Targeting Bacterial Communication: Evaluating Phytochemicals as LuxS Inhibitors to Disrupt Quorum Sensing

Abstract

Bacterial quorum sensing (QS) is a critical communication process that regulates gene expression in response to population density, influencing activities such as biofilm formation, virulence, and antibiotic resistance. This study investigates the inhibitory effects of five phytochemicals—apigenin, carnosol, chlorogenic acid, quercetin, and rosmarinic acid—on the S-ribosylhomocysteinase (LuxS) enzyme, a key player in AI-2 signaling across both Gram-positive and Gram-negative bacteria. Using molecular docking studies, we identified that these phytochemicals interact with the LuxS enzyme, with apigenin, carnosol, chlorogenic acid, and rosmarinic acid binding within the substrate-binding pocket and exhibiting binding scores below −7.0 kcal/mol. Subsequent in vitro assays demonstrated that these compounds inhibited AI-2 signaling and biofilm formation in Escherichia coli MG1655 in a concentration-dependent manner. Notably, carnosol and chlorogenic acid showed the most potent effects, with IC₅₀ values of approximately 60 μM. These findings suggest that these phytochemicals may serve as potential QS inhibitors, providing a foundation for developing new anti-pathogenic agents to combat bacterial infections without promoting antibiotic resistance. Further studies are warranted to explore the therapeutic applications of these compounds in both clinical and agricultural settings.

1. Introduction

Quorum sensing (QS) is a bacterial communication mechanism that involves the production, detection, and response to extracellular signaling molecules known as auto-inducers (AIs) [1]. This process enables bacterial populations to coordinate behavior changes synchronously in response to shifts in density and species composition within their community. QS orchestrates gene expression programs that underpin collective behaviors, such as the expression of virulence factors in pathogenic bacteria [2]. By modulating gene expression based on cell density, bacteria can efficiently undertake energy-intensive activities, maximizing their impact on the environment or host. This collective capability ensures that energetically costly processes are initiated only when they are most likely to be effective [3]. Both Gram-positive (G⁺) and Gram-negative (G⁻) bacteria utilize QS to regulate gene expression and coordinate collective behaviors. However, they employ different mechanisms and signaling molecules for this process [4].

G⁺ bacteria use small, post-translationally modified peptides called autoinducing peptides (AIPs) as signaling molecules [5]. These AIPs are often part of a histidine kinase two-component signal transduction system. In this system, secreted precursor AIPs are processed into mature AIPs by extracellular proteases [6]. These mature AIPs can be imported back into the cell and identified by cytoplasmic transcription factors. Typically, AIPs bind to membrane-bound sensor kinase receptors, which autophosphorylate and pass the phosphate to a cognate cytoplasmic response regulator. The phosphorylated response regulator then activates the transcription of genes in the QS regulon. Additionally, G⁺ bacteria secrete various peptides, including pheromones, that regulate specific genes involved in processes such as virulence factor biosynthesis, bacterial competence, and bacterial conjugation [7,8].

In contrast, G⁻ bacteria use small molecules as autoinducers (AIs), primarily acyl-homoserine lactones (AHLs) or other molecules synthesized from S-adenosylmethionine (SAM) [3]. AIs are produced within the cell and diffuse freely across bacterial membranes. When AI concentration is high, typically at a high cell density (HCD), they bind to specific receptors located either in the inner membrane or cytoplasm. These receptors are often transcription factors that regulate the expression of genes in the QS regulon [9]. In some cases, AIs are detected by two-component histidine kinase receptors similar to those in G⁺ bacteria [10]. A common feature in G⁻ quorum sensing is autoinduction, where the AI-driven activation of QS leads to increased AI synthesis, creating a feed-forward loop that promotes synchronous gene expression across the population [3].

The only QS mechanism that is employed by both G⁺ and G⁻ bacteria is the production of autoinducer-2 (AI-2) signaling molecules by the enzyme S-ribosylhomocysteinase (LuxS) (EC 4.4.1.21), which catalyzes the cleavage of S-ribosylhomocysteine (SRH) into homocysteine and 4,5-dihydroxy-2,3-pentanedione (DPD) [11]. This system enables collective decision making among bacterial populations regarding gene expression through the secretion and detection of AI-2 [12]. In Vibrio harveyi, AI-2 is identified as a furanosyl-borate-diester, whereas in Escherichia coli, it appears as a furan molecule [13]. The synthesis of AI-2 involves converting S-adenosylhomocysteine (SAH) into homocysteine through either a single-step reaction using SAH hydrolase (SahH) or a two-step process involving SAH nucleosidase (Pfs) and LuxS, which specifically cleaves the thioether linkage of SRH to produce 4,5-dihydroxy-2,3-pentanedione (DPD). DPD can then rearrange to form R- or S-2-methyl-2,3,3,4-tetrahydroxytetrahydrofuran (R- or S-THMF), collectively known as AI-2 [14,15]. The ability of different bacterial species to sense the AI-2 produced by others suggests its role as a potential universal signal for interspecies communication [16].

Furthermore, LuxS is prevalent in a diverse array of both G⁺ and G⁻ and is integral not only to the production of AI-2 but also to critical metabolic functions. It catalyzes the hydrolysis of S-adenosyl homocysteine into S-adenosylmethionine (SAMe), a key methyl donor. This activity supports the methyl cycle, vital for the synthesis of vitamins and polyamines within bacterial central metabolism [17]. The presence of the LuxS/AI-2 system underscores its importance as a global regulatory network, responding to bacterial population changes and significantly influencing cellular behavior [18]. Given the pivotal role of LuxS in this universal signaling system, targeting it with specific inhibitors can profoundly impact bacterial communication and pathogenicity. Thus, understanding the structural and functional nuances of LuxS is paramount given its universal role in QS across diverse bacterial species, leading directly into our investigations of specific inhibitors [19].

The crystal structures of LuxS homologues [20,21,22] reveal a homodimer architecture that is distinctive to a specific group of enzymes with related functions. These enzymes, including LuxS, typically engage in metabolic processes related to the handling of SAMe-derived molecules, playing roles in various bacterial signaling and metabolic pathways. This structural setup is particularly suited to their function in cleaving thioether bonds, a critical step in producing signaling molecules like AI-2, which is fundamental for bacterial communication. Each enzyme forms two identical active sites at the interface of the dimer. These sites incorporate residues from both subunits within a conserved HXXEH motif, where an Fe²⁺ ion is tetrahedrally coordinated. The enzyme’s catalytic activity, involving a series of proton-transfer reactions facilitated by the Fe²⁺ ion and two residues from LuxS, leads to the predicted chemical transformations [22]. These structural and functional characteristics confirm that LuxS is the enzyme previously identified as the ribosylhomocysteine-cleavage enzyme [22,23], which is known to catalyze the formation of DPD in Escherichia coli [24]. Examples of LuxS include but are not limited to that of Haemophilus influenzae (UniProt: P44007), Salmonella typhi (UniProt: Q8Z4D7), Vibrio cholerae (UniProt: Q9KUG4), Campylobacter concisus (UniProt: A7ZGE2), and Neisseria meningitidis (UniProt: A9M0P7).

QS is a pivotal mechanism in bacteria that regulates activities such as biofilm formation and virulence factor production. Notably, recent research indicates that QS systems may play a role in this resistance [25,26]. Globally, the rise of antibiotic-resistant bacterial pathogens poses a significant health challenge [27]. Targeting QS offers a novel antibacterial approach that not only prevents resistance development but also suppresses virulence factors dependent on bacterial population density [28]. By disrupting QS, this strategy aims to diminish bacterial virulence and disease-causing potential without promoting the emergence of resistance, as it does not exert selective growth pressure on bacteria [29]. Furthermore, combining anti-QS strategies with traditional antibiotics presents a promising avenue to tackle bacterial infections and the burgeoning problem of antibiotic resistance [30]. Anti-QS strategies encompass a wide array of approaches, including natural compounds, antibody-mediated quorum quenching (QQ), computer-aided drug design for QQ, repurposing FDA-approved drugs as anti-QS agents, and the modulation of quorum-sensing molecules [30]. Notably, phytochemicals from aromatic and medicinal plants have recently been recognized for their potential as QS inhibitors, with several compounds already identified as effective in disrupting QS [31,32].

We have recently explored the anti-QS properties of ethanolic extracts from eight aromatic plants native to Cyprus: Origanum vulgare subsp. hirtum (oregano), Rosmarinus officinalis (rosemary), Salvia officinalis (common sage), Lavandula spp. (lavender), Calendula officinalis (calendula), Melissa officinalis (lemon balm), Sideritis cypria (Cypriot mountain tea), and Aloysia citriodora (lemon verbena) [33]. Our findings revealed that the extracts from oregano, rosemary, and common sage exhibited the most potent inhibitory effects on AI-2 signaling. These extracts also significantly reduced biofilm formation by more than 60% and inhibited both swimming and swarming motilities of E. coli MG1655 in a concentration-dependent manner, distinguishing them as effective anti-QS agents. These activities were achieved without affecting bacterial growth, suggesting a mechanism of action that targets QS pathways rather than bacterial viability. The remaining plant extracts demonstrated varying degrees of inhibitory activity. Furthermore, we identified five key phytochemicals that were the major components in the ethanolic extracts that could be responsible for these effects: carnosol, chlorogenic acid, and quercetin were common to all three plants, while apigenin and rosmarinic acid were uniquely detected in the oregano extract. Given the critical role of the LuxS/AI-2 system in QS, we extend our investigation herein to determine the specific anti-QS activities of each identified phytochemical and their potential for selectively inhibiting the LuxS protein. By elucidating the specific interactions between these phytochemicals and the LuxS enzyme, we anticipate not only advancing our understanding of QS modulation but also paving the way for novel anti-QS therapies that are less likely to contribute to resistance. This ongoing research not only contributes to our understanding of plant-based QS inhibitors but also enhances the potential for developing novel therapeutic agents to combat infections with QS-dependent pathogens, particularly those resistant to conventional antibiotics.

2. Materials and Methods

2.1. LuxS Structure Prediction

Since the experimental (crystallographic) 3D structure of the LuxS protein from E. coli MG1655 (UniProt ID: P45578) is not available in the Protein Data Bank (PDB), we utilized the AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk/entry/P45578, accessed on 1 August 2024) to obtain the predicted structure of the protein [34]. AlphaFold provides a per-residue model confidence score (pLDDT) ranging from 0 to 100, which indicates the confidence level of each residue’s positional accuracy. The model confidence for E. coli LuxS is very high, as reflected by the pLDDT (>90) score. However, the predicted protein models in the AlphaFold database lack coordinates for small molecules essential for molecular structure or function. Specifically, in the case of LuxS, the Fe²⁺ ion is missing. To address this, we utilized AlphaFill (https://alphafill.eu/; accessed on 1 August 2024) [35] to add the missing ion using a known PDB complex template (PDB ID: 5E68) and optimizing it with the YASARA feature (Supplementary Figure S1). As the template protein features a Zn²⁺ ion at its active site, the model predicted by AlphaFill also initially featured the same ion. However, since E. coli LuxS contains Fe²⁺ as its metal ion, we addressed this issue by replacing the Zn²⁺ with a Fe²⁺ ion using ChimeraX software (UCSF ChimeraX v1.18, Resource for Biocomputing, Visualization, and Informatics, University of California, San Francisco, CA, USA).

Subsequently, we used the HDock server (http://hdock.phys.hust.edu.cn/, accessed on 1 August 2024) [36] to build the protein dimer. Among the generated models, “Model 0” was selected for further analysis (Supplementary Figure S2). This model was constructed using a known PDB complex template (PDB ID: 5E68), which HDock indicated as having high confidence. Due to its foundation on a validated PDB structure and the absence of the docking and confidence scores typically calculated for de novo models, “Model 0” was deemed the most reliable and was therefore utilized in all subsequent docking studies. The resulting protein model was visualized and analyzed using UCSF ChimeraX v1.18 (Resource for Biocomputing, Visualization, and Informatics, University of California, San Francisco, CA, USA) [37,38] to evaluate the structural integrity and potential functional sites.

2.2. Docking Studies

The predicted 3D model of LuxS was used to assess the binding mode of five phytochemicals: apigenin, carnosol, chlorogenic acid, quercetin, and rosmarinic acid. As mentioned above, these compounds were identified as the major components in the ethanolic extracts of medicinal plants from Cyprus flora [33]. Docking studies were performed using PyRx v1.1 [39], employing the AutoDock Vina option with an exhaustiveness level of 8. The protein was prepared for docking by adding polar hydrogens and correcting charges under physiological pH conditions (pH 7.4), as previously described [40], using the DockPrep feature of ChimeraX software.

The two-dimensional chemical structures of the phytochemicals in Structured Data Format (SDF) were retrieved from the PubChem-NCBI database (https://pubchem.ncbi.nlm.nih.gov/, accessed on 2 August 2024) [41]. The SDF files were minimized and converted into the PDB format using the OpenBabel feature of PyRx. The phytochemicals were then screened for blind docking across the entire surface of the protein dimer. The grid box was centered at x = 12, y = 1, and z = 12, with dimensions of 65 Å × 50 Å × 65 Å. The docked structures were retrieved and visualized using BIOVIA Discovery Studio 2021 (Dassault Systèmes BIOVIA, San Diego, CA, USA). All compounds exhibited a binding score of less than −6.5 kcal/mol and were evaluated further in vitro for their anti-QS potential, as detailed in the subsequent sections.

2.3. Chemicals, Bacterial Strains, Media, and Culture Conditions

Unless otherwise stated, all chemicals, including pure phytochemicals, were purchased from Merck (Darmstadt, Germany). These included apigenin (Cat. #10798), carnosol (Cat. #49702), chlorogenic acid (Cat. #C3878), quercetin (Cat. #Q4951), and rosmarinic acid (Cat. #R4033), all of analytical grade. Phytochemical stock solutions (10 mM) were prepared in DMSO.

E. coli MG 1655 (ATCC-700926) and V. harveyi BB-170 (ATCC-BAA-1117) were obtained from the American Type Culture Collection (ATCC; Wesel, Germany).

E. coli was grown in a Luria Bertani (LB) medium consisting of 1% tryptone, 0.5% yeast extract, and 1% NaCl at 37 °C. V. harveyi BB-170 was grown at 30 °C in an autoinducer bioassay (AB) medium (ATTC medium: 2034) consisting of (per 1 L) 17.53 g of ΝaCl, 6.02 g of MgCl₂, and 2.0 g of casamino acids (vitamin-free). The pH of the medium was adjusted to pH 7.0 with 1 M KOH and autoclaved at 121 °C for 15 min. The solution was cooled to room temperature, and 10 mL of a 1 M potassium phosphate buffer (pH 7.0), 10 mL of a 0.1 M sterile arginine solution, and 20 mL of 50% v/v sterile glycerol were added to the medium.

2.4. Autoinducer-2 Bioassay

The AI-2 bioassay was performed as previously described [42,43], with minor modifications to assess the effects of five phytochemicals on AI-2 production in E. coli. In brief, E. coli was grown overnight at 37 °C in LB medium supplemented with 0.5% glucose. The following day, the overnight culture was used to inoculate (1:100) fresh LB medium containing 0.5% glucose. The culture was then incubated at 37 °C for 5 h with continuous shaking at 250 rpm [33]. To prepare the cell-free supernatant (CFS), the culture was centrifuged at 16,000× g for 15 min at 4 °C. The resulting supernatant was filtered through 0.2 μm syringe filters (VWR, West Chester, PA, USA), aliquoted, and stored at −20 °C until the AI-2 bioluminescence assay was conducted.

To assess the effects of the five phytochemicals on AI-2 production, the reporter strain V. harveyi BB170 was grown for 16 h at 30 °C in AB medium and subsequently diluted (1:5000) into fresh AB medium. A total of 90 μL of the diluted cells was added to the wells of a 96-well plate and mixed with 9 μL of CFS from E. coli MG1655 and 1 μL of each phytochemical solution at various concentrations (ranging from 5 to 200 μΜ), as described in the results section. Blank controls (9 μL of CFS + 1 μL of DMSO) and a negative control (9 μL of AB medium + 1 μL of DMSO) were included in each experiment.

The plates were incubated at 30 °C with continuous shaking at 100 rpm, and luminescence readings (in relative light units, RLU) were recorded every 20 min using a PerkinElmer VictorX3 2030 Multiplate Reader (PerkinElmer, Waltham, MA, USA) in chemiluminescence mode. The inhibition of AI-2 activity was expressed as a percentage relative to the blank control and calculated using the following equation (Equation (1)) [44].

$$\%\mathrm{AI-2} \mathrm{inhibition}=\left(1-{\displaystyle \frac{\mathrm{RLU} \mathrm{of} \mathrm{sample}}{\mathrm{RLU} \mathrm{of} \mathrm{blank} \mathrm{control}}}\right)\times 100$$

(1)

The IC₅₀ values (i.e., the concentration of inhibitor required to reduce the measured activity to 50% of its maximum value) for the tested phytochemicals were calculated using Prism v8.0.2 (GraphPad, La Jolla, CA, USA) by utilizing the software’s built-in equations and the raw data obtained from the enzymatic assays. These values were determined by fitting the dose–response curves with non-linear regression analysis.

2.5. Biofilm Formation Assay

The effects of each of the five phytochemicals on biofilm formation were assessed using sterile 96-well flat-bottom polystyrene plates, as previously described [33,45]. Positive controls (bacterial cells + LB), medium controls (LB only), and solvent controls (cells + LB + DMSO) were included in the experiments. All experiments were conducted in triplicate.

The appropriate concentration of each phytochemical was added to the test wells before inoculation. Plates were then incubated at 37 °C with continuous shaking at 100 rpm. Following 48 h of cultivation, the content of each well was discarded, and the wells were rinsed three times with phosphate-buffered saline (PBS). The wells were then fixed by drying for 1 h at 37 °C in the incubator. Once fully dry, 200 μL of 0.1% crystal violet stain was added to each well and incubated for 15 min at 25 °C. Excess dye was rinsed off with tap water. Subsequently 200 μL of 96% ethanol was added to each well to solubilize the stain adhering to the biofilm biomass. The ethanol containing the solubilized stain was transferred to clean wells of a 96-well plate, and the absorbance at 570 nm (A570 nm) was measured using a PerkinElmer VictorX3 2030 Multiplate Reader (PerkinElmer, Waltham, MA, USA). The biofilm inhibition rate was calculated using the following equation (Equation (2)).

$$\%\mathrm{Biofilm} \mathrm{inhibition}=\left(1-{\displaystyle \frac{\mathrm{A}570 \mathrm{nm} \mathrm{of} \mathrm{the} \mathrm{sample}}{\mathrm{A}570 \mathrm{nm} \mathrm{of} \mathrm{the} \mathrm{positive} \mathrm{control}}}\right)\times 100$$

(2)

2.6. Effects of the Five Phytochemicals on Bacterial Growth

The effects of the five phytochemicals at the highest concentration tested (200 μΜ) on the growth of E. coli MG1655 were evaluated in liquid culture (200 μL) in the wells of a 96-well plate following 20 h of cultivation. Growth controls (bacteria cells + LB), medium controls (LB only), and solvent controls (cells + LB + DMSO) were used. All experiments were carried out in triplicate. Optical density values at 600 nm (OD_600nm) were obtained using a Perkin Elmer VictorX3 2030 Multiplate reader (PerkinElmer, Waltham, MA, USA) at 0 and 20 h post-inoculation. The % growth on inhibition was determined using the following equation (Equation (3)) [46]:

$$\% \mathrm{inhibition}=(1 -\left({\displaystyle \frac{{\mathrm{OD}}_{\mathrm{t}20\mathrm{h}}-{\mathrm{OD}}_{\mathrm{t}0\mathrm{h}}}{{\mathrm{OD}}_{\mathrm{gc}20\mathrm{h}}-{\mathrm{OD}}_{\mathrm{gc}0\mathrm{h}}}}\right)) \times 100$$

(3)

where OD_t20h and OD_t0h are the OD_600nm values of the test well at 20 h and 0 h post-inoculation, respectively, while OD_gc20h and OD_gc0h are the OD_600nm values of the growth control well at 20 h and 0 h post-inoculation, respectively.

The effects of five phytochemicals on bacterial growth were further assessed by performing viable plate counts [47]. Phytochemicals at a final concentration of 200 μΜ and cultures of E. coli (10⁷ CFU/mL) were added to the wells of a 96-well plate (200 µL per well). Bacteria (10⁷ CFU/mL) without any of the compounds were used as a control. The plates were incubated at 37 °C without shaking for 24 h. Following incubation, bacterial suspensions were transferred to clean Eppendorf tubes, centrifuged at 5000× g for 5 min at 4 °C, washed three times with PBS, and resuspended in 200 μL of fresh LB medium. Each suspension was subsequently serially diluted in LB and plated on LB agar. After incubation at 37 °C for 16 h, the number of viable bacteria was determined and expressed as CFU/mL.

2.7. Statistical Analysis

Unless otherwise stated, experiments were carried out in triplicate, and the data are presented as the mean values ± standard deviation (SD). A one-way ANOVA followed by Tukey’s multiple-comparison test was used to compare the effects of the five phytochemicals on the biofilm formation by E. coli MG1655. Statistical significance was set at p < 0.05. The statistical analysis was performed using GraphPad Prism (v.8.2, GraphPad Software Inc., San Diego, CA, USA).

3. Results

3.1. Predicted 3D Structure of LuxS

We have previously used E. coli MG1655 as a model organism to investigate the QS activities of ethanolic extracts from several aromatic plants, including oregano, rosemary, common sage, lavender, calendula, lemon balm, Cyprus mountain tea, and lemon beebrush. Notably, extracts from oregano, rosemary, and common sage were found to inhibit AI-2 signaling by over 60%. To further explore the impact of the major phytochemical components of these extracts on QS mechanisms and specifically on the LuxS/AI-2 signaling system, we obtained the predicted 3D structure of the LuxS protein in E. coli (Figure 1A). As aforementioned, because the experimental structure of the E. coli LuxS protein was not available in the PDB, we obtained its predicted 3D structure from AlphaFold (https://alphafold.ebi.ac.uk/entry/P45578; accessed on 1 August 2024). The missing ion (Fe²⁺) was added using AlphaFill while the LuxS dimer was developed with the HDOCK server, as described in Section 2.1. “LuxS Structure Prediction”. LuxS is a small metalloenzyme, consisting of 171 amino acids in E. coli, that forms homodimers with two identical active sites at the interface between subunits [48]. High-resolution crystal structures of LuxS, both in its free form and when bound to S-ribosyl-homocysteine (SRH), have been determined at 1.2 Å for various bacterial species [20,21,22]. These studies confirm the enzyme’s consistent homodimeric structure and reveal each active site’s coordination of a divalent metal ion. Initially thought to contain Zn²⁺ based on inductively coupled plasma (ICP) metal analysis, further research by Pei and colleagues identified Fe²⁺ as the actual native metal [49]. The active sites are highly conserved across different species, characterized by a similar overall fold and an invariant HXXEH motif, which is typically found in Zn²⁺-binding proteins. The metal ion is tetrahedrally coordinated by the side chains of His 54 and His 58 in the HXXEH motif, a cysteine residue (Cys128 in E. coli; Supplementary Figure S1), and a water molecule.

Several conserved residues from the symmetry-related chain contribute to the lining of the cavity that likely accommodates the substrate (Figure 1B). Notably, both chains of the enzyme contribute amino acids to form the binding site. The “A” chain provides catalytic residues, including His54, Glu57, and His58 from the HXXEH motif, as well as Cys128. Additionally, four conserved residues from the “B” subunit—Ser6, Phe7, His11, and Arg39—are located in the active site and are potentially important for substrate binding and/or catalysis [24]. It is important to note that Glu57 and Cys83 (Cys-84 in Bacillus subtilis LuxS) are crucial residues for LuxS catalysis, as indicated by mutagenesis studies [49].

The sequence alignment (Figure 1C) of E. coli LuxS with LuxS proteins from other bacterial species highlights the conserved residues involved in binding and catalysis. In addition to the aforementioned conserved residues, other amino acids that have been implicated in binding and/or catalysis, including Phe80 [48], Gly82, and Gly93 [50], are also noted.

3.2. Molecular Docking

We subsequently employed molecular docking studies to explore the interaction and binding modes of the five phytochemicals—apigenin, carnosol, chlorogenic acid, quercetin, and rosmarinic acid—that were identified as the major components in the ethanolic extracts of medicinal plants from the Cyprus flora, as previously described [33], with the predicted LuxS dimer. The structures of the five tested compounds are illustrated in Figure 2.

By performing blind docking across the entire surface of the protein, all the tested compounds except quercetin were found to bind within the same active pocket in the cavity between the two chains (Figure 3). We further reviewed the docking poses generated for quercetin to ascertain if any structures were located in the same active site as the other compounds. Notably, our analysis confirmed that none of the quercetin poses were located in the active site or interacted with the residues crucial for catalysis, suggesting a distinct mode of interaction. The binding scores of the five compounds ranged from −6.7 (quercetin) to −8.3 kcal/mol (rosmarinic acid), as illustrated in

Table 1. Properties and binding scores of the five phytochemicals against LuxS.

Compound	PubChem CID	Molecular Mass (g/mole)	cLogP a	Binding Score b (kcal/mol)	Main Binding Amino Acid Residues c
					Chain A	Chain B
Apigenin	5280443	270.24	2.90	−7.2	His54, Glu57, His58	Ser6, Phe7, His11, Cys83
Carnosol	442009	330.40	3.16	−7.0	His54, Glu57, His58, Cys128	Ser6, His11, Arg39, Cys83,
Chlorogenic acid	1794427	354.31	−1.88	−7.3	His54, Glu57, His58, Cys128	Ser6, Phe7, His11, Cys83
Quercetin	5280343	302.23	1.50	−6.7	-	-
Rosmarinic acid	5281792	360.30	1.10	−8.3	Glu57, His58, Cys128	Ser6, Phe7, His11, Arg39

^a Calculated with ChemDraw v18.2; ^b Calculated with PyRx v1.1 (AutoDock Vina v.1; https://vina.scripps.edu/, accessed on 2 August 2024) as described in the text; ^c While several amino acids participate in substrate binding, only conserved residues are illustrated. All contributing amino acids involved in the binding of each compound are shown in Figure 4.

.

The molecular interactions of the five phytochemicals with the amino acids in the active pocket of LuxS are illustrated as 2D diagrams in Figure 4. Most amino acids in the active pocket formed hydrogen bonds and van der Waals interactions with the selected natural products. Additionally, other types of interactions, such as pi-anion and pi-sigma bonds, were predicted, as shown in Figure 4.

Consistent with previous studies, both chains of the LuxS dimer were involved in substrate (phytochemical) binding. Chain A contributed His54, Glu57, His58, Glu122, and Cys128, although not all these residues participated in the binding of each phytochemical, as shown in Table 1 and Figure 4. Similarly, Chain B was involved with Ser6, Phe7, His11, Arg39, and Cys83, but not all these conserved amino acids were implicated in binding every substrate (Table 1 and Figure 4).

However, the key conserved residues previously reported to be involved in substrate binding—Phe80 [48] Gly82, and Gly93 [50]—did not make significant contributions in binding the tested phytochemicals, possibly due to the use of a different strain. Notably, His11 in Chain B participated in the binding of all phytochemicals except quercetin, underscoring the importance of this residue in substrate binding. A mutation of His11 has been shown to result in a ~1000-fold reduction in LuxS activity [51]. Interestingly, none of the conserved amino acids were implicated in the binding of quercetin (binding score of −6.7 kcal/mol), as it had been placed on a different area on the dimer.

3.3. Effects of Phytochemicals on AI-2 Signaling

We subsequently evaluated the anti-QS potential of the five phytochemicals by monitoring AI-2 inhibition using the well-established V. harveyi BB170 bioassay [52]. This assay relies on the ability of the reporter strain V. harveyi BB170 to produce bioluminescence specifically in response to AI-2. At cell densities of 10⁶ to 10⁷ CFU/mL, bioluminescence can be detected when AI-2 is added (spiked) [53].

In our experiments, V. harveyi BB170 was incubated with a known concentration of exogenous AI-2 (i.e., cell-free supernatant (CFS) from an E. coli culture) to induce luminescence, along with either a phytochemical or its respective blank medium (CFS without the phytochemical). Inhibition was determined to have occurred when the luminescence of the sample was lower than that of its corresponding blank. Based on literature precedents, phytochemicals are typically tested at concentrations ranging from 5 to 200 μM [53,54,55]; therefore, we assessed the effects of the five phytochemicals on AI-2 production within this range.

Previous studies have shown that the concentration of the AI-2 signaling molecule in E. coli peaks during the mid-to-late growth phase, followed by a significant decrease during the stationary phase [56]. In our earlier work [33], we also demonstrated that the concentration of AI-2 in E. coli MG1655 increased with incubation time up to 6 h but began to decrease afterward. Based on these observations, in this work, we used CFSs from E. coli cultures grown for 5 h at 37 °C in the presence of 0.5% glucose.

The results illustrated in Figure 5 demonstrate that apigenin, carnosol, chlorogenic acid, quercetin, and rosmarinic acid inhibit AI-2 signaling in a concentration-dependent manner with varying degrees of efficacy. Notably, the IC₅₀ values for these compounds provide valuable insights into their relative potencies. Carnosol and chlorogenic acid exhibited similar IC₅₀ values of 58.04 μM and 60.52 μM, respectively, indicating that they are among the more potent inhibitors of AI-2 signaling. Apigenin, in contrast, showed a higher IC₅₀ value of 70.63 μM, likely due to its lack of interaction with key catalytic residues such as Ang39 and Cys128, as indicated by our docking studies, which are essential for LuxS activity.

Quercetin displayed an IC₅₀ value greater than 200 μM, indicating a minimal inhibitory effect under the tested conditions. This lack of efficacy might be due to its poor interaction with key amino acid residues within the LuxS active site, as suggested by the docking results (Figure 3 and Figure 4). Interestingly, despite rosmarinic acid showing the strongest binding score in the docking studies (−8.3 kcal/mol), its IC₅₀ value was higher (80.63 μM) than those of carnosol and chlorogenic acid. This suggests that while binding score (energy) is important, the specific interactions with crucial residues, such as Cys83, are also vital for effective inhibition. The docking studies further indicate that rosmarinic acid does not interact with Cys83, unlike carnosol and chlorogenic acid, which may explain its comparatively lower inhibitory potency despite its strong binding score.

These findings emphasize that both the binding energy and specific interactions with key residues are crucial in determining the effectiveness of these compounds in inhibiting AI-2 signaling. Compounds like carnosol and chlorogenic acid, which interact with critical residues such as Cys83 and Cys128, demonstrate more potent inhibitory effects, underscoring the importance of these interactions in the LuxS catalytic mechanism. Overall, the data suggest that while all tested compounds can bind to the LuxS enzyme, their ability to inhibit AI-2 signaling is significantly influenced by their interactions with specific key residues that are essential for LuxS activity.

3.4. Effects of Phytochemicals on Biofilm Formation

Previous studies have demonstrated that QS plays a crucial role in biofilm formation and differentiation in various bacterial species, including E. coli [57,58]. Notably, AI-2 signaling is closely associated with biofilm formation in many bacteria [59,60], including E. coli [61]. It has been shown that AI-2 production and uptake influence several E. coli phenotypes, such as biofilm formation, motility, and virulence [62,63]. During autoaggregation or biofilm formation, AI-2 functions as a chemoattractant, drawing planktonic cells to growing cell aggregates [61,64].

Given the significant role of AI-2 signaling in biofilm formation, we investigated the effects of five phytochemicals—apigenin, carnosol, chlorogenic acid, quercetin, and rosmarinic acid—on the formation of E. coli MG1655 biofilms. Biofilm formation was quantified using crystal violet staining at final concentrations of 50, 100, and 200 μM, as illustrated in Figure 6. The results indicate that all five compounds inhibited biofilm formation to varying degrees. Specifically, apigenin reduced biofilm formation by approximately 50% at 50 μM, with further reductions observed at 100 and 200 μM, though these differences were not statistically significant. Among the tested compounds, carnosol showed the highest inhibitory effect, reducing biofilm formation by over 60% even at the lowest concentration of 50 μM, with only a slight additional reduction at higher concentrations. Chlorogenic acid inhibited biofilm formation in a concentration-dependent manner. Although quercetin had only a minimal effect on AI-2 signaling (as shown in Figure 5D), it still significantly reduced biofilm formation, with increased inhibition at higher concentrations. Lastly, rosmarinic acid also displayed a concentration-dependent inhibition of biofilm formation.

3.5. Effects of the Five Phytochemicals on Bacterial Growth

The main difference between QS inhibitors (QSIs) and antibiotics is that QSIs inhibit the production of virulence factors without killing pathogenic bacteria, thereby reducing the likelihood of bacteria developing drug-resistant mutations [33]. We have previously demonstrated that the ethanolic extracts of oregano, rosemary, and common sage, which contain the five compounds tested here as major components, had minimal impact on bacterial growth [33]. Therefore, in this study, we further investigated whether the observed anti-QS activities of the five compounds (at the maximum concentration tested, viz. 200 μΜ) were correlated with bactericidal activity by examining their effects on the growth of E. coli MG1655.

Consistent with our previous observations [33], none of the phytochemicals exhibited any bactericidal activity, as determined by the inhibition-of-growth assay (Figure 7A) and viable plate counts (Figure 7B).

4. Discussion

Bacteria use QS to coordinate gene expression programs that underlie collective behaviors. QS is a cell–cell communication process that depends on the production, release, detection, and group-level response to extracellular signaling molecules known as AIs. These small chemical signals regulate a wide variety of cellular activities, including motility, virulence, antibiotic production, biosurfactant formation, and biofilm formation [3,65].

AIs vary among bacterial species. For example, acyl-homoserine lactones are typically produced by G⁻ bacteria, while oligopeptides are secreted by G⁺ bacteria. These molecules primarily facilitate intraspecies communication [66]. However, AI-2 is a well-conserved QS signal synthesized by a wide range of G+ and G⁻ bacteria. AI-2, produced via the activated methyl cycle by the enzymes Pfs and LuxS, enables communication at both intra- and interspecies levels, thereby promoting coordination among different bacterial species [67,68].

Inhibiting QS pathways among bacteria presents a potential strategy for combating multi-drug-resistant bacterial infections. Compounds that can inhibit QS activity, particularly without affecting bacterial growth rates, are considered potential QS inhibitors [69]. Notably, several plant extracts and essential oils have been reported to exhibit both antimicrobial and anti-QS activities, highlighting the importance of identifying anti-QS compounds from natural sources, such as aromatic plants [70].

Considering the central role of LuxS in AI-2 signaling, in this study, we examined the inhibitory potential of five phytochemicals—apigenin, carnosol, chlorogenic acid, quercetin, and rosmarinic acid—previously identified as the major components of ethanolic extracts from oregano, rosemary, and common sage in the flora of Cyprus. We initially employed molecular docking studies to explore the interaction and binding models between the five phytochemicals and LuxS. Except for quercetin, the other four compounds were positioned in the cavity between the two protein subunits (chains) and exhibited binding scores of lower than −7.0 kcal/mol. Consistent with previous studies, both subunits contributed amino acids for substrate binding, including His54, Glu57, His58, and Cys128 from Chain A and Ser6, Phe7, His11, Arg39, and Cys83 from Chain B (Table 1 and Figure 4). In contrast to Meng et al. [71], who tested various natural products as inhibitors of LuxS in Lactobacillus reuteri and reported metal complexes between some compounds (e.g., norathyriol, mangiferin, and kaempferol) and Zn²⁺, we did not detect any complexes between the Fe²⁺ ion of E. coli LuxS and the five phytochemicals. This discrepancy is likely due to differences in the docking methods used (Schrödinger software version 2021-1 in Meng’s study vs. AutoDock in our study), the bacterial strains tested (L. reuteri vs. E. coli), and the specific phytochemicals examined.

We subsequently assessed the effects of the five phytochemicals on AI-2 signaling in vitro using the well-established V. harveyi bioassay [52] with the reporter strain BB170 which is highly sensitive to AI-2 (with a QS phenotype of AI-1⁻, AI-2⁺), allowing even low levels of AI-2 to be detected in this bioassay. Inhibition was considered significant when the luminescence of a tested compound (i.e., phytochemical) was lower than that of the corresponding blank control. Consistent with our docking studies, apigenin, carnosol, chlorogenic acid, and rosmarinic acid inhibited AI-2 signaling in a concentration-dependent manner (Figure 5). Carnosol and chlorogenic acid were the most potent, with IC₅₀ values of approximately 60 μM, while apigenin and rosmarinic acid exhibited slightly higher IC₅₀ values of around 70 μM and 80 μM, respectively, despite apigenin having the lowest binding score (−8.3 kcal/mol) in our docking studies. This discrepancy can be partly explained by the fact that carnosol and chlorogenic acid interact with both Cys83 of Chain B and Cys128 of Chain A, which have been identified as essential amino acid residues for LuxS activity. However, the exact mechanism by which each of these phytochemicals inhibits LuxS activity is currently under investigation in our laboratory.

The five phytochemicals demonstrated a dose-dependent inhibition of biofilm formation under the experimental conditions (Figure 6). Although quercetin did not significantly impact AI-2 signaling—likely because it binds to a site away from the active site according to our docking studies (Figure 3 and Figure 4) and does not interact with any of the essential amino acids involved in substrate binding—it still had a significant effect on biofilm formation in E. coli. Additionally, quercetin exhibited only a minor impact on LuxS activity (Figure 5). To ensure that the observed inhibition of biofilm formation was not due to reduced bacterial growth, we assessed the impact of each phytochemical on E. coli growth separately and found that none of them inhibited bacterial growth (Figure 7).

To the best of our knowledge, while the five phytochemicals examined in this study have been explored for their general antimicrobial and anti-QS activities, there is limited direct research on their specific effects on LuxS activity and AI-2 signaling. Most available studies have focused on broader quorum sensing and biofilm inhibition without directly measuring LuxS activity or AI-2 levels. It should be noted that that the actions of phytochemicals against QS are diverse, multiple, and depend on the bacterial strain tested and the experimental method used.

Flavonoids such as quercetin [72] and apigenin [73] have been shown to act as quorum-sensing (QS) inhibitors, suppressing bacterial cell-to-cell communication and exerting antagonistic effects on bacterial signaling. Both compounds have been demonstrated to inhibit biofilm formation in E. coli O157 and V. harveyi [74]. Furthermore, quercetin has been found to inhibit QS-controlled virulence factors such as violacein, elastase, and pyocyanin in Chromobacterium violaceum CV12472 and P. aeruginosa PAO1 [75]. It has also been found to reduce the expression levels of QS-related genes such as lasI, lasR, rhlI, and rhlR at a concentration of 16 μg/mL. Additionally, quercetin inhibits QS circuitry by interacting with the transcriptional regulator LasR in P. aeruginosa [76].

The anti-QS properties of rosmarinic acid and chlorogenic acid have also been documented [77]. For example, rosmarinic acid was shown to inhibit biofilm formation and reduce the QS-mediated production of hemolysin, lipase, and elastase in Aeromonas hydrophila strains. It also downregulated virulence genes such as ahh1, aerA, lip, and ahyB [78].

In P. aeruginosa, rosmarinic acid has been shown to induce QS-dependent gene expression, increase biofilm formation, and enhance the production of virulence factors like pyocyanin and elastase. Another study revealed that rosmarinic acid induced the expression of 128 genes, including numerous virulence factor genes, and triggered a broad QS response in Pseudomonas aeruginosa PAO1. It also induced seven small RNAs (sRNAs), all of which were encoded near QS-induced genes [79].

Chlorogenic acid has been shown to inhibit biofilm formation in P. aeruginosa, suppress swarming motility, and reduce the production of virulence factors, including protease, elastase, rhamnolipids, and pyocyanin [80]. It has also exhibited similar inhibitory effects in Chromobacterium violaceum, reducing biofilm formation, swarming motility, chitinolytic activity, and violacein production. Specifically, chlorogenic acid was found to suppress QS in Chromobacterium violaceum (CECT 494) by inhibiting violacein production [81].

Likewise, direct studies on carnosol’s effects on LuxS activity or AI-2 signaling are limited; however, it has been observed to influence quorum-sensing mechanisms more broadly. For example, carnosol has been demonstrated to inhibit the Staphylococcus aureus quorum-sensing agr operon at a low concentration of 5 μM [82].

5. Conclusions

Bacterial resistance and pathogenesis are closely linked to their ability to sense and respond to population density through QS. The inhibition of the QS system is considered a novel strategy for developing antipathogenic agents, particularly for combating drug-resistant bacterial infections. Several mechanisms can achieve QS inhibition using natural compounds (phytochemicals), including [83,84,85]: (i) the inhibition of the biosynthesis of QS signaling molecules; (ii) competitive inhibition, where certain natural compounds share structural similarities with QS signaling molecules and compete for binding to receptor proteins, thereby disrupting the signal transduction pathways; and (iii) the inhibition of QS signal reception. Notably, phytochemicals exhibit diverse and multiple actions against QS, with their effects varying depending on the bacterial strain and experimental conditions used. In this study, we demonstrated for the first time that four phytochemicals—apigenin, carnosol, chlorogenic acid, and rosmarinic acid—abundant in ethanolic extracts from aromatic plants bind to the substrate-binding pocket of the LuxS enzyme of E. coli MG1655, thereby inhibiting AI-2 signaling. Importantly, AI-2 is a signal molecule produced by LuxS, an enzyme found in many bacterial species, and is essential for enabling interspecies communication.

These findings highlight the potential of these phytochemicals as promising candidates for further research in vitro and in vivo, as they may offer significant potential for new drug development aimed at mitigating bacterial resistance and pathogenesis through QS inhibition. Expanding our understanding of how these compounds interact with bacterial QS systems could lead to the development of novel therapeutic agents capable of targeting resistant bacterial strains without the selective pressure that often leads to antibiotic resistance.

It should be noted that besides anti-QS activity, polyphenolic compounds including apigenin, carnosol, chlorogenic acid, quercetin, and rosmarinic acid are also well known for their strong antioxidant properties and extensive pharmacological profiles. Their ability to interact with multiple and diverse molecular targets, known as promiscuity, presents both advantages and challenges. On the one hand, this characteristic enables these molecules to exert a wide range of therapeutic effects, such as anti-inflammatory, anticancer, and cardiovascular protective activities. However, it also complicates the prediction of side effects and interactions with other drugs, as these compounds can bind to various enzymes, receptors, and other cellular components. This complexity can hinder the therapeutic application of polyphenols, as their beneficial properties may also lead to unintended off-target effects. For example, while quercetin’s modulation of several signaling pathways can be beneficial in cancer treatment, it may also inadvertently affect metabolic pathways. Therefore, a thorough understanding of the promiscuity of these compounds is essential to fully exploit their therapeutic potential while minimizing adverse effects. Thus, research on the potential uses of polyphenols as therapeutic agents must be vigilant, considering both the polyvalent nature of these compounds and their extensive implications in pharmacodynamics.