Cranberry-derived proanthocyanidins impair virulence and inhibit quorum sensing of Pseudomonas aeruginosa

Bacteria have evolved multiple strategies for causing infections that include producing virulence factors, undertaking motility, developing biofilms, and invading host cells. N-acylhomoserine lactone (AHL)-mediated quorum sensing (QS) tightly regulates the expression of multiple virulence factors in the opportunistic pathogenic bacterium Pseudomonas aeruginosa. Thus, inhibiting QS could lead to health benefits. In this study, we demonstrate an anti-virulence activity of a cranberry extract rich in proanthocyanidins (cerPAC) against P. aeruginosa in the model host Drosophila melanogaster and show this is mediated by QS interference. cerPAC reduced the production of QS-regulated virulence determinants and protected D. melanogaster from fatal infection by P. aeruginosa PA14. Quantification of AHL production using liquid chromatography-mass spectrometry confirmed that cerPAC effectively reduced the level of AHLs produced by the bacteria. Furthermore, monitoring QS signaling gene expression revealed that AHL synthases LasI/RhlI and QS transcriptional regulators LasR/RhlR genes were inhibited and antagonized, respectively, by cerPAC. Molecular docking studies suggest that cranberry-derived proanthocyanidin binds to QS transcriptional regulators, mainly interacting with their ligand binding sites. These findings provide insights into the underlying mechanisms of action of a cerPAC to restrict the virulence of P. aeruginosa and can have implications in the development of alternative approaches to control infections.

. (a) Inhibition of virulence determinants and (b) growth curves of P. aeruginosa PA14 in absence or presence of different cerPAC concentrations. LasA: staphylolytic protease, LasB: elastase and AprA: alkaline protease. Results are expressed as means and standard deviations (SD) of triplicate enzyme assays (* p < 0.001). Bacterial growth (OD 600 ) was monitored at 37 °C for 18 h in TSB medium. Error bars with average data points of growth kinetics represent the standard deviation of values obtained from four replicates. Abbreviations: cerPAC x, Cranberry extract rich in proanthocyanidins at x μ g mL −1 (e.g., cerPAC 300 indicates cerPAC at 300 μg mL −1 ).
Scientific RepoRts | 6:30169 | DOI: 10.1038/srep30169 The difference in the treated or untreated PA14 strains' ability to kill D. melanogaster in this feeding assay may have been due to modified survival of the bacteria on the filter papers used for exposure during incubation. To address this possibility, we analyzed the survival of PA14 on the paper discs without and with 200 μ g mL −1 cerPAC under the same conditions as the fly feeding assay. There was no significant difference (F 5,30 = 0.54, p = 0.74) in culturability of the bacterium on the filter paper discs in the absence and presence of cerPAC during incubation (see Supplementary Fig. S1), indicating that an alteration in survival ability of bacteria could not account for the observed differences in fly killing. Overall, these results indicate that cerPAC protect D. melanogaster from P. aeruginosa infection.
Cranberry extract rich in proanthocyanidins modulates the AHL-mediated quorum sensing system in P. aeruginosa PA14. Since QS regulates multiple virulence determinants in P. aeruginosa, we hypothesized that cerPAC contain molecule(s) that might interfere with QS in P. aeruginosa PA14. Therefore, to determine the ability of cerPAC to modulate the production of the two principal AHL molecules by P. aeruginosa PA14, we determined AHL production kinetics in absence or presence of 200 μ g mL −1 cerPAC. As shown in Fig, 3A,B, cerPAC significantly impairs the production of 3-oxo-C 12 -HSL (t = 7.45, df = 4, p < 0.001) and C 4 -HSL (t = 3.54, df = 4, p < 0.05), in P. aeruginosa PA14 at exponential and late stationary phase, respectively. This reduction in the production of the QS signals was observed without affecting bacterial growth (Fig. 3C).
To understand the mechanism for the reduction in AHL levels, β-galactosidase transcriptional fusion reporters of lasI (3-oxo-C 12-HSL synthase) and rhlI (C 4 -HSL synthase) were assayed in P. aeruginosa PA14 bioreporter strain with the same 200 μ g mL −1 cerPAC exposure. These bioassays revealed that expression of both AHL synthase genes (lasI and rhlI) is repressed by cerPAC (Fig. 4A,B). Similarly, we also investigated whether presence of cerPAC affects the expression of the two cognate transcriptional regulator genes lasR and rhlR using lacZ transcriptional fusion reporters. Expression of both regulator gene fusions was partially repressed in the presence of cerPAC (Fig. 4C,D). Thus, cerPAC inhibits both AHL synthases and partially represses the LuxR-type regulator genes associated with the production of the two AHL signals in P. aeruginosa PA14.
Cranberry extract rich in proanthocyanidins act as an antagonist of AHL-mediated quorum sensing in P. aeruginosa PA14. Considering that AHLs act as autoinducing ligands of LasR and RhlR, we hypothesized that cerPAC component(s) interfere with LasR/RhlR activation by AHLs. We thus investigated whether cerPAC affects LasR and/or RhlR induction by exogenous AHLs using bioreporter AHL-negative PA14 mutants with lacZ fusions. As expected, when 3-oxo-C 12 -AHL or C 4 -HSL were supplied to their respective bioreporters, they activated the expression of lasI and rhlI, respectively (Fig. 5A,B). While cerPAC had no effect on the activity of the reporters, there was a significant (p < 0.05) reduced activation by either AHLs in presence of cerPAC (Fig. 5A,B). This indicates that cerPAC partially inhibits the activation of both LasR-and RhlR-directed transcription of lasI and rhlI, respectively, the primary targets of these LuxR-type regulators. Additionally, LasR and RhlR activation titration was performed in absence and presence of three different concentrations of cerPAC, which resulted in lower activation of LasR and RhlR (Fig. 5C,D). This indicates that cerPAC can reduce the activation of both regulators by their native AHLs, likely as a potential antagonist.
Cranberry extract rich in proanthocyanidins inhibits LasR activity without binding to AHL molecules and also interact with LasI. To assess a possible physical interaction between cerPAC components and either AHL molecule, we quantified C 4 -HSL and 3-oxo-C 12 -HSL in cell-free growth medium using an ethyl acetate extraction procedure followed by LC-MS analysis. If the cerPAC binds to the AHLs, we would expect to   observe a reduction in AHL concentration due to compromised extraction. As shown in Supplementary Fig. S2, there was no difference in the concentration of AHLs with or without cerPAC, demonstrating that cerPAC components do not bind to the AHLs and therefore do not inhibit QS by physical interaction.
Inhibition of Las-type QS regulators' activities by cerPAC may be due to structural interactions, important for the functional activity of transcriptional regulatory proteins. To address this possibility, in silico docking analysis was performed using protein structures of LasR (2UV0 46 ), LasI (1RO5 47 ), the monomer and dimer of epicatechin molecules (important components of cPACs 48 ). To test our docking method, we compared the interaction energy scores (obtained using Moldock tools) of the predicted docking complex and the known crystallographic complex structures of the LasR with ligand 3-oxo-C 12 -HSL. The Moldock interaction energy score of − 144.1 kcal mol −1 for the predicted complex of LasR with 3-oxo-C 12-HSL was marginally lower than the Moldock interaction energy score of − 157.5 kcal mol −1 obtained for the crystallographic complex of LasR with ligand 3-oxo-C 12 -HSL ( Fig. 6a and see Supplementary Table S1). The epicatecin and its dimer (proanthocyanidin) molecules were docked separately in the internal cavity of LasR (Fig. 6b,c). Ligand binding domain (LBD) of LasR with a volume of 653 Å 3 , exhibits sufficient space to accommodate the monomer or dimer of epicatechin with a volume of 225 Å 3 or 466 Å 3 , respectively. The in silico docking analysis suggests that the complex formation between the epicatechin and LasR, with a Moldock interaction energy score of − 127.1 kcal mol −1 , is more favorable than LasR-proanthocyanidin complex with Moldock score of − 68 kcal mol −1 (see Supplementary Table S1). The proanthocyanidin formed six hydrogen bonds at the internal binding cavity of LasR compared to four hydrogen bonds of the LasR-3-oxo-C 12 -HSL or LasR-epicatechin complex (Fig. 6). The increase in the Moldock score for the docking complex of LasR with proanthocyanidin compared to LasR-epicatechin complex was observed due to the steric constraints of the proanthocyanidin structure in the internal cavity space of LasR identified by the comparison of their internal energies (see Supplementary Table S1). Due to the lack of crystallographic structure of LasI protein bound with its natural substrates or functional analogues, we performed in silico docking analysis to predict a complex of LasI with its natural substrate S-adenosyl L methionine (SAM) (Fig. 7a). This putative complex with LasI was used as a reference for both docking analyses of epicatechin and proanthocyanidin. The best five structural positions of SAM with higher interaction energies occupied the same binding cavity on the LasI protein. The docking analysis showed the formation of hydrogen bonds of SAM with residues that surround the putative binding cavity with Moldock interaction energy score of − 126.2 kcal mol −1 ( Fig. 7a and see Supplementary Table S1). The binding cavity known for the second substrate of LasI, the acyl-acyl carrier protein (acyl-ACP) was not identified as a potential binding site for either of the tested cerPAC components (epicatechin or proanthocyanidin). The LasI-epicatechin complex showed single hydrogen bond with Moldock interaction energy score of − 106.8 kcal mol −1 (Fig. 7b and see Supplementary Table S1). The docking complex of the LasI protein with the proanthocyanidin molecule suggests the more favorable complex formation with five hydrogen bonds and Moldock interaction energy score of − 153.6 kcal mol −1 compared to the LasI-SAM complex (Fig. 7c). This in silico docking analysis suggests that both main components of cerPAC have the potential to form complexes with LasR and LasI proteins to compete with their native ligands 3-oxo-C 12 -HSL and SAM, respectively. Cranberry extract rich in proanthocyanidins impairs AHL production in other pathogenic strains. According to our data, cerPAC can act as a general QS inhibitor by interfering with the binding of the AHL ligand to LuxR-type transcriptional regulators. To verify that cerPAC is able to impede QS in other bacterial species, we performed an AHL production kinetics assay to examine the effect of administering cer-PAC to wild type strains of Burkholderia ambifaria and Chromobacterium violaceum. The addition of cerPAC to growth medium significantly impairs the production of the two main AHLs (C 8 -HSL and C 6 -HSL) in B. ambifaria (Fig. 8A) and C 6 -HSL in C. violaceum (Fig. 8B). Since the primary target of LuxR regulators are luxI homologues, these observations validate the capacity of cerPAC to interfere with AHL-mediated QS in different bacterial species.

Discussion
The identification of selective anti-virulence therapies that abolish the production of virulence determinants without affecting the viability of pathogenic bacteria would be extremely useful in combating bacterial infections caused by broad-spectrum antibiotic-resistant pathogens. The search for QS inhibitors or quenchers is a promising strategy aimed at development of innovative anti-microbial strategy to attenuate the virulence of infectious bacteria. Here, we showed that a cranberry extract enriched in PACs restricts virulence of P. aeruginosa in a fruit fly animal model and inhibits QS mechanisms. In addition, the cerPAC does not perturb cell viability of P. aeruginosa, indicating that use of these molecules may provide less selective pressure towards the development of resistance than conventional antibiotics (bactericidal and bacteriostatic, which pose strong selective pressure in any environment 45 ). However, it would be naive to presume that QS inhibition is unlikely to put any selective pressures on pathogenic bacteria that may lead to resistance against QS inhibition, as quite a few studies suggest the possibility of resistance development against QS inhibition 44,45 . Nonetheless, it is important to continue to expand our strategies for combating pathogen resistance by identifying novel anti-microbial and anti-virulence agents 49 . Our findings have elucidated the likely mechanisms of action behind the anti-virulence efficacy of cerPAC: 1) it reduces the production of AHL signaling molecules; 2) it represses the expression of the QS regulators LasR and RhlR and autoinducer synthases LasI and RhlI; 3) it antagonizes the activation of LasR and RhlR by their cognate autoinducers; and 4) epicatechin and proanthocyanidin, the main components of cerPAC, are modeled in silico to interact with the LBD of LasR and LasI. In addition, cerPAC also inhibits AHL production in strains of the Gram-negative species B. ambifaria and C. violaceum. Thus, cerPAC could have anti-virulence activity against various pathogens with clinical importance.
For in vivo study, we used a fly feeding assay which represents a long-term infection model and involves feeding starved flies with bacterial cultures. This method is better adapted to chronic infections compared to the fly nicking model 50 . A dose of cerPAC was supplied at the start of infection and virulence was subsequently reduced, indicating that a cranberry extract enriched in PACs could function as a prophylactic. These results, when considered with other literature, indicate that the use of effective prophylactic molecules with anti-virulence activity, specifically for P. aeruginosa, could be a best practice in the clinical setting 51 . It is noteworthy that the extract used herein contains approximately 95% proanthocyanidins, and thus, it is presumed that the bioactivity observed can be mostly attributed to these molecules. Nonetheless, there is also a possibility that other, as yet unidentified molecules present in the cranberry extract, also act as potent quorum sensing inhibitors. Additional studies conducted with fractionated cranberry extracts are needed to achieve more insights into the bioactivity of specific fractions of cranberry derivatives.
That a cerPAC alone inhibits QS has not been previously reported in the peer-reviewed scientific literature. An earlier study reported the impairment of QS in P. aeruginosa in the presence of a commercial cranberry juice cocktail 52 ; however, the mechanisms of action of this complex mixture of molecules was not identified. We targeted Las and Rhl QS systems because they are at the top of the P. aeruginosa quorum sensing hierarchy 33 . Both AHL molecules induce their own production and activate the corresponding LuxR-type transcriptional regulators LasR and RhlR 36,53 . In the presence of the cranberry extract, we observed impairment in AHL production, along with reduced gene expression of AHL synthase (LasI and RhlI) and partial repression of their regulators (LasR and RhlR). Similar observations have been made using other QS inhibitor molecules 51,[54][55][56] . Interestingly, we found that cerPAC, a potent in vivo inhibitor, is an effective antagonist of both LasR and RhlR, two regulators that act reciprocally on key virulence determinants 33,51,57 . We propose that anti-virulence efficacy of the cerPAC is essentially due to its interference with the LasIR and RhlIR-dependent QS regulatory circuitry, though we cannot exclude a multifactorial effect.
Previous studies have identified anti-QS and anti-biofilm candidate molecules effective against P. aeruginosa 17,43,44,58 . Among them, plant-derived natural flavonoid and phenolic compounds are commonly studied for the inhibition of P. aeruginosa biofilm 17,43 and QS 43 . The structure of a typical PAC is tetrameric in nature, composed of epicatechin units with one A-type linkage 48 . Catechin, an epimer of epicatechin, inhibits biofilm formation and QS-controlled virulence factors 59 . Consequently, other structurally-similar flavonols, such as baicalein 60 , naringenin 61 and quercetin 55 , are also anti-QS and anti-biofilm compounds. Thus, we docked monomer and dimer (proanthocyanidin) of epicatechin molecule onto LasR and LasI proteins to assess the specific interaction. Our hypothesis was that these molecules would bind to LasR and LasI, occupying the crucial LBD, which would result in inhibition of P. aeruginosa QS systems. The successful molecular docking of epicatechin or proanthocyanidin with LasR suggested that the inactivation of transcriptional regulators may be the primary mechanism of action for the cerPAC as anti-virulence factors in vivo.
Our results show that a cerPAC protects D. melanogaster from P. aeruginosa likely through an inhibition of QS without negative effect on bacterial growth. Antagonist activity and in silico analysis projected the potential mechanism of action to the inhibition of AHL regulators. As hopeful as these anti-virulence strategies are, they emerge with a new set of preclinical and regulatory development challenges. It remains to be verified whether plant-derived compounds cause significant side-effects in animals or humans. However, one can also conceive of important clinical application for treatment of acute and chronic infections caused by P. aeruginosa and potentially other bacterial pathogens, as our preliminary data suggest cerPAC is a general inhibitor of AHL-mediated QS. Taken together, our results present a promising case for cerPACs as efficient QS inhibitors for the regulation of bacterial pathogenicity.

Infection of Drosophila melanogaster.
Fruit flies (D. melanogaster) were infected orally in fly feeding assay as before 50,72 , with some modifications. Briefly, flies were anesthetized under a gentle stream of carbon dioxide. Male flies (3-to 5-days-old) were starved of food and water for 5-6 h and separated into vials (10 per vial) containing 5 ml of 5% sucrose agar (sterile) without and with 200 μ g mL −1 cerPAC and 2.3-cm filter paper disks (sterile) containing freshly grown bacterial culture suspension. To achieve this freshly grown culture, an overnight PA14 culture was inoculated in 6 mL TSB culture and incubated at 37 °C and 100 rpm until OD 600 = 3.0. This culture was centrifuged at 12,000× g for 1 min and the resulting pellet resuspended in 150 μ L of sterile 5% sucrose, without and with 200 μ g mL −1 cerPAC. All filters were soaked appropriately with this culture suspension, along with sucrose agar, in feeding vials prior to transferring flies into the vial. Separate feeding vials soaked with 150 μ L of 5% sucrose without and with 200 μ g mL −1 cerPAC were used as negative controls for each experiment. Post-infection mortality of flies was monitored daily for 14 days, with each treatment tested twice in triplicate.

LC-MS analyses.
Specific estimation of AHL molecules was achieved by LC-MS in the positive electrospray ionization (ESI+ ) mode, combined with the MRM mode, as described previously 64,65,73 .
Samples of PA14 culture exposed to cerPAC were retrieved at different time points and OD 600 was measured. An aliquot of methanolic internal standard was mixed with each sample to adjust final concentration 3 mg L −1 of 5,6,7,8-tetradeutero-4-hydroxy-2-heptylquinoline (HHQ-d 4 ) and 6 mg L −1 of 5,6,7,8-tetradeutero-3,4-dihyd roxy-2-heptylquinoline (PQS-d 4 ) 73 . All culture samples were vortex-mixed and extracted twice with ethyl acetate (1:1, vol:vol), each ethyl acetate extract pooled and evaporated at 30 °C under a gentle stream of nitrogen. The residue was then resuspended in acidified acetonitrile (HPLC grade, containing 1% ACS grade acetic acid) at ten times the initial concentration and 20 μ L aliquots were injected for LC-MS analysis 73 (more details are in Supplementary Information). β-galactosidase assay for LacZ expression. β-galactosidase activity was measured as described by Miller 74 , with slight modifications as reported previously 53  Antagonists/agonists assay. To evaluate the activity of cerPAC as antagonists/agonists against the natural AHL ligand of LasR or RhlR, 3-oxo-C 12 -HSL (Sigma-Aldrich) and C 4 -HSL (Cayman Chemical) were used as inducers in this assay. The AHL-deficient strain that has been engineered to produce β -galactosidase upon activation of LasR by 3-oxo-C 12 -HSL [∆ lasI (lasI-lacZ; pME3853)] and RhlR by C 4 -HSL [∆ rhlI (rhI-lacZ; pME3846)], were grown overnight in TSB medium. The overnight culture was diluted in fresh TSB and was grown to achieve an OD 600 = 0.3. An appropriate amount of sterilized cerPAC stock solution prepared in MilliQ water was added to sterile culture tube containing TSB. For control condition, either 3-oxo-C 12 -HSL or C 4 -HSL (stock solution in DMSO as a control) was added to sterile culture tube containing TSB, final DMSO concentration (after addition of cells) did not exceed 0.5% v/v. Bacterial cells were added to TSB (final OD 600 = 0.05) without and with cerPAC approximately 30 min prior to the addition of the AHL at a final concentration of 3.13-50 nM (for 3-oxo-C 12 -HSL) or 62.5-1000 nM (for C 4 -HSL), to achieve final volume of 2 mL. Culture tubes were incubated at 37 °C for 3 h under shaking at 200 rpm, measurement of cell OD 600 and β -galactosidase assay were performed at the regular time intervals after 3 h of incubation. The concentration of cerPAC that reduced or increased the β -galactosidase activity compared to controls containing 3-oxo-C 12 -HSL or C 4 -HSL (without cerPAC) was considered to determine antagonist or agonist activity.
In silico docking analysis. To understand the interaction between components of the cerPAC with LasR and LasI protein structures, we performed virtual docking using the Piecewise Linear Potential and Lennard-Jones algorithms that can identify steric and hydrogen bonding interactions, and the Coulomb potential for electrostatic forces. In silico docking analysis was performed using the Molegro Virtual Docker 6.0 suite without the incorporation of water molecules. To maintain the search robustness, twenty rounds of iteration were used for each docking process. The S-adenosyl L methionine (NCBI Pubchem CID 34756) and the 3-oxo-C 12 -HSL (NCBI Pubchem CID 127864) molecular structures were used as native ligand molecules, for the LasI (RCSB Protein data base ID 1RO5 47 ) and LasR (RCSB Protein data base ID 2UV0) proteins, respectively. The components of cer-PAC, epicatechin (NCBI PubChem CID 182232) and proanthocyanidin (NCBI PubChem CID 108065) molecular structures were used as ligands in virtual docking for both proteins. The MolDock search tool 75 , that combines guided differential evolution and a cavity prediction algorithm was used for docking scores (in kcal mol −1 ) based on the interaction energies of each complex. The best five positions with high Moldock interaction energies were sampled and compared in every complex computed. The Computed Atlas of Surface Topography 76 of proteins was used to explore the volumes of the cavities in the target proteins (http://sts.bioe.uic.edu/castp/index.php). The molecular graphics and analyses were performed using the UCSF Chimera version 1.1 77 .

Statistical analysis.
All experiments were repeated in quadruplicate, unless otherwise specified. Fruit fly survival curves were prepared using GraphPad Prism 6 (GraphPad Software, Inc., San Diego, CA) to perform a statistical Log-rank (Mantel-Cox) test. Data represented are the means of replicates and the differences among the control, and results were analyzed by using one way or two way ANOVA as required.