Quorum sensing and iron-dependent coordinated control of autoinducer-2 production via small RNA RyhB in Vibrio vulnificus

Roles for the non-coding small RNA RyhB in quorum-sensing and iron-dependent gene modulation in the human pathogen V. vulnificus were assessed in this study. Both the quorum sensing master regulator SmcR and the Fur-iron complex were observed to bind to the region upstream of the non-coding small RNA RyhB gene to repress expression, which suggests that RyhB is associated with both quorum-sensing and iron-dependent signaling in this pathogen. We found that expression of LuxS, which is responsible for the biosynthesis of autoinducer-2 (AI-2), was higher in wild type than in a ryhB-deletion isotype. RyhB binds directly to the 5′-UTR (untranslated region) of the luxS transcript to form a heteroduplex, which not only stabilizes luxS mRNA but also disrupts the secondary structure that normally obscures the translational start codon and thereby allows translation of LuxS to begin. The binding of RyhB to luxS mRNA requires the chaperone protein Hfq, which stabilizes RyhB. These results demonstrate that the small RNA RyhB is a key element associated with feedback control of AI-2 production, and that it inhibits quorum-sensing signaling in an iron-dependent manner. This study, taken together with previous studies, shows that iron availability and cell density signals are funneled to SmcR and RyhB, and that these regulators coordinate cognate signal pathways that result in the proper balance of protein expression in response to environmental conditions.

RyhB, a non-translated small RNA, was first identified as important for iron metabolism in E. coli 26 . Since that time, genes homologous to RyhB have been found and characterized in the context of the iron regulon in many other bacteria such as V. cholerae, Salmonella typhimurium, Yersinia pestis, and Shigella dysenteriae [27][28][29][30] . Transcription is negatively regulated by Fur in the presence of iron. Under iron-limiting conditions, this repression is relieved and RyhB then represses genes encoding iron-containing proteins, such as superoxide dismutase and succinate dehydrogenase 26,31 . In addition, RyhB also activates the expression of shiA, a gene encoding shikimate permease, which is associated with siderophore synthesis, by directly pairing with the 5′-untranslated region of the shiA mRNA in E. coli 32 . RyhB also translationally down-regulates fur expression 33 . In V. cholerae, a ryhB mutant showed decreased mobility and poor biofilm formation compared to wild type 27 . RyhB represses SodB expression by pairing with the 5′-untranslated region of the sodB mRNA and causing the coupled degradation induced by RNaseE 34,35 . The sRNA chaperone Hfq is essential for this process as it binds and protects RyhB from RNase E degradation, thereby extending its half-life 34,[36][37][38] .
Iron is essential for the growth of living organisms, but low solubility in the environment of a living cell makes this element a limiting factor for growth. High cell density leads to an even more severe iron stringency. Meanwhile, when intracellular iron is in excess, radicals that compromise the viability of cells are generated 39 . Therefore, it is expected that cell density and iron availability are intertwined and that these factors work together to control the expression of related genes. Here we report a new connection between quorum-sensing and the iron stress response mediated by the small non-coding RNA RyhB in V. vulnificus. The results of this study provide evidence for a close relationship between iron levels and cell density in terms of gene regulation and also add to the increasingly long list of roles for non-translated small RNA in pathogenic bacteria.

Results
Identification of the transcription start site of RyhB and evidence for Hfq-dependent stabilization of the RNA. Our previous study showing the repression of smcR by the Fur-iron complex in V. vulnificus 7 led us to further investigate the relationship between quorum-sensing and iron in this pathogen. In both E. coli and V. cholerae, the Fur-iron complex represses ryhB, a gene encoding a small RNA [26][27][28] . Therefore we speculated that if any small RNAs affect quorum-sensing in an iron-dependent manner in V. vulnificus, the most likely candidate would be RyhB. A search of the genome sequence of V. vulnificus revealed a 224-bp gene with 56% similarity to the ryhB of V. cholerae. This gene mapped between the genes encoding DNA polymerase I (GenBank accession number VVMO6_02895) and porphobilinogen (VVMO6_02896). The transcription start site was identified through primer extension experiments and typical − 10 and − 35 consensus sequences also were observed to be present ( Supplementary Fig. S1). The chaperone Hfq is required to stabilize RyhB in E. coli 35 . Therefore, we looked for a similar role in V. vulnificus and compared levels of the RyhB transcript in wild type and a Δhfq mutant at several time points after rifampicin was added to pause transcription. As shown in Fig. 1, RyhB from wild type is still detected up to 30 min after rifampicin treatment with a half-life of about 30.7 min. On the other hand, in the hfq mutant RyhB was degraded quickly with a half-life of about 7.5 min, and was barely detectable at 15 min following the rifampicin treatment ( Fig. 1).

Fur represses RyhB expression in the presence of iron.
We assessed the expression levels of ryhB under different iron conditions. RNA was extracted from V. vulnificus grown in either iron rich medium (LB broth with or without 25 μM of FeSO 4 ) or under iron limiting conditions (LB broth with 200 μM of 2, 2′-dipyridyl) (Fig. 2a). In wild type MO6-24/O cells, the ryhB transcript is barely detectable in cells grown in LB broth with or without 25 μM of FeSO 4 , while an abundance of transcript was detected when the iron chelator 2, 2′-dipyridyl was added. In contrast, a fur-deletion mutant (Δfur) had high amounts of the RyhB transcript regardless of iron concentrations. These results suggest that iron strongly represses the transcription of ryhB and that Fur is necessary for this repression.
Fur regulates gene expression in response to environmental iron conditions by binding to a specific site called the 'Fur box' in the promoter region of target genes 6,40,41 . To verify that Fur regulates ryhB expression in a similar manner, a gel-mobility shift assay was performed using a DNA fragment containing the region upstream to ryhB and purified Fur protein in the presence of the divalent ion Mn 2+ (Fig. 2b). Fur specifically bound to this DNA sequence in the presence MnSO 4 and binding was abolished if EDTA was added to sequester the divalent ion (Fig. 2c). The binding site for Fur within the ryhB promoter region was defined through a DNase I footprinting assay in the presence of 100 μM MnSO 4 ( Supplementary Fig. S2). The result showed that Fur protected a region from − 64 to + 6 with respect to the transcription start site, covering the consensus − 10 and − 35 promoter sequences ( Supplementary Fig. S3).

SmcR represses transcription of ryhB by binding to the promoter region. Our previous study
showed that regulation of vvsAB, which encodes a biosynthetic enzyme for the siderophore vulnibactin, is coordinately regulated by both iron and quorum-sensing 6 . Furthermore, expression of the qrr genes, encoding small RNAs associated with quorum-sensing modulation 9 , also are regulated by iron 8 . These results point to a connection between RyhB and quorum-sensing regulation and that these are somehow associated with iron levels. In support of this prediction, analysis of the nucleotide sequences upstream of ryhB revealed similarities to the consensus binding motif for the quorum-sensing master regulator SmcR 42 ( Supplementary Fig. S3). www.nature.com/scientificreports/ To verify this, we measured the expression of ryhB using a luxAB transcriptional fusion as a reporter in the following strains: wild type MO6-24/O; a mutant with a deletion in luxO (ΔluxO) such that it lacks the cytoplasmic signal transducer that degrades SmcR at low cell density via Qrr and Hfq 8 ; an smcR-deletion mutant (ΔsmcR); and a luxOsmcR double mutant (ΔluxOΔsmcR). In both ΔsmcR and ΔluxOΔsmcR, transcription of ryhB was significantly higher than that of wild type at early stationary phase (A 600 ≅ 1.0) (Fig. 3). Conversely, in a ΔluxO mutant, transcription was just half that of wild type. These results indicated that SmcR represses ryhB expression at the transcriptional level. A DNA fragment containing the upstream region of ryhB was incubated with increasing amounts of purified SmcR and analyzed in a gel shift assay (Fig. 4a). As expected, SmcR bound directly to the promoter region of ryhB. DNase I footprinting identified the binding region as − 27 to + 1 with respect to the transcriptional start site, including a putative − 10 promoter sequence ( Fig. 4b and Supplementary Fig. S3). These results indicate that SmcR represses ryhB transcription by directly binding to a cis-element upstream of ryhB to prevent the binding of RNA polymerase.
RyhB promotes the production of AI-2 in V. vulnificus. The results described above suggested that RyhB is involved in the iron-dependent regulation of the quorum-sensing pathway in V. vulnificus. To examine this further, we compared the transcription of genes related to the quorum-sensing pathway in wild-type and a ryhB-deletion mutant using transcriptome analysis. Expression levels of luxPQUO and SmcR were not significantly different in the two isotypes. However, the expression of luxS, encoding an AI-2 biosynthetic enzyme, was approximately 40% lower in a ΔryhB mutant than in wild type (Supplementary Table S3). To verify this result, we compared both AI-2 production and LuxS expression in wild type and ΔryhB mutant backgrounds. Supernatants from each of the two V. vulnificus strains cultured in AB minimal medium with a low iron concentration were assessed for the production of AI-2 using the bio-indicator V. harveyi BB170 43 and measuring luminescence. The luminescence induced by the wild type culture supernatant was significantly higher than that www.nature.com/scientificreports/ www.nature.com/scientificreports/ of ΔryhB, and introduction of an exogenous ryhB-expressing clone into ΔryhB in trans restored luminescence to wild type levels, whereas introduction of vector alone did not (Fig. 5a). These results indicate that RyhB promotes expression of LuxS under iron-limiting conditions. We then assessed the effects of RyhB on the expression of a gene normally regulated by the AI-2 quorum-sensing system. We quantitatively measured the expression of vvpE, which encodes elastase, a virulence factor that is positively regulated by quorum sensing 7,44 . Iron repressed vvpE expression regardless of RyhB. However, in the absence of iron, expression in the ΔryhB mutant was significantly lower than in wild type (Fig. 5b). This result supports the idea that enhanced production of AI-2 by RyhB promotes quorum-sensing signaling. In summary, these results indicate that RyhB increases AI-2 mediated quorum-sensing signaling by enhancing the production of AI-2 itself, and that SmcR represses RyhB as a mechanism for feedback inhibition to repress AI-2 production. The presence of iron, as part of the Fur-iron complex, down-regulates overall quorum-sensing signaling by inhibiting the transcription of both SmcR and RyhB.
RyhB delays the decay of luxS mRNA by binding directly to the 5′-UTR of LuxS. The next question was how RyhB enhances the production of AI-2 at the molecular level. The non-coding sRNA RyhB posttranscriptionally regulates gene expression by affecting the turnover of mRNAs in E. coli 26 . We assumed that RyhB would affect the luxS mRNA in the same manner. To test this, levels of the luxS transcript were measured over time after cells were treated with rifampicin. As shown in Fig. 6, luxS mRNA degraded faster in ΔryhB than in wild type. Previous studies reported that Hfq is required for the functioning of RyhB. To test for this possibility in V. vulnificus, we also compared the levels of luxS mRNA following rifampicin treatment in wild type, ΔryhB, Δhfq, and a ΔryhBΔhfq double mutant. As shown in Fig. 6, in the absence of Hfq, luxS mRNA levels decreased significantly, independent of RyhB, indicating that even though RyhB appears to stabilize luxS mRNA, this function requires Hfq.
Generally, non-coding RNAs modulate gene expression by base pairing with target mRNAs at the 5′-UTR 45 , and the sRNA chaperone Hfq either stabilizes the sRNA or facilitates binding of the sRNA to the target 46 . We predicted that RyhB delays the decay of luxS mRNA by binding directly to the 5′-UTR. A 32 P-labeled 102-bp luxS RNA fragment including bases − 62 to + 40 with respect to the luxS translation start site was expressed by in vitro transcription using T7 RNA polymerase. Full-length RyhB was also transcribed using in vitro transcription, and increasing amounts of this transcript were incubated with the labeled luxS mRNA probe in the presence or absence of Hfq. In the presence of Hfq, RyhB binds to the 5′-UTR of luxS mRNA in a concentration-dependent www.nature.com/scientificreports/ manner ( Fig. 7a), whereas no binding was observed in the absence of Hfq. We also assessed the effect of Hfq on the half-life of RyhB and observed that stability was significantly reduced in Δhfq compared to wild type (Fig. 7b). Together these indicate that RyhB binds directly to the 5′-UTR of luxS mRNA and stabilizes it with the assistance of Hfq.  was used to predict secondary structures that may form in the 5′-UTR of luxS mRNA (Fig. 8a), one of which is a stem-and-loop structure (SL2) that would obscure the start codon. Hybridization between RyhB and the 5′-UTR of luxS mRNA was predicted to include base pairing in three regions labeled HR (Hybridized Region) 1, 2, and 3 ( Fig. 8b). If hybridization occurs at these sites, loops SL1 and SL2 would be resolved and the start codon of luxS mRNA would be exposed (Fig. 8b).

SmcR-probe complex
To confirm these predictions, we performed a primer extension in the presence or absence of the 5′-UTR of luxS mRNA by reverse transcription using RyhB RNA as a template and primer RyhB-PE (Supplementary www.nature.com/scientificreports/ Table S3, Supplementary Fig. S3) which is complementary to the 3′-end of RyhB RNA. We expected that binding of RyhB to the luxS mRNA in the presence of Hfq would produce shorter immature cDNA products due to the formation of secondary structures between two RNA molecules that partially block the primer extension. In fact, we observed partially extended cDNA molecules that terminated at cytosine at position 108 and uracil at position 84 relative to the first base of the ryhB transcript (Fig. 9a,b). These two sites correspond to the 3′-end of the HR2 and HR3 regions of RyhB, respectively. cDNA terminated at adenine at position 97 also was observed. This residue is the first base at the 3′-end of Loop 2 (Fig. 9b). These results suggested that hybridization to luxS mRNA does indeed occur at HR2 and HR3. We speculate that there may be weak hybridization at HR1 such that termination of cDNA at this region was barely detectable. To confirm the involvement of the HR2 and 3 regions of RyhB in regulation of the luxS expression, we constructed derivatives of RyhB with mutations in either or both of HR2 or 3 (named HR2m, HR3m, and HR2&3m) by site-directed mutagenesis. A mutation called HR0m at a site outside of the hybridized region (HR0) was generated as a positive control ( Fig. 9b and Supplementary Fig. S3). Wild type RyhB and each of the RyhB mutations (HR0m, 2m, 3m, and 2&3m) were introduced into a ryhBnull mutant strain (ΔryhB) and assessed for both AI-2 production using the V. harveyi indicator strain BB170 (Fig. 10a) and levels of LuxS by western hybridization using antibody against purified LuxS (Fig. 10b). RyhB with HR0m did not differ significantly from wild type. However, the presence of either HR2m or 3m led to significant decreases in both AI-2 production and LuxS expression, suggesting that HR2 and 3 are critical for LuxS regulation. However, the results of luxS qRT-PCR for wild type and the ryhB mutants showed that these mutations did not affect transcription levels of luxS under experimental conditions (Fig. 10c), suggesting that the regulation of the LuxS expression by RyhB is exerted at the translational level.

Discussion
It is well known that an adequate concentration of iron is required for the survival of bacteria 2,48 . Nevertheless, iron is not readily available in natural environments due to its low solubility at neutral pH 49 . Furthermore, it is very difficult for pathogenic bacteria to compete with host cells for iron 50 . Therefore, pathogenic bacteria are equipped with various mechanisms to more effectively scavenge iron in the host environment. Our previous studies showed that iron levels affect quorum-sensing pathways 6,8 . This study more specifically defines the relationship between iron levels and quorum-sensing by showing that iron inhibits the production of AI-2, an initiation signal for quorum-sensing. Our observations lead us to propose that one of the most important roles for quorum-sensing is to control the timing of expression of virulence factors in pathogens as they prepare to attack host cells to acquire nutrients such as iron.
Fur is the most common regulator of iron-dependent target genes. We previously showed that cell density signals and iron converge on the AI-2 quorum-sensing pathway in V. vulnificus, and that the regulation is also mediated by Fur 7,8 . In the presence of iron, Fur represses the expression of major components in this pathway such as luxO, five qrr genes, and smcR. However, a deletion in fur did not fully abolish this repression 8 . This observation led us to search for additional factors known to be involved with iron-dependent regulation. We examined these factors and found that RyhB affects the expression of luxS, which encodes the biosynthesis of AI-2. Neither genetic analysis using a reporter fusion to luxS nor gel-shift assays using purified Fur-iron along with the upstream region of luxS provided any evidence that Fur directly controls luxS expression in V. vulnificus (our unpublished data). Interestingly, unlike other components of the quorum-sensing pathway, luxS expression appears to be controlled by RyhB at the translational level.
This mode of regulation may allow for a continuous basal level of AI-2 expression even in the presence of iron rather than the tight suppression that would occur if Fur were involved. If cells do not produce any AI-2 at all, the quorum-sensing pathway cannot quickly resume once it is shut down by the Fur-iron complex. This scenario may also explain the presence of the strong promoter sequence of ryhB, which is very similar to the canonical consensus promoter sequence in this organism. In this way, expression of ryhB is initiated swiftly www.nature.com/scientificreports/ once the repression exerted by Fur-iron is relieved. Employing RyhB may also provide a particular advantage in that this small RNA functions at the translational level to alter expression quickly and, due to the short half-life of the molecule, repression is rapidly relieved if the SmcR or Fur-iron signals disappear. The chaperone Hfq is absolutely required for RyhB activity. Furthermore, levels of luxS mRNA in a hfq-deletion mutant were even lower than in a ryhB-deletion isotype (Fig. 6), suggesting that yet other factor may exist, possibly an additional small RNA specific to luxS. A search for non-coding small RNAs modulated by RyhB may provide more insight into the iron-dependent modulation of genes. Quorum-sensing signaling pathways are a valuable way for pathogenic bacteria to adapt to host environmental conditions but they are energetically expensive due to the numerous factors and variety of regulatory mechanisms that are employed. Therefore, it is necessary for cells to be equipped with the ability to control this signaling when it is not needed. For example, once a quorum-sensing signaling molecule accumulates within a closed environment, it is possible that the cell would continue to sense the signal even when expression of target genes are no longer required. The best way to avoid such a waste of energy is to lower the concentration of signaling molecules. Two possible mechanisms are either to degrade the signal molecule or to cease synthesis of the signal molecule. It is known that bacteria harbor enzymes to degrade the homoserine lactone signal molecule 51,52 . However, whether the AI-2 molecule is degraded in a similar way remains to be elucidated. To the best of our knowledge, ours is the first study to show that feedback inhibition of quorum signaling occurs through the inhibition of AI-2 synthesis. www.nature.com/scientificreports/ Other small RNAs called Qrrs are also known to be involved in the feedback control of quorum-sensing 8,53,54 . However, these appear to be involved in elaborate control of the signal to properly adapt to changes in cell density and not in feedback control of the quorum-sensing pathway as a whole. AI-2 is a signal that plays a very important role in bacterial community structure, affecting inter-species communication as well as intra-species communication 9 . AI-2 signaling from pathogenic bacteria can be transmitted to other recognizable bacteria, and a decrease in AI-2 levels in response to the presence of iron can certainly affect other microbiota present in the infected regions of a host. A comparison of the gut microbiota in mice infected with wild type V. vulnificus and mice infected with a luxS-deleted isotype are significantly discernable (our unpublished data). This suggests that the presence of iron alone can indeed affect the community structure of the microbiota, in addition to influencing levels of the quorum-sensing signal molecules and related target functions. Studies on the effect of AI-2 on host microbiota in the context of iron concentrations may provide interesting insight into host-microbe interactions.
Small RNAs have been widely studied especially in the context of virulence regulation in pathogenic bacteria 31,37,38,55 . In Vibrionacea, small RNAs has been intensively studied in V. cholerae 56 . However, in other species, the role of small RNAs awaits further elucidation. The nucleotide sequences of RyhB and luxS mRNAs and possible hybridization between these two molecules in three related Vibrio spp. (V. cholerae, V. parahaemolyticus, and V. harveyi) demonstrated that, for all three, the start codon for luxS is hidden within stem-loop structures, unless binding with RyhB resolves this secondary structure ( Supplementary Fig. S4), implying that this type of regulation is common among Vibrionaceae. Considering that major virulence factors are controlled by quorum sensing within these species, appreciation of the role of RyhB in mechanisms related to pathogenicity will provide useful information about the disease process.
Iron, or the molecules with which it interacts, may be good targets for the development of agents to control pathogenic bacteria. This study demonstrated that iron inhibits the quorum sensing signaling pathway associated with expression of virulence factors, and any tactics to enhance iron solubility may be of value to control pathogenic bacteria. Development of molecules such as anti-sense RNAs that interfere with the action of RyhB or possibly direct inhibitors of Hfq may be useful clinical approaches.

Materials and methods
Strains and culture condition. Strains and plasmids used in this study are listed in Supplementary   Table S1. Escherichia coli cells were cultured in Luria-Bertani (LB) medium at 37 °C. V. vulnificus strains were grown at 30 °C in LB medium supplemented with 2.0% (w/v) NaCl (LBS) or in AB minimal medium (0.3 M NaCl, 0.05 M MgSO 4 , 0.2% casamino acid, 1 mM KPO 4 , 1 mM l-arginine, pH 7.5) 57 . All media components were purchased from Difco (Detroit, USA), and antibiotics were purchased from Sigma (St. Louis, USA).
Determination of the transcription start site of ryhB. RNA was isolated from V. vulnificus using the RNase Easy Mini Kit (Qiagen, Valencia, CA), and RNA concentration was determined using a Biophotometer (Eppendorf, Hamburg, Germany). A 500 ng sample of RNA extracted from V. vulnificus was incubated with www.nature.com/scientificreports/ 5′-labeled primer RyhB-PE at 65 °C and chilled on ice. Reverse transcription was performed using a PrimeScript RT reagent kit (Takara, Tokyo, Japan). The resulting product and sequencing ladder were resolved on a 6% polyacrylamide sequencing gel to identify the transcription start site. Plasmid pGEM-T-RyhBup was digested with restriction enzymes PstI and BamHI, and the resulting DNA fragment containing the ryhB upstream region was ligated into plasmid pGEM-T-RyhBdown to construct pGEM-T-RyhBKO. The construction was digested with XhoI and XbaI, and subsequently ligated into pDM4 to obtain pDM4-RyhBKO which was subsequently introduced into E. coli S17-1 λpir to be mobilized into V. vulnificus by conjugation. Double crossover selection was performed on a 10% sucrose plate as described previously 58 . The ryhB deletion mutant, ΔryhB, was confirmed by PCR and DNA sequencing.

Gel shift assay.
A 336-bp DNA fragment including the ryhB promoter region (− 213 to + 123 with respect to the transcriptional start site) was PCR-amplified using the 32 P-labeled primers RyhB-F1 and RyhB-R1 (Supplementary Table S2). For the gel shift assay, 10 ng of labeled DNA fragment was incubated with increasing amounts of purified SmcR (0 to 1 μM) or Fur (0 to 1 μM) in a 20 μl reaction for 30 min at 37 °C. The SmcR binding reaction buffer contained 10 mM HEPES, 100 mM KCl, 200 μM EDTA, and 10% glycerol at pH 7.5. The Fur binding reaction contained 10 mM HEPES, 100 mM KCl, and 10% glycerol at pH 7.5 and was supplemented with either 100 μM MnSO 4 or 1 mM EDTA. The binding reaction was terminated by the addition of 3 μl sucrose dye solution (0.25% bromophenol blue, 0.25% xylene cyanol, 40% sucrose) and the samples were resolved on a www.nature.com/scientificreports/ 6% neutral polyacrylamide gel. For gel shift assays of RyhB and LuxS 5′-UTR, 10 ng of 32 P-labeled LuxS 5′-UTR and increasing amounts of RyhB were incubated at 70 °C for 10 min and then chilled on ice for at least 1 min. Then 2 μl of 10 × structure buffer (100 mM Tris-Cl, 50 mM magnesium acetate, 1 M ammonium chloride, 5 mM DTT, pH 7.5) and Hfq were added and incubated at 30 °C for 10 min. The binding reaction was terminated by the addition of 4 μl of sucrose dye and resolved on a 6% neutral polyacrylamide gel.

Construction of ryhB::luxAB and vvpE::luxAB transcriptional fusions. Primers PryhB-F and
PryhB-R were used for PCR amplification of the ryhB promoter region, which is − 251 to + 123 relative to the transcription start site. Primers PvvpE-F and PvvpE-R were used for PCR amplification of the vvpE promoter region. The resulting products were digested with KpnI and XbaI (Takara, Ohtsu, Japan) and cloned into vector pHK0011 to construct pHryhB and pHvvpE, respectively, and each was conjugated into V. Detection of AI-2 production. The AI-2 assay was performed using the V. harveyi reporter strain BB170 as previously described 43 . Briefly, BB170 was prepared by culturing overnight in LB broth at 30 °C, then washing twice and diluting 1:3000 in fresh AB minimal medium. Test cultures of V. vulnificus were grown overnight in LBS broth and then washed once before diluting 1:250 in fresh AB broth. Each hour, A 600 was measured and 10 μl of cell-free supernatant was collected and mixed with 90 μl of diluted BB170, prepared as described above, in 96 well plates at 30 °C. The luminescence was measured using a Mithras LB 940 multimode microplate reader (Berthold, Germany).

RNA synthesis by in vitro transcription.
Template DNA of ryhB and luxS was prepared for in vitro transcription by PCR using primers containing the T7 promoter sequences as shown in Supplementary Table S2. RNA was synthesized by in vitro transcription using T7 RNA polymerase (Takara, Tokyo, Japan) using the prepared template DNA at 37 °C following the protocol provided and was further purified using a Monarch RNA Cleanup Kit (NEB, Cambridge, USA).
Determination of the regions of RyhB RNA and luxS mRNA that hybridize with each other using primer extension with reverse transcriptase. RyhB (50 ng) was incubated with LuxS 5′-UTR and 32 P-labeled primer RyhB-PE, which is complementary to RyhB from residues 167 and 189 relative to the transcription start site of ryhB, at 65 °C for 10 min and then chilled on ice. Hfq and 10 × structure buffer (100 mM Tris-Cl, 50 mM magnesium acetate, 1 M ammonium chloride, 5 mM DTT, pH 7.5) were added and the mixture was incubated at 30 °C for 10 min. Lastly, dNTPs and 1 unit of SuperScript III reverse transcriptase (Invitrogen, Carsbad, USA) were added and incubated at 30 °C for 1 h. Reactions were terminated at 85 °C for 5 min, and samples were resolved on a 6% polyacrylamide sequencing gel alongside a sequencing ladder generated by the same primer. www.nature.com/scientificreports/ 8.0). The eluted proteins were assessed for purity using 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). To purify the Hfq protein, DNA spanning the 261-bp Hfq ORF encoding 86 amino acids of Hfq was PCR-amplified using primers STREP-HFQF and STREP-HFQR (Supplementary Table S2). The amplified fragment was cloned into pASK-IBA-7 to generate pASK-IBA-Hfq (Supplementary Table S1), and Hfq was purified as described above.

Construction of a ryhB-deletion and series of site-directed mutations in ryhB.
A DNA fragment containing the complete sequence of ryhB was PCR-amplified using primers CryhB-F and CryhB-R (Supplementary Table S2). The resulting product was ligated into pGEM-T easy vector (Promega). After confirmation by sequencing, the DNA fragment containing the ryhB sequence was cut with KpnI and XbaI and cloned into pRK415, generating pRK-RyhB. Nucleotides that potentially bind to LuxS 5′-UTR were mutated using the EZchange Site-directed Mutagenesis Kit (Enzynomics, Deajeon, Korea). Primers ryhB_SDM2_F and ryhB_SDM2_R were used to construct pRK-RyhB2m; primers ryhB_SDM3_F and ryhB_SDM3_R were used to construct pRK-RyhB3m; all four primers (ryhB_SDM2_F&R, ryhB_SDM3_F&R) were used to construct pRK-RyhB2&3m; and primers ryhB_SDM_CON_F and ryhB_SDM_CON_R were used to construct the control pRK-RyhB0m (Supplementary Table S2). The resulting constructs were mobilized from S17-1 to mutant strain ΔryhB. Conjugants were selected in thiosulfate-citrate-bile salts-sucrose agar (TCBS) plate medium supplemented with 1 μg/ml tetracycline.
Preparation of polyclonal rabbit antibody against purified LuxS and western hybridization. Purified LuxS was used to produce polyclonal rabbit antibodies (Ab Frontier, Seoul, South Korea www.nature.com/scientificreports/