Research ArticlesHost-Pathogen Interactions

Rosmarinic acid is a homoserine lactone mimic produced by plants that activates a bacterial quorum-sensing regulator

See allHide authors and affiliations

Science Signaling  05 Jan 2016:
Vol. 9, Issue 409, pp. ra1
DOI: 10.1126/scisignal.aaa8271

Plants send out a bacterial mimic

Plants and microbes have evolved mechanisms to communicate. Corral-Lugo et al. determined that a plant compound, rosmarinic acid, bound to a protein in the bacterial quorum-sensing pathway, which bacteria use to regulate “community” interactions, such as the formation of biofilms. In vitro analysis showed that rosmarinic acid bound to RhlR, a transcriptional regulator in the quorum-sensing pathway of the plant and human pathogen Pseudomonas aeruginosa. Both reporter gene analysis and in vivo analysis of quorum-sensing responses showed that rosmarinic acid stimulated RhlR activity, thereby functioning as a mimic of the bacterial ligands. Identification of this molecular mimic has both agricultural and biomedical implications by enabling strategic disruption of bacterial communication.


Quorum sensing is a bacterial communication mechanism that controls genes, enabling bacteria to live as communities, such as biofilms. Homoserine lactone (HSL) molecules function as quorum-sensing signals for Gram-negative bacteria. Plants also produce previously unidentified compounds that affect quorum sensing. We identified rosmarinic acid as a plant-derived compound that functioned as an HSL mimic. In vitro assays showed that rosmarinic acid bound to the quorum-sensing regulator RhlR of Pseudomonas aeruginosa PAO1 and competed with the bacterial ligand N-butanoyl-homoserine lactone (C4-HSL). Furthermore, rosmarinic acid stimulated a greater increase in RhlR-mediated transcription in vitro than that of C4-HSL. In P. aeruginosa, rosmarinic acid induced quorum sensing–dependent gene expression and increased biofilm formation and the production of the virulence factors pyocyanin and elastase. Because P. aeruginosa PAO1 infection induces rosmarinic acid secretion from plant roots, our results indicate that rosmarinic acid secretion is a plant defense mechanism to stimulate a premature quorum-sensing response. P. aeruginosa is a ubiquitous pathogen that infects plants and animals; therefore, identification of rosmarinic acid as an inducer of premature quorum-sensing responses may be useful in agriculture and inform human therapeutic strategies.


Plants live in association with fungi and bacteria, and it is believed that plant evolution was influenced by the presence of these associated microorganisms (1). During this evolution, diverse signaling systems emerged that permitted mutual plant-microorganism sensing. Quorum sensing (QS) is a mechanism of communication between bacteria and is based on the synthesis, detection, and response to QS signals (QSS). As cell density increases, QSS accumulate in the environment and are sensed by bacterial proteins called QS regulators, which in turn control the expression of genes; the products of these genes direct activities that are beneficial when performed by groups of bacteria acting in synchrony (2). Homoserine lactones (HSLs) produced by Gram-negative bacteria are the best studied and possibly the most common group of bacterial QSS (3). Frequently, pairs of genes encoding the HSL synthase and the HSL-sensing transcriptional regulator are found close to each other in bacterial genomes. In addition, many bacteria have additional paralogs of HSL-sensing regulator genes that are not associated with an HSL synthase gene and that were consequently termed solo or orphan regulators (4).

Bacteria-to-plant and plant-to-bacteria signaling are also based on QS systems. A proteomics study showed that HSLs modulate the expression of a large number of genes in the legume Medicago truncatula (5). Similarly, a transcriptomic study revealed that C6-HSL, a bacterial QS molecule produced in the rhizosphere, changed gene expression in Arabidopsis thaliana (6). HSL signaling processes are, in part, responsible for the induced systemic resistance of plants toward bacterial pathogens (7); these processes also modulate plant growth (8).

In addition, different plants produce compounds that interfere with the bacterial QS mechanism. Extracts (9, 10) and macerates of different plants, plant parts, and seeds (1115), as well as exudates from seeds (16) or seedlings (17), interfere with bacterial QS mechanisms. Additionally, leaf washings from 17 different plants stimulated or inhibited HSL-dependent activities in bacteria (18). Furthermore, QS-dependent gene expression is altered when pathogenic bacteria grow in their host plants (19). Biofilm formation and HSL production increase in the presence of different plant-derived phenolic compounds (20); however, these compounds do not act as HSL mimics. In contrast, most of the experiments using extracts from plants stimulated rather than inhibited QS-dependent gene expression (1), with the data suggesting that these plant compounds act as HSL mimics and bind to the autoinducer-binding domain of QS regulators. A molecular docking study identified rosmarinic acid (RA), naringin, morin, mangiferin, and chlorogenic acid (21) as plant-derived compounds that were predicted to bind to QS regulators. Although in that study each of these compounds inhibited QS-mediated phenotypes, suggesting that they function as QS antagonists, potential toxic effects were not evaluated. Experimental confirmation of the binding of any of these compounds to QS regulators has not been done. The algal compound lumichrome, which is a riboflavin derivative, stimulates the activity of a QS regulator of Pseudomonas aeruginosa (22).

Here, we used P. aeruginosa PAO1 as a model organism to screen for plant-derived HSL mimics. This bacterium is a ubiquitous pathogen that infects a wide range of species, including humans and different plants such as barley, poplar tree (23), and lettuce (24, 25). As a model for studying the effect of QS on pathogenic traits, P. aeruginosa has a multisignal QS system that is based on the synthesis and detection of signals that belong to two different classes, namely, Pseudomonas HSLs and quinolone signals (26, 27). The HSL response is mediated by two pairs of synthases and regulators—the synthase LasI and the regulator LasR (LasI/LasR) and the synthase RhlI and the regulator RhlR (RhlI/RhlR)—as well as by the orphan regulator QscR. RhlI produces the signaling molecule N-butanoyl-homoserine lactone (C4-HSL), and LasI produces N-3-oxododecanoyl-homoserine lactone (3-Oxo-C12-HSL) (26). The P. aeruginosa QS system is hierarchically organized with LasR at the top of the signaling cascade: LasR activation stimulates transcription of multiple genes, including rhlR, rhlI, and lasI. The QS cascade then modulates multiple QS phenotypes, including changes in the amounts of elastase, pyocyanin, rhamnolipid, and hydrogen cyanide (28).

Here, we used ligand-free LasR and RhlR purified from Escherichia coli without added HSL (29) for microcalorimetric binding studies of plant-derived compounds that were selected on the basis of in silico docking experiments. We identified RA as a plant-derived compound that binds with nanomolar affinity to RhlR. In transcription assays with RhlR, RA significantly more effectively stimulated transcription at lower concentrations than C4-HSL. In bioassays, RA, but not the closely related compound chlorogenic acid, stimulated biofilm formation and the production of the virulence factors pyocyanin and elastase. RA is produced exclusively in plants and not in bacteria (30, 31). Thus, these data showed that RA acts as a QS regulator agonist, thereby providing the molecular identification of a plant QSS mimic.


RA binds to purified RhlR with high affinity

A major limitation in the study of the HSL-sensing regulators is their instability in the absence of HSL (3234). Because recombinant regulator purified from E. coli cultures binds HSL added to the culture medium, using most methods, the bound HSL copurifies with the protein, resulting in partially saturated regulators and thus hampering ligand-binding studies. We developed a method that enables the purification of recombinant RhlR and LasR without the addition of HSL, thereby providing a system for performing ligand-binding analysis (29). Recombinant RhlR and LasR isolated with this method in the absence of HSL bound C4-HSL and 3-Oxo-C12-HSL, respectively, with dissociation constant (KD) values of 1.66 ± 0.4 μM and 1.14 ± 0.2 μM, as determined by isothermal titration calorimetry (ITC) (Fig. 1, A and B).

Fig. 1 Microcalorimetric binding studies of LasR and RhlR.

(A) Titration of 8 μM RhlR with 100 μM C4-HSL. (B) Titration of 8 μM LasR with 100 μM 3-Oxo-C12-HSL. (C) Structural superimposition of the homology model of the autoinducer-binding domain of RhlR (in orange) with the structure of the analogous domain from LasR [in pink; Protein Data Bank (PDB) ID: 3IX3]. The alignment was done using Subcomb (69). (D) Titration of 19 μM RhlR with 0.66 mM RA (I) or chlorogenic acid (II). Lower panel plots the titration data for RA. (E) Titration of buffer (I) and 8 μM LasR (II) with 0.66 mM RA. For the titration data (A, B, D, and E), the upper panels show the raw titration data and the lower panels are concentration-normalized and dilution heat-corrected integrated peak areas of the titration data fitted with the “One binding site model” of the MicroCal version of Origin.

To identify potential ligands, we conducted in silico docking experiments of ligands present in a database of natural compounds to the structure of LasR and a model of RhlR. We used the structure of the LasR autoinducer-binding domain in complex with 3-Oxo-C12-HSL (PDB ID: 3IX3) to generate a homology model of the analogous domain of RhlR, which could be closely superimposed onto the template (Fig. 1C). We used 3-Oxo-C12-HSL and C4-HSL as controls in the docking experiments (Table 1) and then we screened the Natural Compounds subset of the ZINC compound database (5391 compounds) and selected those of plant origin and with docking scores below −8 for further analysis. Most of the tested compounds had lower docking scores at RhlR or LasR when compared to those of their cognate HSL ligands (Table 1).

Table 1 Results from in silico docking and experimental binding studies of plant-derived compounds to RhlR and LasR.

Shown are XP scores for the in silico docking of different ligands to a homology model of the RhlR autoinducer domain and to the structure of the analogous domain of LasR. Binding parameters are derived from ITC experiments of purified LasR and RhlR with the ligands listed. Data shown are means and SD from three independent experiments.

View this table:

Microcalorimetric binding studies with the selected compounds were performed to assess binding to ligand-free LasR and RhlR. We found that RA bound only to RhlR (Fig. 1D) and not to LasR (Fig. 1E). No other selected compound bound to RhlR or LasR (Table 1).

We calculate that RA bound to RhlR with a KD of 0.49 ± 0.08 μM and had small favorable enthalpy changes (ΔH = −0.4 ± 0.05 kcal/mol). Similar to the HSL ligands, the binding stoichiometry was close to 1:1, which can be observed as the point of inflection of the sigmoidal binding curves in Fig. 1D with respect to the lower x axis. To assess the specificity of this interaction, we titrated RhlR with chlorogenic acid, a compound structurally similar to RA (Fig. 2) and that had a low docking score (Table 1). Chlorogenic acid did not cause significant heat changes, indicating that this compound did not bind RhlR (Fig. 1D).

Fig. 2 In silico docking of RA to the autoinducer-binding domains of RhlR or LasR.

The RhlR autoinducer-binding domain is a homology model (see Materials and Methods) and contains docked C4-HSL, and the LasR structure (PDB ID: 3IX3) contains bound 3-Oxo-C12-HSL. The best binding positions of docked ligands with minimal glide score (XP G score) and glide energy (G energy) are displayed. The docking scores are shown in Table 1. The structures of different ligands are shown in the lower part of the figure.

The RhlR model containing the best fit of docked RA and C4-HSL showed that both ligands overlap (Fig. 2). Although the docking simulations with LasR predicted that RA could overlap with bound 3-Oxo-C12-HSL (Fig. 2), the ITC studies showed an absence of binding of RA for LasR (Fig. 1E).

RA stimulates RhlR-mediated transcription

To determine whether RA behaved as an agonist or antagonist, we conducted in vitro transcription assays with a 490–base pair (bp) DNA fragment containing the hcnABC promoter, which is activated by RhlR (35). We performed the experiments with a constant RhlR concentration of 25 μM in the presence of either equimolar or half-equimolar concentrations of C4-HSL and RA. Because the KD values for both ligands are much less than the protein concentration, we expect that protein saturation at equimolar ligand concentrations is comparable and almost complete. Both C4-HSL and RA stimulated RhlR-dependent transcription (Fig. 3A), but quantification revealed that RA exhibited a significantly greater activity compared to C4-HSL when the response to equal concentrations of ligand was compared to the transcription detected in the control (absence of ligand) condition (Fig. 3B). In the presence of half-equimolar C4-HSL or RA to RhlR concentrations, transcriptional activity was less than at equimolar concentrations of ligand to regulator, thus indicating dose dependence of the response. Consistent with the lack of binding to RhlR, chlorogenic acid did not increase transcription. These data showed that compared to the bacterial ligand C4-HSL, RA was more effective at activating RhlR-dependent transcription from this promoter.

Fig. 3 The capacity of C4-HSL and RA to stimulate RhlR-mediated transcription in vitro.

(A) Representative acrylamide gel of an in vitro transcription assay using a DNA fragment containing the hcnABC promoter that is induced by RhlR (35). Conditions match those listed below the graph in (B). Ligands tested included C4-HSL, RA, and chlorogenic acid. (B) Densitometric analysis of in vitro transcription assays. Shown are means and SD from three individual experiments. *P < 0.05, Student’s t test.

RA stimulates QS gene expression in vivo

To assess the capacity of RA to modulate QS-dependent gene expression in vivo, we conducted experiments in E. coli and P. aeruginosa. RhlR controls the expression of the gene encoding its cognate HSL synthase RhlI (36). Therefore, we transformed E. coli BL21 with either plasmid pPET28b-RhlR (RhlR expression plasmid) or pPET28b (empty plasmid as control) and pMULTIAHLPROM containing an rhlI::lacZ transcriptional fusion. β-Galactosidase measurements showed statistically significant increases in gene expression in the presence of C4-HSL or RA, whereas chlorogenic acid did not stimulate expression (Fig. 4A).

Fig. 4 RA- or C4-HSL–mediated activation of a QS reporter in bacteria.

(A) Transcriptional activation in E. coli. β-Galactosidase measurements at 2 hours after induction of E. coli BL21 containing pET28b-RhlR (expression plasmid for RhlR) or the empty expression plasmid and pMULTIAHLPROM containing a rhlI::lacZ transcriptional fusion. The ligand concentrations were 100 μM. CA, chlorogenic acid. (B) Transcriptional activation over time by the indicated ligands in P. aeruginosa lasI/lasR containing pMULTIAHLPROM. Bars represent the β-galactosidase measurements at different time intervals after the addition of dimethyl sulfoxide (DMSO; control), C4-HSL, or RA. The line graphs represent growth curves of the corresponding cultures. (C) Concentration-dependent transcriptional activation by the indicated ligands in P. aeruginosa lasI/lasR containing pMULTIAHLPROM. Dose-response curves for each ligand from samples taken 4 hours after the addition of the ligand. Shown are means and SD from three independent experiments conducted in duplicate. **P < 0.01, ***P < 0.001, Student’s t test.

To study gene expression in P. aeruginosa, we introduced the rhlI::lacZ reporter plasmid into the lasI/lasR double mutant and measured β-galactosidase activity in samples taken at different time intervals after the addition of either DMSO, C4-HSL, or RA. The data indicated that the bacteria exhibited differential kinetics in response to the two RhlR ligands. Induction of the reporter in the cultures exposed to RA peaked within 1 hour and was significantly greater than that of the control and the C4-HSL–exposed cultures at both 1 and 2 hours (Fig. 4B). At subsequent time points, reporter activity decreased in the RA-containing cultures. We predicted that the reduction in β-galactosidase activity of RA-containing cultures after 2 hours indicated that RA was metabolized. The activity of C4-HSL cells was comparable to that of the control after 1 hour but was significantly greater for time points 6 to 8 hours, which may be due to slow uptake.

We determined the dose-response relationships for the increase in gene expression induced by C4-HSL and RA (Fig. 4C). Because of the differences in kinetics, we took measurements for the concentrations of ligand tested at 4 hours, the time point at which C4-HSL and RA induced similar amounts of β-galactosidase activity (Fig. 4B). At concentrations of 1 to 100 μM, transcriptional activities were comparable, which may be due to metabolization of RA (Fig. 4C). However, at 0.5, 1, and 2 mM, the β-galactosidase activity in response to RA was higher than that induced by C4-HSL (Fig. 4C). At 5 mM RA, we observed a further increase in β-galactosidase activity, but we could not perform similar measurements with C4-HSL due to the solubility limit of this compound. These data are consistent with the in vitro transcription experiments and support the conclusion that the capacity of RA to stimulate transcription is superior to that of C4-HSL.

To confirm that P. aeruginosa could metabolize RA, we analyzed bacteria grown in minimal medium containing 1 to 10 mM RA as the only carbon source. P. aeruginosa grew with 1 to 5 mM RA as the only carbon source (Fig. 5A), consistent with metabolism of this compound and suggesting that metabolism of RA may be responsible for the reduction of its gene induction activity over time. The bacteria did not grow in 10 mM RA. Although our growth data are consistent with those of Annapoorani et al. (21), they conflict with those of Walker et al. (31) who reported that RA is toxic at low micromolar concentrations. Therefore, we examined cell viability as a function of RA concentration. We found that viability was not affected by RA concentrations up to 7.8 mM, whereas viability dropped at 15.6 mM (Fig. 5B). Furthermore, we confirmed that the presence of C4-HSL and RA at the concentrations used for gene expression studies did not change growth kinetics (Fig. 5C). These discrepancies in the RA tolerance of P. aeruginosa PAO1 may be due to differential evolution of the strain in different laboratories.

Fig. 5 The effect of RA on P. aeruginosa PAO1 survival and growth.

(A) Growth curve of P. aeruginosa PAO1 in M9 minimal medium supplemented with the indicated concentration of RA. Shown are means and SD from three experiments conducted in triplicate. (B) Growth in LB medium supplemented with the indicated concentrations of RA (and the corresponding DMSO-containing controls). Cell survival after 24 hours was determined by plating out on solid medium and cell counting. Shown are means and SD from three independent experiments each conducted in quintuplicate. (C) Impact of RA on bacterial growth on rich medium. Shown is a growth curve in LB medium supplemented with different concentrations of C4-HSL or RA. Shown are means and SD from three experiments conducted in triplicate.

The transcriptional reporter studies so far used the rhlI promoter. We performed analogous experiments in P. aeruginosa PAO1 expressing lacZ controlled by lasB (37), rhlA (38), or hcnABC (35), which are all induced by RhlR. RA produced a significant increase in β-galactosidase activity for each of the three reporters (Fig. 6, A to C), whereas control experiments with bacteria transformed with the empty plasmid did not show an increase in β-galactosidase activity upon RA addition (Fig. 6D).

Fig. 6 Effect of RA on the expression from other QS-regulated promoters.

(A to D) β-Galactosidase measurements of P. aeruginosa PAO1 containing the plasmids pβ01 (lasB-lacZ) (A), pβ02 (rhlA-lacZ) (B), pME2823 (hcnA-lacZ) (C), or pQF50 (empty vector) (D). Measurements were made 2 hours after the addition of RA or the corresponding amount of DMSO. Shown are means and SD from three independent experiments conducted in duplicate. **P < 0.01, ***P < 0.001, Student’s t test.

RA increases biofilm formation, pyocyanin production, and elastase synthesis

Increased biofilm formation and the production of the virulence factor pyocyanin are characteristic features of HSL-mediated QS responses (39, 40). We therefore tested these traits in P. aeruginosa grown in the presence and absence of different concentrations of RA or chlorogenic acid. We quantified pyocyanin, which is green, by measuring the absorbance at 520 nM and found that RA, but not chlorogenic acid, stimulated a dose-dependent increase in the intensity of the green color when corrected for cell density (Fig. 7A).

Fig. 7 RA induces QS-regulated phenotypes.

(A) Pyocyanin production measured in P. aeruginosa PAO1 grown in the presence of different concentrations of RA or chlorogenic acid. Shown are means and SD from three independent experiments of the absorbance at 520 nm (pyocyanin) of P. aeruginosa supernatants relative to the absorbance at 660 nm (cell density). The inset shows culture tubes after 8 hours of growth. (The amount of DMSO equivalent to that added in the 2 mM RA condition was added as a control to the first tube “0.”) (B) Biofilm formation quantified in the presence and absence of RA or chlorogenic acid at different time points. Shown are means and SD from three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, Student’s t test. The inset shows representative crystal violet–stained tubes containing an 8-hour culture of P. aeruginosa in the presence and absence of RA or chlorogenic acid. In the lower part, the tubes show the RA concentration dependence of biofilm formation (after 8 hours). (C) Elastase synthesis measured as enzymatic activity of P. aeruginosa PAO1 cultures grown in the presence or absence of RA. Data are normalized for culture density. Shown are means and SD from three individual experiments conducted in triplicate. **P < 0.01, Student’s t test.

Visual inspection of culture tubes (inset in Fig. 7A) showed that RA also stimulated biofilm formation. To quantify the effect of RA on biofilm formation, we grew bacterial cultures in borosilicate glass tubes in the presence or absence of either 2 mM RA or chlorogenic acid and quantified biofilm formation at different time points. We found that RA stimulated biofilm formation between 2 to 8 hours of growth, whereas chlorogenic acid had no significant effect (Fig. 7B, upper). After 24 hours, the amount of biofilm formed in these three conditions was comparable. The RA-mediated stimulation of biofilm formation contradicted a previous report in which the compound was used at higher concentrations than what we used here and inhibited biofilm formation (21). To assess this discrepancy, we determined the dose-response relationship of the RA-mediated effect on biofilm (Fig. 7B, lower). RA stimulated biofilm formation at concentrations up to 2 mM, but above this concentration, biofilm formation was inhibited by RA.

Elastase synthesis is also stimulated by RhlR-mediated QS activity (37). Our reporter analysis indicated that RA enhanced the activity of the promoter controlling the expression of lasB, encoding elastase (Fig. 6A). To verify whether this results in changes in elastase synthesis, we measured elastase activity in P. aeruginosa grown in the absence and presence of RA. RA stimulated the amount of elastase activity in density-normalized cultures (Fig. 7C).


Multiple lines of evidence indicate that plants produce HSL mimics, which interfere with the bacterial QS system (917). However, little information is available regarding the molecular identity of the active compounds that are responsible for this interference (1).

We report the identification and characterization of a plant compound that directly interacts with a bacterial QS regulator, stimulating its transcriptional activity. RA bound purified RhlR with a higher affinity than C4-HSL, and this translated into a greater stimulatory activity of RA on RhlR-mediated transcription and gene expression compared with that induced by C4-HSL. Furthermore, RA stimulated biofilm formation and the synthesis of pyocyanin and elastase, which are phenotypic characteristics that are regulated by QS mechanisms in P. aeruginosa (40, 41).

Because interference with bacterial QS mechanisms by plant-derived compounds has been observed for various different plant pathogens, it was proposed that this represents a plant defense strategy (1). HSL mimics acting as agonists of HSL-mediated sensing may decrease pathogenicity because these mimics would stimulate premature expression of genes encoding proteins involved in QS-controlled functions (1). Molecular identification of HSL mimics will enable their application in medicine and agriculture, for example, in the generation of pathogen-resistant plants. However, is this mimicking of a bacterial QSS by RA of physiological relevance? Walker et al. (31) monitored the consequences of sweet basil root infection by P. aeruginosa strains PAO1 and PA14 and determined that infection with either strain induced RA secretion. RA concentration in root exudates gradually increased, reaching a maximum of ~40 μM 6 days after infection (31). Thus, plant infection triggers RA release, suggesting that RA release forms part of a plant defense strategy and supporting the model that plant-derived HSL mimics decreased pathogenicity by stimulating premature QS-responsive gene expression (1). We found that RA was both a growth substrate and a signaling molecule. Bacterial consumption of RA may provide a mechanism to eliminate the signaling effects of this compound, which is consistent with the kinetics of gene induction by RA that we observed.

Reports of the toxic effects of RA to P. aeruginosa PAO1 differ. Our data and those reported by Annapoorani et al. (21) indicate an effective inhibitory concentration of 2.1 mM with regard to bacterial growth, whereas Walker et al. (31) reported minimal inhibitory concentration of 8 μM. These discrepancies may be due to differential strain evolution in different laboratories. Annapoorani et al. (21) showed a reduction of P. aeruginosa PAO1 biofilm formation and elastase activity in the presence of RA and a number of related compounds, including chlorogenic acid; however, our data showed no activity for chlorogenic acid in binding to or stimulating RhlR or in promoting biofilm formation, and we found that RA functioned as an HSL mimic.

P. aeruginosa is one of the pathogens that infect a wide range of species, including plants and animals (25). Common pathogenic mechanisms enable bacteria to infect phylogenetically different hosts, and there are also parallels in the key features underlying host defense responses in plants, invertebrates, and mammalian hosts (25). The production of HSL mimics is considered a plant defense strategy. The detailed knowledge of plant defense mechanisms may enable the development of strategies to protect human from this pathogen. The effect of HSL mimics, including R, on the virulence properties toward mammals is of interest.

RA is synthesized by many plants (42) and can accumulate to high concentrations (43). RA has multiple biological activities, including antibacterial (31), antiviral (44), antiallergy (45), anticarcinogenic (46), antigenotoxic (47), anti-inflammatory (48), and antioxidant effects (49) activity. In addition, RA was found to be effective against amyloid-β peptide–induced neurotoxicity that is associated with Alzheimer’s disease (50), reduces atopic dermatitis (51), and protects keratinocytes from ultraviolet radiation damage (52). Consequently, plants rich in RA are used as medicinal herbs and by the food industry (43).

Solo or orphan QS regulators recognize plant-derived HSL mimics (1). Solo QS regulators are abundant in plant-associated bacteria, which supports the view that they are involved in interkingdom signaling between plants and bacteria (1). Our data showed that RhlR has a double function and mediates responses to bacterial HSL and plant-derived RA, thus indicating that recognition of HSL mimics is not limited to solo QS regulators.

Several synthetic HSL and non-HSL ligands have been identified that modulate the activity of P. aeruginosa QS regulators (5358). These compounds behaved as either agonists or antagonists. Structurally, the agonists (53, 58) are similar to RA because part of these compounds were linear with aromatic moieties at each extension of the molecule.

Here, we identified RA as an HSL mimic. Indeed, RA mimicked the action of C4-HSL in vitro and in vivo despite being dissimilar structurally. RA is not the only plant-derived HSL mimic. Gao et al. (16) estimated that there are 15 to 20 separable compounds with the capacity to affect HSL-based QS processes. Although RA can serve as a lead compound, this work demonstrated that mimics can have a structure very different from that of the cognate ligand.



The strains and plasmids used in this study are provided in Table 2. HSL and plant-derived compounds (Table 1) were purchased from Sigma-Aldrich.

Table 2 Strains and plasmids used in this study.

Antibiotic resistance: Ap, ampicillin; Gm, gentamicin; Km, kanamycin; Tc, tetracycline; Cb, carbenicillin.

View this table:

Purification of LasR and RhlR

The DNA fragments encoding LasR and RhlR were amplified using the primers 5′-GTTTAAGAAGAACGTGCTAGCATGGCCTTG-3′ and 5′-CTGAGAGGGATCCTCAGAGAGTAATAAGAC-3′ (LasR) and 5′-TATCGAGCTAGCCTTACTGCAATGAGGAATGAC-3′ and 5′-CGAGCTCTGCGCTTCAGATGAGACC-3′ (RhlR), respectively. These primers contained restriction sites (underlined) for Nhe I and Bam HI (LasR) and Nhe I and Sac I (RhlR). Polymerase chain reaction (PCR) products were digested with these enzymes and cloned into the expression plasmid pET28b(+). E. coli BL21 (DE3) was transformed with the resulting plasmids, and cultures were grown in LB medium supplemented with kanamycin (50 μg/ml) at 37°C until an OD660 (optical density at 660 nm) of 0.4. The temperature was then lowered to 18°C, and growth continued until an OD660 of 0.6 to 0.8, at which point protein production was induced by the addition of 0.1 mM isopropyl-β-d-thiogalactopyranoside. Growth was continued at 18°C overnight, and cells were harvested by centrifugation at 10,000g for 30 min. Cell pellets were resuspended in buffer A [20 mM tris-HCl, 0.1 mM EDTA, 500 mM NaCl, 10 mM imidazole, 5 mM β-mercaptoethanol, 5% (v/v) glycerol, 1 mM dithiothreitol (DTT) (pH 7.8)] and broken by French press at 1000 psi. After centrifugation at 20,000g for 1 hour, the supernatant was loaded onto a 5-ml HisTrap column (Amersham Biosciences) previously equilibrated with buffer A. The column was then washed with buffer A containing 45 mM imidazole before protein elution with a linear gradient (20 min) of 45 to 500 mM imidazole in buffer A at a flow of 1 ml/min. Protein-containing fractions were pooled and dialyzed into buffer B [50 mM tris-HCl, 500 mM NaCl, 1 mM DTT (pH 7.8)] and applied to a HiPrep26/60 Sephacryl S-200 high resolution gel filtration column previously equilibrated with the same buffer. Protein was eluted by a constant flow (1 ml/min) of buffer B at 4°C.

Isothermal titration calorimetry

Experiments were conducted on a VP-microcalorimeter (MicroCal) at 25° and 30°C. Proteins were dialyzed in analysis buffer [50 mM K2HPO4/KH2PO4, 150 mM NaCl, 1 mM DTT (pH 7.8)]. HSL ligands were prepared at a concentration of 1 mM in DMSO and subsequently diluted 1:10 with analysis buffer. The corresponding amount of DMSO [10% (v/v)] was added to the dialyzed protein sample. Typically, 8 to 15 μM protein was titrated with 0.1 mM HSL solution. Control experiments involved the titration of dialysis buffer containing 10% (v/v) DMSO with HSL ligand. For the non-HSL ligands, a solution of 660 μM was prepared directly in dialysis buffer and used for the titration of the dialyzed protein. The mean enthalpies measured from the injection of ligands into the buffer were subtracted from raw titration data before data analysis with the MicroCal version of Origin.

Molecular docking, homology modeling, and structural alignment

The atomic structure of LasR was obtained from the PDB (; PDB ID: 3IX3). The structure was refined and optimized with the Protein Preparation Wizard of the Schrödinger Suite (Schrödinger Suite 2012 Protein Preparation Wizard; Epik version 2.3, Schrödinger, LLC, New York, NY, 2012; Impact version 5.8, Schrödinger, LLC, New York, NY, 2012; Prime version 3.1, Schrödinger, LLC, New York, NY, 2012). Ligands were obtained from the Natural Compounds (Metabolites) subset of the ZINC database (59), optimized by LigPrep (LigPrep, version 2.5, Schrödinger, LLC, New York, NY, 2012), and then submitted to virtual screening docking experiments to LasR using the Glide dock SP mode (Glide, version 5.8, Schrödinger, LLC, New York, NY, 2012) (60). The best hits were subsequently docked using the Glide dock XP mode. A homology model of the autoinducer domain of RhlR was generated using Swiss-Model (61) and the structure of QscR (PDB ID: 3SZT) as template. RA was docked onto this structure in the Glide dock XP mode. The structural alignment of the autoinducer domains of LasR and RhlR was generated by PyMOL (PyMOL Molecular Graphics System, version, Schrödinger, LLC).

In vitro transcription assay

A 490-bp DNA fragment of the hcnABC promoter of P. aeruginosa (35) was amplified by PCR using the primers 5′-GCACTGAGTCGGACATGACGGAA-3′ and 5′-CGTGTTGACGTTCAAGAAGGTGCATTGC-3′ and used as a template for these assays. Transcription reactions (20 μl) were performed in binding buffer [20 mM tris-HCl, 50 mM KCl, 1 mM EDTA, 1 mM DTT, 10% (v/v) glycerol (pH 7.8)] containing 50 nM E. coli RNA polymerase holoenzyme saturated with σ70 sigma factor (Epicentre Technologies), 5 nM linear hcnABC DNA, 25 μM RhlR, and different effector molecules (12.5 or 25 μM C4-HSL, RA, and chlorogenic acid). The mixtures were incubated at 30°C for 20 min before the addition of 0.1 mM adenosine triphosphate, cytidine triphosphate, and guanosine triphosphate; 0.05 mM uridine triphosphate (UTP); and 3.6 μCi of [α-32P]UTP (10 μCi/μl) (1 Ci = 37 GBq). After incubation for 50 min, the reactions were stopped by transferring them to a 95°C thermoblock and then subsequently chilled at 4°C, at which point 4 μl of formamide sequencing dye was added. Samples were separated on 6.5% (w/v) polyacrylamide gel electrophoresis gels for 2 hours. Gels were dried and then exposed on a phosphorimager, and the resulting images were processed with the Quantity One software 4.6.2 (Bio-Rad Laboratories). The densitometric analysis was carried out using the program ImageJ (62).

Gene expression studies

Gene expression experiments were conducted with E. coli BL21 harboring pET28b-RhlR (expression plasmid for RhlR) or pET28b (empty plasmid as a control) and pMULTIAHLPROM containing an rhlI::lacZ transcriptional fusion (63), as well as with P. aeruginosa PAO1 lasI/lasR harboring pMULTIAHLPROM. E. coli BL21 were grown in LB containing tetracycline (10 μg/ml) and kanamycin (50 μg/ml), and P. aeruginosa PAO1 lasI/lasR in LB containing tetracycline (40 μg/ml) and gentamicin (20 μg/ml) at 37°C overnight. Stock solutions of RA and chlorogenic acid were prepared in DMSO (100%) and diluted in water to the desired concentration, whereas C4-HSL solutions were prepared in 10% (v/v) DMSO. Fresh LB medium was then inoculated with the resulting cultures (1:100 dilution), grown for 1 hour, and then diluted twofold twice at 30-min intervals to ensure proper dilution of accumulated β-galactosidase after overnight growth. The resulting culture was then grown for another hour before induction with different ligands. To rule out nonspecific effects of DMSO (present in the stock solution), control experiments were performed in which the amount of DMSO corresponding to that present in C4-HSL–, RA-, or chlorogenic acid–containing cultures was added. Growth was continued at 37°C, and samples were taken t different time points for the determination of β-galactosidase in permeabilized whole cells as described in (64). For dose-response experiments, the β-galactosidase activity was measured 4 hours after induction. Data shown are means and SD from at least three independent experiments.

To explore the effect of RA on other RhlR-regulated promoters, plasmids containing lacZ fusions were transferred to the wild-type strain by electroporation. These plasmids were pß01 (lasB::lacZ), pß02 (rhlA::lacZ), pME3823 (hcnA::lacZ), and the insert-free pQF50, which has served to construct the former two plasmids (Table 2). Experimental conditions were as those described above except that cultures were grown for 6 hours after dilution before the induction with different ligands.

Minimal inhibitory concentration assay

These assays were performed in 96-well plates using a modified version of the protocol reported in (65). Wells of a 96-well plate were filled with 200 μl of LB containing different amounts of RA (added using a 250 mM stock solution in DMSO). Control experiments contained the corresponding amounts of DMSO. Wells were inoculated with 10 μl of an overnight culture of P. aeruginosa PAO1 in LB medium. Plate was incubated at 37°C for 24 hours, at which point the viable cell amount was determined by plating out cells on LB agar medium and counting.

Growth experiments

To assess the potential of the bacterium to use RA as sole growth substrate, sterile honeycomb plates (Bioscreen C) containing 200 μl of M9, supplemented with 1 to 10 mM RA, were inoculated with an overnight culture of P. aeruginosa PAO1 grown in M9 minimal medium (66) containing 5 mM citrate at 37°C. Cultures were grown in a Bioscreen C (Thermo Fisher Scientific) instrument under constant shaking at 37°C during which time the OD660 was measured in 1-hour intervals. To assess the effect of different RA concentrations on P. aeruginosa growth, honeycomb plates were filled with LB medium containing 1 to 100 μM C4-HSL or RA, and cultures were carried out as described above.

Biofilm formation

Overnight cultures of P. aeruginosa PAO1 were grown at 37°C and used to inxoculate borosilicate glass tubes containing 2 ml of LB medium (supplemented with either 2 mM RA or chlorogenic acid) to an initial OD660 of 0.05. Both compounds were added as 143 mM solutions in DMSO, and the corresponding control experiments were conducted to assess the effect of the equivalent amount of DMSO on biofilm formation. Cultures were incubated in a Stuart SB3 tube rotator for 2, 4, 6, 8, and 24 hours at 30°C, with an angle of 45° at 40 rpm. Biofilms formed were visualized by crystal violet (0.4%) staining and quantified by solubilizing the dye with 30% acetic acid and measuring the absorbance at 540 nm (67). Data shown are means and SD from three experiments conducted in duplicate.

Quantification of pyocyanin production

Cultures of P. aeruginosa PAO1 were grown in LB at 37°C overnight and used to inoculate glass tubes containing 2 ml of LB medium to an initial OD660 of 0.05. Stock solutions of 143 mM RA or chlorogenic acid were prepared in DMSO, and aliquots were then added to the tubes to final concentrations of 0.5 to 2 mM. The amount of DMSO corresponding to the experiment at 2 mM was added to the control tube. Growth was continued, and pyocyanin production was determined after 8 hours. The OD660 of cultures was determined before centrifugation of cultures at 13,000 rpm for 5 min. The OD520 (indicative of pyocyanin production) was measured, and values were normalized with the cell density (OD660). LB media containing either RA or chlorogenic acid were used as blanks.

Elastolysis assay

Elastase activity in P. aeruginosa PAO1 cultures was determined using a modified version of the elastin–Congo red (ECR) assay (68). Cells were grown with shaking in LB medium at 37°C overnight and used to inoculate glass tubes containing 3 ml of LB medium to an initial OD660 of 0.05. Growth was continued for another 6 hours, and cultures were induced with 500 μM RA. The equivalent amount of DMSO was added to control tubes. Growth was continued until 24 hours, and 1 ml of cell suspension was centrifuged at 13,000 rpm for 15 min. The resulting supernatant was added to tubes containing 10 mg of ECR (Sigma) and 1 ml of buffer [0.1 M tris-HCl, 1 mM CaCl2 (pH 7.0)]. Tubes were incubated at 37°C with shaking (150 rpm) for 24 hours. The reaction was stopped by the addition of 1 ml of sodium phosphate buffer (0.7 M; pH 6.0). Residual, solid ECR was removed by centrifugation, and the OD492 of the supernatant was measured. Shown are means and SD from three replicates conducted in triplicate.


Acknowledgments: We thank J. Kato for providing plasmids pQF50, pβ01, and pβ02; S. Heeb for providing plasmid pME2823; and M. Cámara for plasmid pMULTIAHLPROM and P. aeruginosa PAO1 lasI/lasR. Funding: This work was supported by the FEDER funds and the Fondo Social Europeo through grants from the Junta de Andalucía (grants P09-RNM-4509 and CVI-7335 to T.K. and CVI-7391 to M.E.-U.) and the Spanish Ministry for Economy and Competitiveness (grants BIO2010-16937 and BIO2013-42297 to T.K. and grant BFU2010-17946 to M.E.-U.). Author contributions: A.C.-L., A.D., and A.O. conducted research and analyzed the data, and M.E.-U. and T.K. designed the experiments, interpreted the data, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Signaling

Navigate This Article