Characterization of natural product inhibitors of quorum sensing reveals competitive inhibition of Pseudomonas aeruginosa RhlR by ortho-vanillin

ABSTRACT

Quorum sensing (QS) is a cell-cell signaling system that enables bacteria to coordinate population density-dependent changes in behavior. This chemical communication pathway is mediated by diffusible N-acyl L-homoserine lactone signals and cytoplasmic signal-responsive LuxR-type receptors in Gram-negative bacteria. As many common pathogenic bacteria use QS to regulate virulence, there is significant interest in disrupting QS as a potential therapeutic strategy. Prior studies have implicated the natural products salicylic acid, cinnamaldehyde, and other related benzaldehyde derivatives as inhibitors of QS in the opportunistic pathogen Pseudomonas aeruginosa, yet we lack an understanding of the mechanisms by which these compounds function. Herein, we evaluate the activity of a set of benzaldehyde derivatives using heterologous reporters of the P. aeruginosa LasR and RhlR QS signal receptors. We find that most tested benzaldehyde derivatives can antagonize LasR or RhlR reporter activation at micromolar concentrations, although certain molecules also cause mild growth defects and nonspecific reporter antagonism. Notably, several compounds showed promising RhlR or LasR-specific inhibitory activities over a range of concentrations below that causing toxicity. ortho-Vanillin, a previously untested compound, was the most promising within this set. Competition experiments against the native ligands for LasR and RhlR revealed that ortho-vanillin can interact competitively with RhlR but not with LasR. Overall, these studies expand our understanding of benzaldehyde activities in the LasR and RhlR receptors and reveal potentially promising effects of ortho-vanillin as a small molecule QS modulator against RhlR.

IMPORTANCE

Quorum sensing (QS) regulates many aspects of bacterial pathogenesis and has attracted much interest as a target for anti-virulence therapies over the past 30 years, for example, antagonists of the LasR and RhlR QS receptors in Pseudomonas aeruginosa. Potent and selective QS inhibitors remain relatively scarce. However, natural products have provided a bounty of chemical scaffolds with anti-QS activities, but their molecular mechanisms are poorly characterized. The current study serves to fill this void by examining the activity of an important and wide-spread class of natural product QS modulators, benzaldehydes, and related derivatives, in LasR and RhlR. We demonstrate that ortho-vanillin can act as a competitive inhibitor of RhlR, a receptor that has emerged and may supplant LasR in certain settings as a target for P. aeruginosa QS control. The results and insights provided herein will advance the design of chemical tools to study QS with improved activities and selectivities.

INTRODUCTION

Many bacteria sense and respond to changes in population density using a gene regulation system called quorum sensing (QS). QS can regulate diverse behaviors including light production in marine bioluminescent bacteria, virulence factor production in plant and animal pathogens, and motility in many soil bacteria (1). In Proteobacteria, one type of QS system involves N-acyl L-homoserine lactone (AHL) signals [for reviews, see references (2, 3)]. AHLs are produced by LuxI-type signal synthases and detected by LuxR-type signal receptors, which are cytoplasmic transcriptional factors. At low population densities, AHLs are produced at low levels and accumulate in the local environment with increasing population density. The AHLs diffuse in and out of the cell, although active efflux can also contribute to the export of certain long chain AHLs (4). AHLs bind to the LuxR-type receptor protein, and for most of the known associative-type receptors, when they reach a critical concentration, they cause conformational changes to the protein that enable binding and activation of target gene promoters. AHLs interact with their cognate LuxR protein by making a series of hydrogen-bonding and hydrophobic contacts with residues in the ligand-binding pocket. AHL-binding pockets vary structurally among LuxR family members to ensure specific responses to cognate AHLs, which differ in acyl chain structure.

Pseudomonas aeruginosa is an opportunistic pathogen that can cause debilitating infections in immunocompromised patients and is difficult to treat due to its multi-drug resistance. P. aeruginosa has two LuxI/R-type systems, LasI/R and RhlI/R. The LasI/R system produces and responds to N-(3-oxo)-dodecanoyl L-homoserine lactone (3OC12-HSL), and the RhlI/R system produces and responds to N-butanoyl L-homoserine lactone (C4-HSL). Upon AHL binding, LasR and RhlR activate distinct and overlapping regulons (5, 6). Among those are the genes encoding factors with known roles in virulence, such as the secreted toxins phenazine and hydrogen cyanide, proteases, and biofilm matrix proteins. These systems have been shown to be important for P. aeruginosa virulence in numerous acute animal infection models (7–11). Thus, P. aeruginosa QS has been proposed as an attractive target for the development of novel anti-virulence therapeutics (12).

Over the past 30 years, there has been considerable effort to identify molecules that block QS in P. aeruginosa and other bacteria. These prior studies have identified several promising approaches such as inhibiting LuxI-type synthases (13), destroying or sequestering AHLs (14), or inhibiting LuxR-type receptors (15). The latter strategy has received the most attention to date in P. aeruginosa, with much focus on the LasR receptor, and more recently RhlR, in P. aeruginosa. As a result, several promising molecules have been identified that inhibit these receptors (16–19). These molecules have potencies in the high-nM to mid- to low-μM range. In general, the most potent molecules have been identified as a result of high-throughput screens of small molecule libraries or by making targeted changes to the native AHL or other promising lead compounds via chemical synthesis.

In addition to these synthetic agents, there also has been widespread study of readily available molecules that can be re-purposed as QS inhibitors. Many of these compounds are natural products and were initially identified because of their ability to block QS-dependent phenotypes in the native species, not via studies of their ability to target specific QS pathways. These compounds include halogenated furanones (20), flavonoids such as baicalein (21, 22), and several benzaldehydes such as cinnamaldehyde (23–28). Despite the widespread use of these molecules as chemical tools for studies of QS inhibition, relatively little is known of the specificity, potency, and mechanism of action for most of these compounds. New tools to study QS are of considerable interest, as many of the known chemical modulators have limitations, including relatively low potencies, efficacies, solubilities in aqueous media, and/or chemical stabilities. Consequently, re-purposed bioactive agents and readily available natural products (and analogs) with promising QS inhibitory activities represent a valuable space to search for new chemical probes to study bacterial signaling.

In this study, we used Escherichia coli reporters to evaluate the ability of several naturally occurring benzaldehydes and related derivatives to inhibit the P. aeruginosa QS receptors LasR and RhlR. We focused on compounds reported to disrupt QS-dependent phenotypes in P. aeruginosa, such as cinnamaldehyde and salicylic acid, along with several previously unstudied compounds with some structural similarity, such as orsellinaldehyde and ortho-vanillin (Fig. 1). We observed antagonism of the E. coli LasR and RhlR reporters at concentrations in the mid- to low-μM range, with ortho-vanillin showing the most promising effects. The compounds also caused mild reductions in growth and could nonspecifically antagonize a constitutive reporter at higher concentrations; however, at lower concentrations, there was a suitable window of activity allowing for LasR and RhlR antagonism without any observable toxicity. In follow-up structure-function studies using LasR mutants, we found that critical AHL-binding residues in LasR were not required for ortho-vanillin to antagonize LasR. However, our results support that ortho-vanillin might specifically interact with RhlR. Together, our results indicate that naturally occurring benzaldehydes could have utility in QS inhibition and motivate future studies to develop this chemical scaffold into small-molecule tools to explore LuxR-type protein function and QS pathways.

Fig 1.

Structures of compounds examined in this study. 2 H-4MB, 2-hydroxy-4-methoxybenzaldehyde; 2 H-5MB, 2-hydroxy-5-methylbenzaldehyde.

RESULTS

Construction of cell-based E. coli bioreporters for LasR

To characterize compounds for their potential activity as LasR antagonists, we used heterologous E. coli, which enables LasR to be isolated from other host regulation effects. We used an E. coli strain expressing LasR from an arabinose-inducible promoter (Para-lasR) on plasmid pJN105-L and a second plasmid with the LasR-inducible lasI promoter fused to a promoterless lacZ reporter (PlasI-lacZ) on plasmid pSC11-L (Fig. 2A). In this strain, lacZ expression required LasR and the LasR signal 3OC12-HSL (Fig. 2A), with a half-maximal activation concentration (i.e., EC₅₀ value) of 65 nM. As a control, we also constructed an E. coli strain carrying a plasmid with lacZ expressed from the constitutive aphA-3 promoter (29, 30), pVT19. With this strain, lacZ expression is fully activated in the absence of LasR or 3OC12-HSL (Fig. 2B).

Fig 2.

General schematic and initial characterization of E. coli bioreporter strains used for these studies. (A) An E. coli bioreporter of LasR activity carries plasmid pSC11-L with a LasR-inducible PlasI promoter fused to the lacZ reporter and plasmid pJN105-L with an arabinose-inducible LasR. This strain produces β-galactosidase in response to 3OC12-HSL and 0.4% arabinose. The 3OC12-HSL induction is dose-responsive with a half-maximal concentration of 65 nM. (B) An E. coli strain constitutively expressing lacZ carries the plasmid pVT19 with lacZ fused to the constitutive aphA-3 promoter. This strain produces β-galactosidase in the presence or absence of 3OC12-HSL. Results are the averages of two (A) or three (B) independent experiments, and the error bars represent the SD.

E. coli reporter assays indicate that orsellinaldehyde antagonizes reporter activation nonspecifically

We utilized our E. coli reporters to evaluate the activity of the natural products and related derivatives in Fig. 1 as LasR antagonists, and we initiated our study with orsellinaldehyde, a metabolite produced by the fungus Aspergillus nidulans (31). Given its structural similarity to several known QS inhibitors, such as cinnamaldehyde and salicylic acid, we were interested to examine orsellinaldehyde’s activity as a LasR antagonist. We generated a dose-response curve with orsellinaldehyde in the presence of 100 nM 3OC12-HSL (Fig. 3A; Table 1), a concentration where we observed reproducible results across a wide range of inhibitor compounds. At 100 nM 3OC12-HSL, we found that the concentration of orsellinaldehyde needed to inhibit PlasI-lacZ activity by 50% (i.e., its IC₅₀ value) was 2,370 µM (Fig. 3A, black line), indicating weak antagonist activity toward LasR. However, we observed that orsellinaldehyde caused a dose-dependent reduction of growth yield by about 10%–20% at the highest concentrations (Fig. 3A, gray line). Furthermore, orsellinaldehyde-dependent antagonism of the lasI-lacZ reporter correlated with its increasing effects on growth (correlation coefficient r = 0.9877, P < 0.0001, Fig. 3B). These results suggest antagonism of the LasR bioreporter by orsellinaldehyde may be due to the generalized effects of this compound on growth. To address this possibility, we generated a dose-response curve with orsellinaldehyde and our constitutive lacZ-producing control E. coli strain (with plasmid pVT19). We found that orsellinaldehyde antagonized the constitutive lacZ reporter in a dose-responsive manner with an IC₅₀ of 2,310 µM (Fig. 3A, red line), which was similar to that of the LasR bioreporter (2,370 µM). These results support the conclusion that orsellinaldehyde antagonizes lacZ reporter activation in a nonspecific manner.

Fig 3.

Activity of orsellinaldehyde with E. coli LasR and constitutive reporters. (A) Gray line, orsellinaldehyde-dependent growth inhibition as a percent of cells with no orsellinaldehyde. Dose-response curve of orsellinaldehyde in competition with 100 nM 3OC12-HSL in cultures of E. coli with arabinose-inducible LasR and a LasR-dependent lasI-lacZ reporter (black line) or of E. coli with a constitutive aphA-3-lacZ reporter (red line). IC₅₀ values are given in Table 1. Results show the averages of five (LasR) or three (LacZ control) independent experiments, and the error bars represent the SD. (B) Average values from A (% growth reduction vs % inhibition of the lasI-lacZ reporter) were used to determine Pearson’s correlation coefficient (r value) and significance (P) and generate a fitted line using a simple linear regression model.

TABLE 1.

Potency of benzaldehydes using E. coli LasR and constitutive reporters^a

Compound	IC50 ± CI (µM)b,c
	LasR reporter	Constitutive reporter
Orsellinaldehyde	2,370 (2,280–2,470)	2,310 (2,180–2,450)
Salicylic acid	1,670 (1,500–1,870)	3,650 (3,450–3,860)
Cinnamaldehyde	851 (761–943)	1,460 (1,400–1,530)
ortho-Vanillin	437 (358–523)	1,260 (1,170–1,350)
2-hydroxy-4-methoxybenzaldehyde	1,040 (944–1,140)	1,955 (1,800–2,120)
2-hydroxy-5-methylbenzaldehyde	1,500 (1,300–1,640)	2,020 (1,890–2,160)

^a The E. coli reporter strain for LasR carried plasmid pSC11-L (carrying the lasI-lacZ reporter) and plasmid pJN105-L (expressing LasR from an arabinose-inducible promoter). The E. coli constitutive reporter carried plasmid pVT19 expressing lacZ constitutively from the aphA-3 promoter. Results with both reporters were from experiments carried out in the conditions described for the LasR reporter in the Materials and Methods.

^b Experiments were performed by competing for the compounds at a range of concentrations (25 µM to 50 mM) against 100 nM 3OC12-HSL using conditions described for the LasR reporter in the Materials and Methods. IC₅₀ values determined using a nonlinear best-fit curve with variable parameters with the top and bottom constrained to 100% and 0%, respectively (in all cases the bottom of the computed dose-response curve or maximum inhibition was near 0%). Best-fit curve and IC₅₀ calculations were using Prism v10. Full dose-response curves used to generate these data are shown in Fig. 5; Fig. S2.

^c CI, 95% confidence interval.

To test whether these effects were specific to the lacZ reporter or general to other reporters, we generated an orsellinaldehyde dose-response curve using a strain constitutively expressing GFP (E. coli carrying a constitutive GFP-producing plasmid pUC18T-mini-Tn7T-Gm-gfpmut3). Orsellinaldehyde also antagonized the constitutive GFP reporter with an IC₅₀ of 1,057 µM for GFP (Fig. S1), which was similar to that of the lacZ reporter. These results support the conclusion that the effects of orsellinaldehyde on our LasR bioreporter are related to a generalized effect on gene expression or other cellular processes and not specific to LasR.

Evaluation of other benzaldehyde derivatives in E. coli LasR reporters

We next examined compounds structurally related to orsellinaldehyde and previously reported to modulate QS for antagonistic activity in LasR. In view of the results above, we questioned whether some of the reported inhibitory activities were also largely due to nonspecific toxic effects. We selected several such compounds; cinnamaldehyde (25), salicylic acid (25–28), and the as-yet untested but related compounds ortho-vanillin, 2-hydroxy-5-methylbenzaldehyde, and 2-hydroxy-4-methoxybenzaldehyde (Fig. 1). The results (Fig. 4; Fig. S2; Table 1) show that each of the compounds can antagonize the LasR-dependent lasI-lacZ reporter with IC₅₀s ranging from 437 µM for vanillin to 1,6 µM for salicylic acid. We also observed decreases in growth like that of orsellinaldehyde by ~25% at the highest concentrations (Fig. 4). The effects on growth and inhibition of the lasI-lacZ reporter were significantly correlated for each of the compounds (Fig. 4; Fig. S2; Table 1), although there was a weaker correlation for ortho-vanillin and cinnamaldehyde because the effects on growth were minimal at the lower concentrations (Fig. 4A and C, right side). We also generated dose-response curves of each compound with the control constitutive lacZ reporter strain (Fig. 4). All of these compounds inhibited the LasR-specific reporter at a lower concentration than that of the constitutive reporter by 1.3-fold for 2-hydroxy-5-methylbenzaldehyde to almost threefold for ortho-vanillin. These results suggest that, while all of the compounds also have nonspecific effects at higher concentrations, certain compounds—i.e., ortho-vanillin and cinnamaldehyde—have some specific activity against LasR at lower concentrations.

Fig 4.

Activities of cinnamaldehyde (A), salicylic acid (B), and vanillin (C) with E. coli LasR and constitutive reporters. Left column: dose-response curves for each indicated compound in competition with 100 nM 3OC12-HSL in cultures of E. coli with arabinose-inducible LasR and a LasR-dependent lasI-lacZ reporter (black line) or of E. coli with a constitutive aph-lacZ reporter (red line). Results show the averages of four independent experiments, and the error bars represent the SD. IC₅₀ values from the fit curves are given in Table 1. The right column shows average values from the graphs on the left (% growth reduction vs % inhibition of the lasI-lacZ reporter), which were used to determine Pearson’s correlation coefficient (r value) and significance (P) and generate a fitted line using a simple linear regression model (B) or a second-order polynomial nonlinear regression model (A and C).

Results of LasR mutant reporters support ortho-vanillin not contacting specific residues in the LasR ligand-binding domain

As ortho-vanillin was the most potent LasR antagonist identified above, we sought to further characterize the nature of potential ortho-vanillin/LasR interactions. To our knowledge, no other studies have experimentally addressed the molecular mechanism by which benzaldehyde derivatives antagonize LuxR-type receptors. We began by asking whether ortho-vanillin is acting as a competitive LasR antagonist, similar to the synthetic compound V-06–018 (19) and binding in the native ligand (i.e., 3OC12-HSL) binding site. To this end, we applied an approach of generating a dose-responsive curve for ortho-vanillin competed with varying concentrations of 3OC12-HSL using our LasR reporter assay described above. The ability of ortho-vanillin to antagonize LasR should vary when the 3OC12-HSL concentration is increased if both molecules are competing for binding to the same site in LasR as described for other competitive ligand binding models (32), including for LasR (19, 22). We generated antagonism dose-response curves for ortho-vanillin competed against 3OC12-HSL at 65 nM, 1 µM, and 10 µM (Fig. 5A). Although there was a small difference in the ortho-vanillin IC₅₀ at 1 µM and 10 µM, this difference was not significant (P > 0.07). These results do not support the conclusion that ortho-vanillin is a competitive antagonist of LasR.

Fig 5.

LasR mutant antagonism data for ortho-vanillin and V-06–018. (A) Dose-response curves of ortho-vanillin in competition with 3OC12-HSL at its EC₅₀ value of 65 nM, 1 µM, or 10 µM in E. coli with arabinose-inducible LasR and a LasR-dependent lasI-lacZ reporter. Each curve shows the results of three independent experiments with the SD represented by horizontal bars. (B) Data show averages of IC₅₀ values from each curve shown in panel A. Error bars show the SD. There were no statistical differences between any of the conditions by one-way ANOVA (P > 0.3). (C) The calculated IC₅₀ of ortho-vanillin for each LasR variant is shown as the percent of the wild-type LasR IC₅₀. The ortho-vanillin IC₅₀ values were determined from dose-response curves generated for LasR and each LasR variant at the EC₅₀ determined (from Fig. S4). Full IC₅₀ curves are shown in Fig. S5. There were no significant differences of any of the LasR mutant IC₅₀’s from that of wild type by one-way ANOVA. (D) The LasR inhibitor V-06–018 was tested against 3OC12-HSL as in C. There was insufficient antagonism with LasR S129A to determine an IC₅₀; results with just one of the concentrations (100 µM) is shown here with the full IC₅₀ curves in Fig. S5. Values are reported as the % of reporter activation with the EC₅₀ of 3OC12-HSL with no other compound. *, statistical significance by Student’s t test (P < 0.05).

In addition, we performed in silico docking studies of ortho-vanillin within the ligand-binding domain (LBD) of LasR using the reported full-length LasR structure (PDB ID: 6V7X; see Materials and Methods) and found that this compound could be accommodated. Three residues were identified that could be important for the ortho-vanillin/LasR interaction: Thr75, Thr115, and Ser129 (Fig. S3). These residues were predicted to hydrogen bond with the phenol and aldehyde substituents of ortho-vanillin. Several other residues, such as Tyr56, Trp60, and Tyr93, were also predicted to form close contacts with ortho-vanillin. Ser129 and several other of these residues (e.g., Arg61, Tyr56, and Asp73) were also found to be important for LasR interaction with 3OC12-HSL and other ligands (19, 33, 34) (Fig. S3).

To examine these putative interactions between ortho-vanillin and LasR, we determined the activity of ortho-vanillin in several LasR mutants. In prior studies in our laboratories, a set of LasR mutants were generated in which residues within the ligand-binding pocket were mutated to a different residue of similar steric size but without the capability to hydrogen bond (e.g., Tyr → Phe). These mutants were introduced into E. coli to generate lasI-lacZ reporters analogous to the wild-type LasR reporter above (see Table S1 and Materials and Methods). From that set, we selected five LasR mutant reporters to test ortho-vanillin (W60F, Y56F, T75V, Y93F, and S129A), which included the Thr75 and Ser129 residues predicted to be important for ortho-vanillin interaction in our in silico study. Each of these mutants showed varying degrees of activation by 3OC12-HSL in our reporter experiment (Fig. S4), which was consistent with prior results (34). We generated dose-response curves with ortho-vanillin competed against 3OC12-HSL at the concentration needed to cause half-maximal LasR activation (EC₅₀) for each mutant (Fig. S4). In our experiments, ortho-vanillin antagonism of the LasR mutants was indistinguishable from that of the wild-type LasR (Fig. 5C; Fig. S5). As a control, we also tested the ability of V-06–018 to antagonize the LasR S129A mutant. We used the same approach to test the ability of V-05–018 to antagonize LasR and the LasR S129A mutant. Consistent with prior results (19), V-06–018 was significantly less active with the S129A mutant compared with wild-type LasR (Fig. 5D; Fig. S5) so that an IC₅₀ could not be determined for the S129A mutant. Results from one of the concentrations (100 µM) are shown instead. Together, these results show that some of the LasR ligand-binding site residues that make important contacts with other ligands (AHL and non-AHL agonists or antagonists) are not required for ortho-vanillin activity and are consistent with the idea that ortho-vanillin does not interact with the LasR ligand-binding domain in a mode analogous to other ligands.

Evaluation of benzaldehyde derivatives using E. coli RhlR reporters indicates that ortho-vanillin can antagonize RhlR

We hypothesized that the benzaldehyde derivatives in our studies (Fig. 1) could be poor antagonists of LasR because they have very short or no acyl tail functionality, which has been shown to be important for LasR interactions in studies of the native ligand 3OC12-HSL and other inhibitors, such as V-06–018 (19, 35). We thus turned our attention to RhlR from P. aeruginosa, which is regulated by an AHL with a much shorter 4-carbon tail, C4-HSL. We performed in silico docking studies analogous to those for LasR above using the recently published RhlR structure, which was purified with a non-native agonist meta-bromothiolactone (PDB ID: 8DQ0) (36). We examined docking of ortho-vanillin and the native ligand C4-HSL to RhlR (Fig. S6) and found that both could be accommodated with similar docking scores (about −5.2 kcal/mol). The phenol moiety of ortho-vanillin was predicted to hydrogen bond with Asp81 of RhlR, supporting the idea that this compound could possibly interact with RhlR. Interestingly, Asp81 was previously shown to be important for RhlR activation by C4-HSL (37), supporting that the specific interactions of ortho-vanillin and C4-HSL with RhlR may be similar.

To test the ability of our set of benzaldehydes and related compounds to antagonize RhlR, we generated dose-response curves with these compounds using an E. coli RhlR reporter strain (Fig. 6; Table 2). This strain is analogous to the LasR reporter above, but it carries plasmid pECP61.5 expressing RhlR from the isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible Plac promoter as well as the rhlA-lacZ reporter (38). We also utilized our constitutive lacZ reporter plasmid pVT19 to generate dose-response curves using the RhlR assay conditions (Fig. 6; Table 2). The tested compounds caused a maximal ~20% growth reduction for the RhlR conditions. The potencies of our compounds with the RhlR reporter ranged from an IC₅₀ of 151 µM for ortho-vanillin to ~10 mM for salicylic acid. With the constitutive lacZ reporter, there was a similar spread in potencies for our compounds, with ortho-vanillin having the lowest IC₅₀ and salicylic acid as the highest. However, the IC₅₀ for ortho-vanillin was fivefold lower with RhlR than with the constitutive lacZ reporter. There were also no observed effects of ortho-vanillin on growth until concentrations at which there was >50% antagonism of the RhlR reporter, and there was no antagonism of the constitutive lacZ reporter until concentrations >500 µM. These results support the idea that ortho-vanillin may specifically antagonize RhlR at concentrations below 500 µM.

Fig 6.

Activities of orsellinaldehyde (A) and ortho-vanillin (B) with E. coli RhlR and constitutive reporters. Left column: dose-response curves for each indicated compound in competition with 414 nM C4-HSL in cultures of E. coli with IPTG-inducible RhlR and an RhlR-dependent rhlA-lacZ reporter (black symbols) or of E. coli with a constitutive aph-lacZ reporter (red symbols). Results show the averages of three independent experiments, and the error bars represent the SD. IC₅₀ values from the fit curves are given in Table 2. The right column shows average values from the graphs on the left (% growth reduction vs % rhlA-lacZ reporter antagonism), which were used to determine Pearson’s correlation coefficient (r value) and significance (P) and generate fitted lines using a second-order polynomial nonlinear regression model.

TABLE 2.

Potency of benzaldehyde derivatives using E. coli RhlR and constitutive reporters^a

Compound	IC50 ± CI (µM)b,c
	RhlR reporter	Constitutive reporter
Orsellinaldehyde	956 (826–1,060)	2,750 (2,470–3,100)
Salicylic acid	10,050 (8,670–11,700)	6,260 (4,840–8,300)
Cinnamaldehyde	1,310 (1,200–1,430)	2,850 (2,640–3,090)
ortho-Vanillin	151 (136 to 168)	872 (749–?)d
2-hydroxy-4-methoxybenzaldehyde	458 (409–511)	1,360 (1,190–1,550)
2-hydroxy-5-methylbenzaldehyde	1,082 (974–1,210)	1,789 (1,259–2,775)

^a The E. coli reporter strain for RhlR carried plasmid pECP61.5 (carrying the inducible Ptac-rhlR and the rhlA-lacZ reporter). The E. coli constitutive reporter strain carried plasmid pVT19 expressing lacZ constitutively from the aph promoter. Results with both reporters were from experiments carried out in the conditions described for the RhlR reporter in the Materials and Methods.

^b Experiments were performed by competing for the compounds at a range of concentrations (25 µM to 50 mM) against 400 nM C4-HSL using conditions described for the RhlR reporter in the Materials and Methods. IC₅₀ values were determined using a nonlinear best-fit curve with variable parameters with the top and bottom constrained to 100% and 0%, respectively (in all cases the bottom of the computed dose-response curve or maximum inhibition was near 0%). Best-fit curve and IC₅₀ calculations were using Prism v10. Full dose-response curves used to generate these data are shown in Fig. 6; Fig. S7.

^c CI, 95% confidence interval.

^d CI and slope could not be calculated with a variable parameter model. With a Hill slope set at 1.0, the calculated IC₅₀ was 774 with a CI of 531–1,108.

RhlR reporter data support a competitive mechanism of RhlR antagonism by ortho-vanillin

We were interested to determine whether ortho-vanillin was acting as a competitive RhlR antagonist. As with LasR, we tested whether competing with C4-HSL at different concentrations could elicit changes in the ability of ortho-vanillin to antagonize RhlR in the E. coli lacZ reporter. We generated antagonism dose-response curves for ortho-vanillin competed against C4-HSL at 400 nM, 10 µM, and 100 µM (Fig. 7). We observed a significant C4-HSL concentration-dependent decrease in the potency of ortho-vanillin. These differences were most apparent at the lowest concentrations of ortho-vanillin, which were below the concentration at which nonspecific antagonism of the lacZ reporter was observed. These results are congruent with the ability of ortho-vanillin can act as a competitive antagonist of RhlR.

Fig 7.

Dose-response RhlR antagonism by ortho-vanillin competed with varying C4-HSL concentrations. (A) Dose-response curves of ortho-vanillin in competition with 400 nM, 10 µM, or 100 µM C4-HSL in cultures of E. coli with IPTG-inducible RhlR and an RhlR-dependent rhlA-lacZ reporter. Each curve shows results of three independent experiments with the SD represented by horizontal bars. (B) Data show averages of IC₅₀ values from each curve shown in panel A. Error bars show the SD. Statistical significance by one-way ANOVA; *, P < 0.05; **P < 0.01; ***P < 0.001.

DISCUSSION

The contribution of QS to a wide array of phenotypes, including virulence, in P. aeruginosa has attracted significant attention to the identification of QS inhibitors for use as chemical probes and in therapeutic development. Despite considerable work in this area, there are relatively few highly potent and selective QS inhibitors in P. aeruginosa and related Proteobacteria. Most of these compounds target LuxR-type receptor proteins, including V-06–018 which antagonizes LasR in P. aeruginosa (18) and the chlorolactone AHL analog (CL) that antagonizes CviR from Chromobacterium violaceum (39, 40). Beyond these classes of synthetic compounds, there are many naturally derived compounds or extracts that have reported activities as QS inhibitors in bacteria. For example, salicylic acid can downregulate the production of the QS-controlled virulence factors pyocyanin and elastase and attenuate the ability of P. aeruginosa to infect plants (28). However, detailed studies to determine the molecular mechanisms by which these natural products elicit their effects on QS are limited. In this study, we evaluate the ability of salicylic acid, cinnamaldehyde, and several related benzaldehyde derivatives to antagonize the P. aeruginosa LuxR-type receptors LasR and RhlR using heterologous reporters in E. coli. We provide evidence that one of these compounds, namely ortho-vanillin, can specifically antagonize these receptors within a lower range of concentrations in which they are not generally toxic. These results provide a basis to guide the use of these compounds in QS studies and suggest chemical scaffolds to advance the design of new QS receptor antagonists.

The investigations described here indicate that ortho-vanillin can specifically antagonize LasR, and it does so through a non-competitive mechanism that is independent of some of the critical ligand interaction residues in the ligand-binding domain of this protein (Fig. 5). There are prior reports of other compounds that might inhibit LuxR-type receptors noncompetitively. Halogenated furanones, such as bromofuranone, have been shown to inhibit the Vibrio fischeri LuxR receptor noncompetitively (41). Inhibition might involve a mechanism of increasing the turnover of the receptor protein in the cell (41), although bromofuranone can also be broadly toxic at inhibitory concentrations (42). Some flavonoids also have been reported to inhibit LasR noncompetitively, such as baicalein, although in the case of baicalein the mechanism is not known (22). Our discovery that ortho-vanillin can antagonize LasR noncompetitively adds to this list of noncompetitive antagonists.

In the case of RhlR, ortho-vanillin appears to act as a specific, competitive antagonist in the E. coli reporter (Fig. 7). Competitive inhibition is by far the most invoked mechanism for known LuxR-type inhibitors; the crystal structure of CL bound to CviR and stabilizing an inactive conformation provides perhaps the most compelling support for this mechanism (40). There are several other known competitive inhibitors of RhlR, most of which closely resemble its native ligand C4-HSL, and our prior detailed structure-function studies have revealed portions of the molecules that are essential for strong inhibitory activity (43). With the recently determined crystal structure of RhlR (36), it is now possible to carry out more detailed studies to better understand RhlR-ligand binding interactions, including with the native ligand C4-HSL. Such studies will be interesting to reveal important insight into the mechanism of RhlR-ligand interactions and advance the design of compounds that can modulate RhlR activity.

Our results with E. coli reporters show that ortho-vanillin is more potent against RhlR than LasR. Other compounds, such as salicylic acid and cinnamaldehyde, were less potent against RhlR compared with LasR, supporting that the difference in potency with ortho-vanillin was specific to this compound and not due to broader differences in RhlR vs LasR activities or conditions of each reporter assay. This difference could be due to the relatively small size of this molecule and/or its lack of an acyl tail. The natural ligand of LasR, 3OC12-HSL, has a long 12-carbon acyl tail, whereas the RhlR ligand C4-HSL has a much shorter 4-carbon acyl tail. Prior structure-function studies of LasR and 3OC12-HSL reveal that there are important hydrophobic contacts formed between the long tail of 3OC12-HSL and residues within the LasR-binding pocket (44). These contacts contribute to the strength and specificity of the interaction with LasR. In addition, studies with V-06–018 analogs showed that shorter acyl tails weaken LasR interactions (45). In turn, we have shown that RhlR is both activated and inhibited by AHLs analogs with shorter tails. ortho-Vanillin largely lacks such a hydrophobic tail (Fig. 1), which might weaken its ability to antagonize LasR while enhancing its ability to engage with RhlR. Our results support the idea that the hydrophobic tails of ligands play a critical role in the specificity and strength of interactions with LuxR proteins. As this competitive activity for ortho-vanillin in RhlR, and its non-competitive activity in LasR, were observed in E. coli reporter systems, additional experiments, including in vitro studies, will be necessary to provide further clarity into its molecular mechanisms of action and the hypotheses outlined here.

Our results suggest that ortho-vanillin might have the utility to antagonize RhlR in P. aeruginosa. RhlR is an important regulator of virulence in clinical P. aeruginosa strains (46), warranting significant recent interest in identifying RhlR inhibitor molecules. RhlR activity also can be modulated by protein-protein interactions with another protein, PqsE (37, 47), suggesting that RhlR activity in P. aeruginosa could look different than that in E. coli. There may also be nonspecific effects on P. aeruginosa growth, as observed in E. coli, and/or other barriers, such as membrane permeability, active export, or metabolism of this compound that could impact its activity. A careful evaluation of the activities of ortho-vanillin and related compounds in P. aeruginosa is warranted as the next step to evaluate the potential utility of ortho-vanillin as a RhlR antagonist; these studies are ongoing in our laboratory and will be reported in due course. The relative simplicity of the ortho-vanillin scaffold suggests straightforward routes to alter its structure and examine its impact on potency and specificity, along with reducing any associated toxicity. Such an approach could lead to improved molecules for biological studies of P. aeruginosa QS. Overall, these studies illustrate the importance of performing rigorous studies to determine the specificity and function of small molecule QS inhibitors to inform their use as research tools and other applications.

MATERIALS AND METHODS

Culture conditions and reagents

Unless otherwise noted, bacteria were grown at 37°C in lysogeny broth (LB; 10 g tryptone, 5 g tryptone, and 5 g NaCl per L), or on LB agar [1.5% (weight per volume) Bacto-Agar]. For RhlR bioreporter experiments, growth was at 30°C and in A medium (48) [60 mM K₂HPO₄, 33 mM KH₂PO₄, 7.5 mM (NH₄)2SO₄, 1.7 mM sodium citrate ⋅2H₂O, 0.4% glucose, 0.05% yeast extract, and 1 mM MgSO₄]. All E. coli broth cultures were grown with shaking at 250 rpm, 18 mm test tubes (for 5 mL cultures), or 125 mL baffled flasks (for 10 mL cultures) unless otherwise specified. For selection, 100 µg mL⁻¹ ampicillin, 10 µg mL⁻¹ gentamicin, or 150 µg mL⁻¹ spectinomycin were used. For experiments with the RhlR bioreporter strain, A medium was used as described (38, 49). When needed for induction of LasR or RhlR, we added IPTG at 1 µM final concentration and L-(+)-arabinose at 0.25% final concentration. Native HSLs were suspended in ethyl acetate acidified with 0.01% glacial acetic acid and added to culture tubes and dried down prior to adding growth medium for experiments.

We measured β-galactosidase activity with a Tropix Galacto-Light Plus chemiluminescence kit according to the manufacturer’s protocol (Applied Biosystems, Foster City, CA). Native HSLs (3oxoC12-HSL and C4-HSL) were purchased from Cayman Chemicals (MI, USA), gentamicin was purchased from GoldBio (MO, USA), and ampicillin and spectinomycin were purchased from Sigma Aldrich (MO, USA). Dimethyl sulfoxide (DMSO, solvent for inhibitor compounds), IPTG, and L-(+)-arabinose were purchased from Fisher Scientific (PA, USA). Natural products and benzaldehyde derivatives were purchased from Sigma Aldrich (MO, USA). V-06–018 was synthesized as previously described (19).

Strains and plasmids

Strains and plasmids are listed in Table S1. To assess LasR activation of lasR expression in recombinant E. coli, we used E. coli strain DH5α carrying two plasmids; plasmid pJN105-L (50) with an arabinose-inducible P. aeruginosa lasR and plasmid pSC11-L (51) with the promoter of the LasR-responsive gene lasI fused to a lacZ reporter. For some studies, pJN105-L was replaced with derivatives of this plasmid encoding LasR mutants with single amino acid substitutions (33). To assess RhlR activation of rhlR expression in recombinant E. coli, we used E. coli DH5α with plasmid pECP61.5 (38) with an IPTG-inducible ptac-rhlR and an RhlR-responsive gene rhlA fused to the lacZ reporter. For constitutive expression of the lacZ reporter, we used E. coli DH5α with plasmid pVT19, which has the lacZ gene fused to the constitutive aphA-3 promoter. To construct pVT19, the constitutive aphA-3 promoter (29, 30) was amplified from a pTCV-lac derivative using primers Vlac1 and Vlac2 (30). The resulting amplicon was digested with EcoRI and BamHI and ligated into similarly digested pKS12A (52). The resulting plasmid with the aphA-3 promoter transcriptionally fused to lacZ was designated pVT19. For constitutive expression of the gfp reporter, we used E. coli DH5α with plasmid pUC18T-mini-Tn7T-Gm-gfpmut3 (53).

Transcription reporter assays in E. coli

To assess LasR activation of lasR expression in recombinant E. coli, overnight cultures of E. coli DH5α pSC11-L and pJN105-L were diluted 1:100 into LB containing selection antibiotics gentamicin and ampicillin in 10 mL cultures. When the cultures reached an OD₆₀₀ of 0.2–0.3, L-(+)-arabinose was added to a final concentration of 0.25%. The control did not receive L-(+)-arabinose. The cultures were then grown to an OD₆₀₀ of 0.5–0.6, and 500 µL was added to 1.5 mL micro centrifuge tubes containing dried 3OC12-HSL. Aliquots (5 µL) of increasing concentrations of inhibitor test compound stock solution in DMSO were then added to the designated micro centrifuge tubes containing culture. Tubes containing E. coli with just signal and DMSO were included as controls. After 3 h at 37°C with shaking, OD₆₀₀ was measured using a plate reader, and β-galactosidase activity was measured as described above.

To assess RhlR activation of rhlR expression in recombinant E. coli, overnight cultures of E. coli DH5α pECP61.5 grown at 30°C in A medium containing antibiotic selection (ampicillin) and IPTG to induce RhlR expression were diluted to an OD₆₀₀ of 0.1, and 1 mL was added to culture tubes containing dried C4-HSL. Aliquots (5 µL) of DMSO containing increasing concentrations of inhibitor test compound or DMSO with no test compound were added to the designated Eppendorf tubes containing culture. Tubes containing E. coli with signal and DMSO were included as vehicle control. After 5 h at 30°C with shaking, OD₆₀₀ was measured using a plate reader, and β-galactosidase activity was measured as described above.

Experiments with the LasR mutants and the constitutive lacZ expression plasmid pVT19 or constitutive gfp expression plasmid pUC18T-mini-Tn7T-Gm-gfpmut3 were carried out identically as described above for the LasR or RhlR bioreporter experiments. Results with the pBT19 constitutive reporter strain were different for the LasR vs RhlR bioreporter protocols likely due to differences in growth conditions (temperature and/or growth media).

Computational modeling

The structure of LasR (PDB ID: 6V7X) and RhlR (PDB ID: 8DQ0) was used for docking studies using the Lamarckian protocol and the empirical-free energy function in AutoDock version 4.2. The hit search was refined using an improved docking method. The α-β-α sandwich located near the N-terminal LBD was used as the binding location for docking calculations. The protein target was prepared using AutoDock 4.2. Hydrogen atoms were added, and the water molecules were removed using the AutoDock Tools (ADT) module included in AutoDock. Charges were adjusted using AutoDock’s Gasteiger charges module for proteins, and atom type was modified to ADT type for calculations. In our calculations, we dock the ligand (natural or ortho-vanillin) with ligand-free LasR or RhlR. For each type of atom in the ligand being docked, AutoDock needs a pre-calculated grid map. These maps are calculated using AutoGrid. The Gasteiger-Marsili method was used to determine the atomic charges of the protein. The AutoGrid application created mass-centered grid maps with 80 grid points in each direction and 0.375 spacing. Ten different docking runs for the ligand were carried out, followed by the evaluation of docking results for the binding mechanism and conserved interactions, such as hydrogen bonds and hydrophobic interactions, between the hits and the LasR or RhlR binding site. The common interactions of the ligand-docked complexes were analyzed, and the one with the best binding score based on the binding-free energy was reported.

Statistical analyses

All statistical analyses (one-way ANOVA and Student’s t test) were done using Prism v10. IC₅₀ and EC₅₀ curves were fitted using a nonlinear regression model with a variable slope unless otherwise stated.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/spectrum.00681-24.

Supplemental figures and table. spectrum.00681-24-s0001.docx.

Fig. S1-S7; Table S1.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.