine pyrophosphate (HMP-PP) (38–41). Details regarding thiamine biosynthesis are explored in the following section because VIBHAR_05185 regulates several thi genes. VIBHAR_02617 activates the expression of ThiC; thus, LuxR represses this arm of thiamine biosynthesis via repression of VIBHAR_02617.

VIBHAR_05185 regulates five genes.

The protein encoded by VIBHAR_05185 is annotated as an AraC-type transcriptional regulator. AraC homologues have been shown to activate the expression of the arabinose utilization operons, araBAD, araFGH, and araE, upon binding arabinose (42). Our RNA-seq data suggest that VIBHAR_05185 may also serve as a metabolic regulator, similar to AraC, however in the context of thiamine instead of arabinose. VIBHAR_05185 regulates five genes in V. harveyi (±2-fold, P adjusted ≤ 0.05) (Table 6) that are associated with thiamine biosynthesis, fructose transport, and c-di-GMP degradation. The genes involved in thiamine biosynthesis, the thiCEFSGH operon, are repressed by VIBHAR_05185, although thiC, thiS, and thiG fall slightly below our 2-fold cutoff. The literature regarding the regulation of the thi genes is largely focused on the posttranscriptional level in which the THI-box riboswitch serves as a negative feedback control point (43). A transcriptional regulator that modulates the expression of the thi genes has yet to be identified in bacteria. It is possible that VIBHAR_05185 directly represses expression of the thi operon in V. harveyi. Conversely, it is possible that secondary effects produced by VIBHAR_05185 (i.e., regulation of other genes/cellular processes) influence expression of this operon. The net effect is that these thiamine biosynthesis genes are upregulated as the population reaches quorum. Thiamine and its derivatives function as cofactors for many cellular enzymes (44, 45). Thus, it is probable that QS fine-tunes the production of thiamine to accommodate the enzymatic activity of populations transitioning between cellular densities.

TABLE 6.

Genes regulated by VIBHAR_05185

Locus ID	Old locus ID	Fold regulation	P-adjusted value	Predicted function
VIBHAR_RS19410	VIBHAR_05188	3.0	6.8E−47	PTS fructose transporter subunit IIA
VIBHAR_RS08935	VIBHAR_01904	2.2	4.8E−10	EAL domain-containing protein (RbdA from Pseudomonas)
thiH	VIBHAR_00361	−2.0	3.4E−02	2-Iminoacetate synthase ThiH
VIBHAR_RS01700	VIBHAR_00365	−2.1	2.2E−02	Thiamine phosphate synthase (ThiE)
VIBHAR_RS01695	VIBHAR_00364	−3.0	7.2E−04	Molybdopterin-synthase adenylyltransferase MoeB (ThiF)

In addition to the thi operon, VIBHAR_05185 activates the expression of a fructose transporter, encoded just downstream of VIBHAR_05185 (Fig. 2). Because the expression of the fructose transporter gene is activated by VIBHAR_05185, which is repressed by LuxR, QS effectively reduces the expression of this gene at HCD. Consistent with this hypothesis, LuxR represses other components of the fructose phosphotransferase system (PTS) that neighbor VIBHAR_05185 at this locus (15). Interestingly, the LM medium used in this experiment contains <100 μM sugars (25). If this experiment is performed in other medium with fructose, gene expression changes might be more substantial throughout quorum-sensing stages. We surmise that as populations of V. harveyi mature and reach HCD states, the availability of fructose likely decreases substantially. Thus, QS serves to shut down elements of fructose uptake as the carbon source is depleted.

Another interesting constituent of the VIBHAR_05185 regulon is VIBHAR_01904. This gene encodes an EAL domain-containing protein, and further analysis using Phyre2 predicts significant secondary structure homology to RbdA from P. aeruginosa. RbdA is a PAS-GGDEF-EAL protein that negatively regulates biofilm formation while simultaneously stimulating dispersal (46, 47). Although RbdA contains both diguanylate cyclase and phosphodiesterase motifs, it is primarily responsible for degrading c-di-GMP in vivo. In addition to modulating intracellular concentrations of c-di-GMP, the PAS domain of RbdA serves as an O₂ sensor (46). It has been postulated that RbdA is a component of low-oxygen stress response in Pseudomonas. When comparing the homology between RbdA and VIBHAR_01904, residues are mostly conserved in the PAS and EAL domains. The EAL domain is more strongly conserved, so it is likely that VIBHAR_01904 also functions as a phosphodiesterase in vivo. Conversely, the PAS domain shows weaker conservation—this could be explained by a difference in ligand identity. Investigation of this protein could provide insight on c-di-GMP regulation and its connection to QS in V. harveyi.

VIBHAR_06912 does not regulate gene expression under laboratory conditions.

The protein encoded by VIBHAR_06912 belongs to the Lrp/AsnC family of transcription factors. These proteins are equipped with a DNA binding domain to interact with promoter DNA as well as a ligand binding domain to sense a small molecule/metabolite (e.g., leucine in the case of Lrp) (48). We were surprised to find that VIBHAR_06912 has no significant effect on gene expression in V. harveyi (±2-fold, P adjusted value ≤ 0.05). Thus, we conclude that, under our laboratory conditions, VIBHAR_06912 is not involved in regulating downstream gene expression. This transcription factor is highly conserved in other vibrios including V. parahaemolyticus as well as other marine bacteria such as Ferrimonas, Shewanella, and Bacterioplanes (see Fig. 6). This observation supports the idea that VIBHAR_06912 may be utilized under specific conditions and that these conditions are experienced by Vibrio, Ferrimonas, Shewanella, and Bacterioplanes alike.

FIG 6.

LuxR-regulated transcription factors are conserved across the Vibrio genus. The amino acid sequence of each transcription factor from V. campbellii (ATCC BAA-1116) was used as a template for BLAST analysis (NCBI). The bit-score of each hit is displayed; the maximum bit-score value is represented by the V. campbellii hits (highlighted row). The length of the green bar in each cell represents the degree of conservation of each protein. Phylogenetic analysis was performed by aligning 16S rRNA from each species using Clustal Omega and iTOL (v5).

VIBHAR_06936 controls the expression of 1 gene.

The protein encoded by VIBHAR_06936 belongs to the LacI/MalR transcription factor family. In V. harveyi, this transcription factor regulates 1 gene (±2-fold, P adjusted ≤ 0.05) (VIBHAR_RS01705 is regulated by VIBHAR_06936; old locus identifier, VIBHAR_00366; fold regulation, 2.2; P-adjusted value, 3.2E−03; predicted function, phosphomethylpyrimidine synthase [ThiC]). Interestingly, this gene is thiC, which is also regulated >2-fold by VIBHAR_02617. Similar to VIBHAR_05185, thiE and thiF are also regulated by VIBHAR_06936; however, they fall slightly below our 2-fold change in gene expression cutoff. This result indicates that three of these regulators control thiamine biosynthesis as follows: VIBHAR_02617 and VIBHAR_06936 activate thiC whereas VIBHAR_05185 represses expression of thiEFH. This opposing regulation of these genes is intriguing, and future experiments will be needed to examine thiamine connections to QS.

Second-tier regulators do not contribute to efficient cellular growth.

We next wanted to determine the influence of each of the second-tier regulators on overall cellular growth. We monitored the growth of each mutant strain and compared them to wild-type and ΔluxR strains. In LM (rich) medium, deletion of any second-tier regulator does not influence growth rate or culture yield (Fig. 5A; see also Fig. S1A in the supplemental material). Nearly all of the amino acids are present at high levels (low mM concentrations) in LM (25). This nutritional density could mask growth phenotypes resulting from the deletion of any of the LuxR-regulated transcription factors. To further explore this hypothesis, we performed this experiment under more restricted nutritional conditions using an M9-based minimal medium. With the exception of the ΔluxR ΔmetJ strain, none of the other strains tested show an appreciable growth defect (Fig. 5A; see also Fig. S1B). The ΔluxR ΔmetJ strain displays a severe growth defect in this medium. Interestingly, while there is a small improvement in growth kinetics, the addition of exogenous l-methionine does not completely rescue this mutant (Fig. 5B). This result can be rationalized in the following two nonmutually exclusive ways: (i) LuxR and MetJ are both required for l-methionine transport, or (ii) LuxR and MetJ are involved in essential metabolic pathways outside of methionine biogenesis. In support of the latter rationale, we observed experiment-to-experiment variation in the ΔluxR mutant when grown in minimal medium as the culture reached stationary phase (Fig. 5A). Wild-type V. harveyi displays robust growth in minimal medium (Fig. 5A), which suggests that LuxR and MetJ are required to achieve the highest growth rate and yield under these conditions. Overall, these results indicate that MetJ plays a vital role in central metabolism while the other second-tier regulators likely function in response to external stimuli not present in our laboratory conditions.

FIG 5.

LuxR and MetJ are required for growth in minimal medium. (A) Growth (OD₆₀₀) over time by wild-type (BB120), ΔluxR (KM669), ΔmetJ (RRC176), and ΔluxR ΔmetJ (RRC178) strains in LM (left) or M9 (right) medium was measured. (B) Growth of the ΔluxR ΔmetJ strain in minimal M9 medium supplemented with exogenous l-methionine at the concentrations indicated.

Conservation of second-tier regulators is variable among vibrios.

We next explored the degree of conservation of the second-tier regulators in other vibrios. BLAST analysis was used for the set of 8 transcription factors that we examined by RNA-seq to identify homologues in other common Vibrio species. Additionally, we performed a phylogenetic analysis of the identified Vibrio species using 16S rRNA aligned with Clustal Omega to correlate the presence of these transcription factors with genetic relatedness. The BLAST results indicate that MetJ, VIBHAR_02610, VIBHAR_02611, VIBHAR_05185, and VIBHAR_06936 are the most conserved and widespread across the Vibrio clade (Fig. 6). Conversely, VIBHAR_02617, VIBHAR_02618, and VIBHAR_06912 show conservation in species most closely related to Vibrio harveyi/Vibrio campbellii.

Conclusions.

Our work has shown that 15 transcription factors comprise the second tier of regulation in the LuxR quorum-sensing regulon (Fig. 7). Among these, several were previously known to control important cellular behaviors, such as type III secretion and osmotic stress response. The experiments in this study have expanded our knowledge of the LuxR-regulated genes in the quorum-sensing pathway of V. harveyi to include other pathways and metabolic systems, including methionine biosynthesis, thiamine biosynthesis, and acetate utilization. The second tier of regulation controls at least 75 genes, though we did not examine the regulon of every transcription factor (Fig. 7).

FIG 7.

The hierarchy of LuxR regulation in V. harveyi. All LuxR-regulated transcription factors and their associated regulons are shown. Transcription factors are colored based on their respective regulons; gray indicates that the regulon was determined previously (3, 14–16), red indicates that the regulon was not tested in this study, yellow indicates that the transcription factor did not regulate genes under our tested conditions, and blue indicates a regulon that was determined in this study. Arrows indicate positive regulation while blunt-end arrows indicate repression. Lines emanating from LuxR are solid to denote direct regulation or dotted to denote indirect regulation. Direct/indirect regulation of the second-tier regulons is not inferred in the model.

Some transcription factors regulated by LuxR-type proteins have important connections to the quorum-sensing circuit in vibrios, including AphA in V. harveyi and VqsA in Vibrio alginolyticus. These regulators both feed back on the quorum-sensing system to regulate other components as follows: (i) AphA represses the Qrr small RNAs (sRNAs) and luxR and the Qrrs activate aphA, and (ii) VqsA activates luxR and represses aphA (3, 49). Of note, V. harveyi does not appear to encode a VqsA homologue; the closest conserved gene product shares only 36% amino acid identity, and the gene is not regulated by LuxR (14, 15). Among the 8 transcription factors that we investigated, we did not observe evidence of feedback regulation onto any of the other quorum-sensing components (e.g., aphA or the Qrrs). The experimental setup precluded us from evaluating regulation of luxR because the RNA-seq was performed in the ΔluxR background. However, other studies of Qrr targets did not reveal any of these transcription factor genes as regulated by the Qrrs (50). Thus, it is possible that there is not any regulatory connection of these transcription factors to the quorum-sensing circuit beyond LuxR-directed regulation with the exception of AphA.

Instead of feeding back onto the QS system, our data demonstrates that these LuxR-regulated transcription factors are involved in modulating several cellular processes. Among the genes that make up the second tier of the LuxR regulatory cascade, many play a role in metabolism. This finding is not surprising when considering nutrient availability as cells reach quorum; at HCD, the nutrient profile available to the community changes and, thus, cells reconfigure metabolic programs, via LuxR, to respond accordingly. This is evidenced by the observation that acetate utilization genes are indirectly upregulated by LuxR. In V. fischeri, QS protects cells from acetate toxification; considering the conservation of regulatory schemes (i.e., increased expression of acetate utilization genes through the LuxR homologue) and acetyl-CoA synthase identity, a similar detoxification mechanism is likely present in V. harveyi (33). In addition, our data have uncovered a clear connection between LuxR and methionine biosynthesis. LuxR may be responsible for maintaining methionine homeostasis as the cell transitions into a low-nutrient status.

Overall, this work has shed light on a clear hierarchy of QS regulation that exists in V. harveyi and likely other vibrios. Why would a multileveled system of regulation be used in this fashion? One rationale is that the different hierarchies serve as checkpoints throughout regulatory transitions (i.e., changes in AI concentration in this case). Additional investigation will be required to validate this hypothesis, and investigation of hierarchies/checkpoints present in the LCD arm of the QS system (the AphA subregulon) would also be informative. The discoveries made in this study contribute evidence to an evolving model of QS: amid its many responsibilities, reconfiguration of global cellular metabolism is yet another fundamental behavior that is controlled by LuxR and QS in V. harveyi.

MATERIALS AND METHODS

Bacterial strains and methods.

The strain of V. harveyi used is BB120 (ATCC BAA-1116), which was reclassified as Vibrio campbellii (51). For consistency with the literature, we refer to this strain as V. harveyi throughout this manuscript. E. coli strain S17-1 λpir was grown at 30°C with shaking (250 rpm) in lysogeny broth (LB) medium. V. harveyi and all mutant derivatives were grown at 30°C with shaking (250 rpm) in Luria murine (LM) medium (LB medium with 2% total NaCl). Antibiotics were used at the following concentrations: 10 μg/ml chloramphenicol, 10 μg/ml tetracycline, 50 μg/ml polymyxin B, and 40 or 100 μg/ml kanamycin (E. coli or V. harveyi, respectively). Plasmid DNA was transformed into electrocompetent E. coli S17-1 λpir cells using 0.2-cm electroporation cuvettes (USA Scientific). Transformed E. coli S17-1 λpir cells were used to transfer plasmid DNA into V. harveyi via conjugation on LB plates at 30°C overnight. Exconjugants were selected for using polymyxin B and other antibiotics described above on LM plates grown at 30°C overnight. All strains used in this study are listed in Table S1 in the supplemental material.

Molecular methods.

PCR was performed using Phusion high-fidelity (HF) polymerase (New England Biolabs). All restriction enzymes and T4 polynucleotide kinase were purchased from New England Biolabs and used according to the manufacturer’s protocol. All oligonucleotides were ordered from Integrated DNA Technologies. Oligonucleotides used for EMSAs and quantitative PCR (qPCR) are listed in Table S3 in the supplemental material. Plasmids were constructed via isothermal DNA assembly (IDA) (all enzymes and deoxynucleoside triphosphates [dNTPs] from New England Biolabs). PCR products and plasmids were cleaned up using Qiagen kits and sequenced using Eurofins Genomics. All plasmids used in this study are listed in Table S2 in the supplemental material. DNA samples were resolved using 1% agarose and 1× Tris-borate-EDTA (TBE) buffer.

Mutant construction.

V. harveyi mutants were constructed using the protocol as described previously (15). Briefly, 1,000 bp upstream and downstream of the target gene were cloned into the suicide vector pRE112 flanking a chloramphenicol (CM) resistance marker and a sacB counterselectable marker. pRE112 contains oriR_R6Kγ, which allows for replication in E. coli S17-1 λpir but not V. harveyi. Plasmids were conjugated into V. harveyi from E. coli and exconjugants selected on CM-containing LM plates. Once integrated, plasmid excision was selected for by streaking cells on LM plates containing 20% sucrose. Isolated colonies were screened using PCR to identify cells in which plasmid excision yielded a successful gene deletion. All mutants were confirmed with sequencing.

RNA isolation and qRT-PCR.

For RNA-seq and qRT-PCR experiments, cells were grown to an optical density at 600 nm (OD₆₀₀) of 1.0, and 5 ml of culture were harvested, pelleted, and frozen on liquid N₂. RNA isolation was performed as described in previous studies using a TRIzol-chloroform extraction method (3, 15). DNA in samples was digested and removed using DNase I (Ambion RNase-free kit). The RNA sample was purified using an RNeasy purification kit (Qiagen). The SensiFAST SYBR Hi-ROX one-step kit (Bioline) was used for quantitative real-time PCR (qRT-PCR). Reactions were set up according to the manufacturer’s protocol, and 50 ng of purified RNA from cultures was added to each reaction mixture in a 96-well plate (Applied Biosystems). qRT-PCR was performed using a StepOnePlus real-time PCR system (Applied Biosystems) using thermal cycling conditions according to the SensiFAST protocol (Bioline). The hfq gene was used as an internal standard for normalization, as its expression is constitutive (and unaffected by LuxR) in V. harveyi. To analyze relative fold changes in RNA levels, the standard curve method was used on data acquired from three independent biological replicates.

RNA sequencing.

RNA-seq analysis was performed at the Center for Genomics and Bioinformatics at Indiana University using RNA that was purified as described above. RNA-seq was performed as described previously (15).

Electrophoretic mobility shift assays.

DNA substrates for EMSA experiments were obtained via PCR. PCR products were cleaned with a purification kit (Qiagen) and the 5′ ends of the DNA (25 nM) were radiolabeled using T4 polynucleotide kinase (PNK) (New England Biolabs). Excess, unincorporated [γ-³²P]ATP was removed using G-50 spin columns (GE Life Technologies) by following the manufacturer’s protocol. DNA binding reactions were performed using 1 nM 5′-radiolabeled DNA in LuxR binding buffer (10 mM HEPES [pH 7.5], 100 mM KCl, 2 mM dithiothreitol [DTT], 200 μM EDTA). All reaction mixtures included 100 μg/ml bovine serum albumin (BSA) and 10 ng/μl poly(dI-dC). Reaction mixtures were incubated at 30°C for 30 min and then run on 6% Tris-glycine-EDTA (TGE) nondenaturing gel. Gels were dried at 80°C for 20 min and exposed to a phosphor screen overnight. Phosphor screens were visualized using a Typhoon 9210 (Amersham Biosciences).

Data availability.

Sequence data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus (NCBI GEO) database under accession number GSE144017.