Suppression of bacterial cell–cell signalling, biofilm formation and type III secretion system by citrus flavonoids


  • A. Vikram,

    1.  Vegetable and Fruit Improvement Center, Department of Horticultural Sciences, Texas A&M University, College Station, TX, USA
    Search for more papers by this author
  • G.K. Jayaprakasha,

    1.  Vegetable and Fruit Improvement Center, Department of Horticultural Sciences, Texas A&M University, College Station, TX, USA
    Search for more papers by this author
  • P.R. Jesudhasan,

    1.  Department of Poultry Science, Texas A&M University, College Station, TX, USA
    Search for more papers by this author
  • S.D. Pillai,

    1.  Vegetable and Fruit Improvement Center, Department of Horticultural Sciences, Texas A&M University, College Station, TX, USA
    2.  Department of Poultry Science, Texas A&M University, College Station, TX, USA
    Search for more papers by this author
  • B.S. Patil

    1.  Vegetable and Fruit Improvement Center, Department of Horticultural Sciences, Texas A&M University, College Station, TX, USA
    Search for more papers by this author

Bhimanagouda S. Patil, Vegetable and Fruit Improvement Center, Department of Horticultural Sciences, Texas A&M University, College Station, TX 77843-2119, USA. E-mail:


Aim:  This study investigated the quorum sensing, biofilm and type three secretion system (TTSS) inhibitory properties of citrus flavonoids.

Methods and Results:  Flavonoids were tested for their ability to inhibit quorum sensing using Vibrio harveyi reporter assay. Biofilm assays were carried out in 96-well plates. Inhibition of biofilm formation in Escherichia coli O157:H7 and V. harveyi by citrus flavonoids was measured. Furthermore, effect of naringenin on expression of V. harveyi TTSS was investigated by semi-quantitative PCR. Differential responses for different flavonoids were observed for different cell–cell signalling systems. Among the tested flavonoids, naringenin, kaempferol, quercetin and apigenin were effective antagonists of cell–cell signalling. Furthermore, these flavonoids suppressed the biofilm formation in V. harveyi and E. coli O157:H7. In addition, naringenin altered the expression of genes encoding TTSS in V. harveyi.

Conclusion:  The results of the study indicate a potential modulation of bacterial cell–cell communication, E. coli O157:H7 biofilm and V. harveyi virulence, by flavonoids especially naringenin, quercetin, sinensetin and apigenin. Among the tested flavonoids, naringenin emerged as potent and possibly a nonspecific inhibitor of autoinducer-mediated cell–cell signalling. Naringenin and other flavonoids are prominent secondary metabolites present in citrus species. Therefore, citrus, being a major source of some of these flavonoids and by virtue of widely consumed fruit, may modulate the intestinal microflora.

Significance and Impact of the Study:  Currently, a limited number of naturally occurring compounds have demonstrated their potential in inhibition of cell–cell communications; therefore, citrus flavonoids may be useful as lead compounds for the development of antipathogenic agents.


Flavonoids are prominent secondary metabolites ubiquitously present in the plant kingdom. Health-promoting properties of flavonoids are focus of intensive research since the beginning of 20th century. Several epidemiological studies have also correlated intake of citrus flavonoids with health benefits (Rossi et al. 2007; Benavente-Garcia and Castillo 2008). Over the years, research in various laboratories has demonstrated several health benefits of flavonoids. Studies in our laboratory (Yu et al. 2005; Vanamala et al. 2006; Miller et al. 2008) and elsewhere (Benavente-Garcia et al. 1997; Benavente-Garcia and Castillo 2008; Morin et al. 2008) have provided numerous evidences, suggesting the role of citrus flavonoids in anticancer, antioxidant, anti-inflammatory, cardiovascular diseases and other health benefits.

In recent years, antimicrobial activity of flavonoids against various bacterial species is being explored (Fig. 1a). However, the published data on the antimicrobial properties of flavonoids do not seem to provide consistent results (Cushnie and Lamb 2005). This is evident from the divergent minimum inhibitory concentration (MIC) values reported for Escherichia coli (Basile et al. 1999, 2000, 2003; Rauha et al. 2000; Gatto et al. 2002; Mandalari et al. 2007). This paradoxical situation may be the result of different assay systems employed, and often unreported initial inoculum-size, which significantly affects the outcome of the experiment. Furthermore, it is possible that some of these flavonoids influence the bacterial cells in noninhibitory fashion i.e. they modulate various physiological processes rather than inhibiting the growth. These observations led us to explore the role of flavonoids as modulators of other possible mechanisms. One such intensively investigated antivirulence mechanism is quorum sensing. It has been postulated that interference with quorum sensing or cell–cell signalling will impact the bacterial pathogenicity (Rasmussen and Givskov 2006a,b).

Figure 1.

 (a) Structure of citrus flavonoids used in the study. (Neohesperidose = 2-O-alpha-l-Rhamnosyl-d-glucose; Rutinose = 6-O-l-rhamnosyl-d-glucose). (b) Basic structure of flavonoid nucleus.

Quorum sensing is a coordinated regulation of gene expression as a function of cell-density (Bassler and Losick 2006). Bacterial cells produce and secrete small molecules, termed autoinducers, in the local environment. These autoinducers are recognized by specific two-component signalling systems in a concentration-dependent fashion. Upon reaching a threshold concentration, binding of autoinducers to specific receptors elicits adequate response to initiate signalling cascade. Activation of signalling cascade results in simultaneous regulation of several genes across the population (Surette and Bassler 1998; Mok et al. 2003; Bassler and Losick 2006). This signalling system is used by the bacteria to assess its population as well as density of other bacterial species in a particular niche (Camilli and Bassler 2006). This phenomenon has gained importance in recent years with the enumeration of tight regulation of pathogenic traits such as virulence and biofilm formation by autoinducer-mediated cell–cell signalling (Fuqua et al. 2001; Bassler 2002; Ahmer 2004; Henke and Bassler 2004a; Gonzalez Barrios et al. 2006; Walters and Sperandio 2006; Choi et al. 2007). Therefore, quorum sensing or cell–cell signalling has emerged as an alternative target to control bacterial virulence (Hentzer and Givskov 2003; Rasmussen and Givskov 2006a,b).

Biofilms are complex bacterial communities adhered to surfaces, which pose a critical problem in every day life by causing any economic and health problems (Costerton et al. 1999). Biofilms severely afflict immune system, owing to the reduced expression of polysaccharide matrix and phagocytosis-resistant biofilm-colonies (Mahenthiralingam et al. 1994). Even though there has been a great interest in biofilms, effective treatments are still limited. With the increase in the importance of biofilms, it has become imperative to search for alternative antimicrobials with nonconventional targets. Quorum sensing may be one such target, because quorum sensing, in particular autoinducer-2-mediated cell–cell signalling, may be important regulatory factor for biofilm production in E. coli, Vibrio spp. and Salmonella Typhimurium (Prouty et al. 2002; Hammer and Bassler 2003; Mok et al. 2003; Lu et al. 2005; Gonzalez Barrios et al. 2006).

Research in recent years has targeted towards the identification of synthetic compounds and analogues with quorum sensing inhibitory properties or quorum quenching. However, only a limited number of natural compounds have been tested for their quorum-quenching abilities (Borchardt et al. 2001; Castang et al. 2004; Adonizio et al. 2006; Choo et al. 2006; Defoirdt et al. 2007). On the other hand, a few natural products have been demonstrated to possess biofilm inhibitory property (Chorianopoulos et al. 2008). Considering the potential of the natural compounds in disease prevention, it is imperative to study their potential as quorum sensing and biofilm inhibitors. Because certain flavonoids are biologically active in preventing diseases, we were interested in the elucidation of their effect on the bacterial system. In the current study, we explored the impact of citrus flavonoids on autoinducer-mediated bacterial cell–cell signalling. Furthermore, ability of these flavonoids to curtail cell–cell signalling controlled processes, such as biofilm formation and type three secretion system (TTSS), was investigated.

Materials and methods


Naringin has been purified previously in our laboratory (Raman et al. 2004). Naringenin, kaempferol, quercetin, rutin, hesperidin, apigenin, neohesperidin and neoeriocitrin were purchased from Sigma-Aldrich Co. (St Louis, MO, USA). Sinensetin was purchased from Chromadex Inc. (Irvine, CA, USA). All compounds were dissolved in DMSO (dimethyl sulfoxide) at 10 mg ml−1.

Bacterial strains and media

Escherichia coli O157:H7 ATCC 43895 and biosensor strain Vibrio harveyi MM32 (luxN::Cm, luxS::Tn5Kan) were purchased from ATCC (Manassas, VA, USA). The bioreporter strain V. harveyi BB886 (luxPQ::Tn5) and V. harveyi strain BB120 (wild-type) were kindly provided by B. L. Bassler (Princeton University, Princeton, NJ) (Evans et al. 1977). Escherichia coli K12 was used as a positive control for AI-2 bioassay. The autoinducer bioassay (AB) medium was used to culture the V. harveyi strains (Surette and Bassler 1998; Lu et al. 2004). The colony-forming antigen (CFA) medium was used to culture the E. coli O157:H7 strain (Evans et al. 1977; Jackson et al. 2002), whereas V. harveyi BB120 biofilm was grown in LM (Luria marine) medium.

Cell growth assay

Overnight culture of V. harveyi BB120 and E. coli O157:H7 were diluted in 1 : 100 ratio in LM or LB (Luria Bertani) media and treated with either 100, 50, 25, 12·5 or 6·25 μg ml−1 flavonoids or equivalent volume of DMSO. The cultures were grown for 16 h and OD at 600 nm was measured every 2 h. In addition, 1 : 100 dilution of V. harveyi BB120 and E. coli O157:H7 overnight culture was treated with 100 μg ml−1 of flavonoids and grown at 30 or 37°C with shaking. Samples were drawn every 2 h, and serial dilution were prepared in saline solution, and plated on LM or LB agar plates. The colonies were counted after 24 h of growth.

Assay for inhibition of intercellular signalling

Cell-free supernatant (CFS) was prepared from E. coli strain K12 and V. harveyi BB120 as described earlier (Surette and Bassler 1998). Overnight cultures of V. harveyi and E. coli were diluted at 1 : 100 ratio with LM or LB media and grown further at 30°C (for V. harveyi) or 37°C (for E. coli) to achieve high concentration of autoinducer molecules. CFS was collected by the centrifugation of cells at 2500 g for 20 min followed by filtration through 0·2- μm membrane filter and stored at −20°C.

The citrus flavonoids at five different concentrations (6·25, 12·5, 25, 50 and 100 μg ml−1) were tested for their ability to inhibit the luminescence in V. harveyi reporter strains BB886 (AHL) or MM32 (AI-2). Bioluminescence assay was used essentially as reported previously (Surette and Bassler 1998; Lu et al. 2005) with one modification. The reporter strains were cultured overnight at 30°C with aeration in AB media and diluted in 1 : 5000 ratio with fresh AB medium. Each well received 5% CFS, 0·5% flavonoids in DMSO or DMSO, 4·5% sterile AB media and 90% freshly diluted culture. In case of negative control, CFS was replaced with sterile AB media. The microtiter plates were incubated for 4 h at 30°C at 100 rev min−1. At the end of incubation period, light production was measured by using a Victor2 1420 multilabel counter (Beckman Coulter, Fullerton, CA, USA) in luminescence mode. The values were recorded as relative light units (RLU) and used in calculation.

Biofilm formation

Overnight cultures of E. coli O157:H7 strain ATCC 43895 and V. harveyi BB120 were diluted with fresh CFA media or LM media (Evans et al. 1977). The diluted culture was placed in 96-well microtiter plates. All the wells received either 0·5% of DMSO (control) or test compound dissolved in DMSO. The plates were incubated at 26°C for 24 h without shaking. Total biofilm mass was quantified by washing the plates with phosphate buffer (0·1 mol l−1, pH 7·4) and staining with 0·3% crystal violet (Fisher, Hanover Park, IL, USA) for 20 min. The extra dye was removed by washing with phosphate buffer (0·1 mol l−1, pH 7·4). All of the dye associated with the attached biofilm was dissolved with 200 μl of 33% acetic acid, and an absorbance reading at 570 nm by Synergy™ HT Multi-Mode Microplate Reader (Bio-Tek, Winooski, VT, USA) was used to quantify the total biofilm mass. Each experiment was carried out at least thrice with three replicated wells per plate. Each data point was averaged from three replicates and expressed as mean ± standard deviation.

Semi-quantitative realtime RT-PCR

Three genes encoded in three V. harveyi TTSS locus vopD, vscO and vscR were selected for study (Table 1). Relative transcript amounts of three genes were measured by qRT-PCR. Primers were designed by Primer3 software program (Rozen and Skaletsky 2000). Primers designed were targeted to the unique sequences as determined by the alignment with the V. harveyi BB120 genomic sequence.

Table 1.   Sequence of primers used in the study
Gene nameSequence 5′–3′

One ml samples from naringenin (100 μg ml−1), hesperidin (100 μg ml−1) or DMSO (10 μl ml−1) treated cultures were collected at each time point during 8- h growth period as mentioned in ‘Cell Growth Assay’ section. Total RNA was extracted from the samples using RNeasy minikit (Qiagen Inc.) according to the manufacturer’s instructions. RNA quantity was measured using NanoDrop ND 1000 spectrophotometer (Thermo Fischer Scientific, Pittsburg, PA). Reverse transcription was performed using MuLV reverse transcriptase enzyme and random hexamer on a programmable thermocycler (Gene Amp PCR system 2700; Applied Biosystems, Foster City, CA, USA), under following conditions: 60 min at 42°C for annealing and elongation followed by 5 min at 99°C to inactivate the reverse transcriptase enzyme. Amplification of target sequences was carried out on ABI-Prism 7000 HT (Applied Biosystems). A total of 25 ng cDNA was amplified using 10 pmol of primers and 10 μl of SYBR® GREEN PCR mix (Applied Biosystems, Warrington, UK). All data were normalized to the 16s gene. After completion of 40 PCR cycles, melt curve data were generated. All the measurements were carried out on three biological replicates consisting of three technical replicates each.

Statistical analysis

The inhibition of AI activity was calculated from the formula 100 − [(relative AI activity/relative activity of positive control) × 100] (Lu et al. 2004) and expressed as percentage and SD values. The percentage inhibition of biofilm formation was calculated as 100 − [(OD570 of sample well/OD570 of positive control) × 100] and expressed as percentage and SD values.

For semi-quantitative real-time PCR, the cycle threshold (Ct) was calculated as the cycle number when primary fluorescent curve crossed a threshold of 30 fluorescence units. The Ct values for primers were normalized against that of 16S RNA. Fold change in the gene expression was calculated by 2(−ΔΔCt) (Livak and Schmittgen 2001) and expressed as fold change ±SD.


Interference with bioluminescence production in Vibrio harveyi

To evaluate the effect of flavonoids on cell–cell communication system, the bioluminescence production in reporter strains V. harveyi BB886 and MM32 in the presence of flavonoids were measured. All the flavonoids except hesperidin (Fig. 2e) inhibited either HAI-1- or AI-2-mediated bioluminescence, significantly. Naringenin, kaempferol, quercetin and apigenin (Fig. 2a, h, f and i) significantly (P < 0·01) inhibited both HAI-1- and AI-2-induced bioluminescence. Naringin and neohesperidin (Fig. 2b,d) depicted stronger inhibitory activity against HAI-1. In contrast, sinensetin (Fig. 2j) was more effective antagonist against AI-2-mediated bioluminescence (Fig. 2j). Naringin and sinensetin demonstrated maximal response at 100 μg ml−1 against HAI-1 and AI-2, respectively, whereas neohesperidin depicted a saturated response at higher than 25 μg ml−1 doses against HAI-1. Moreover, naringin and rutin (Fig. 2b,g) demonstrated a concentration-dependent inhibition of HAI-1 and AI-2. At the same time, hesperidin and neoeriocitrin were relatively ineffective (Fig. 2e,c).

Figure 2.

 Inhibition of HAI-1 (grey bars) and AI-2 (black bars) mediated bioluminescence in Vibrio harveyi mutant strains BB886 and MM32. (a) Naringenin, (b) naringin, (c) neoeriocitrin, (d) neohesperedin, (e) hesperidin, (f) quercetin, (g) rutin, (h) kaempferol, (i) apigenin and (j) sinensetin. The bars represent mean of 9 data points and SD.

Response of Vibrio harveyi and Escherichia coli O157:H7 growth to flavonoids

Growth rate of V. harveyi BB120 was measured at OD600 over a period of 16 h to determine the inhibitory effect of flavonoids is presented in Fig. 3a. Apigenin, quercetin and kaempferol inhibited the growth of V. harveyi BB120 significantly (P < 0·01), whereas neoeriocitrin induced the growth rate significantly (P < 0·01). Naringenin, naringin, neohesperedin, sinensetin, hesperidin and rutin did not show inhibitory effect on V. harveyi BB120 growth (P > 0·05). To further confirm the influence of naringenin, quercetin and kaempferol, V. harveyi was grown upto 16 h in the presence of these flavonoids, and viable cell count was determined by plating on LM agar plates every 2 h (Fig. S1). Quercetin and kaempferol demonstrated significant effect on V. harveyi growth, whereas naringenin was ineffective at 100 μg ml−1. These results supported the earlier observations using OD600 (Fig. S1). However, a dose-dependent response of quercetin and kaempferol was observed (Fig. S2). In contrast, only quercetin and kaempferol demonstrated growth inhibitory activity against E. coli 0157:H7 at 100 μg ml−1 (Figs 3b and S3), but not at lower concentrations (Fig. S4).

Figure 3.

 Growth of (a) Vibrio harveyi BB120, (b) Escherichia coli O157:H7 measured at 600 nm in the presence of citrus flavonoids. The data represent mean of three biological replicates. (a) (inline image) Control; (inline image) apigenin; (inline image) neohesperidin; (inline image) quercetin; (inline image) neoeriocitrin; (inline image) naringin; (inline image) naringenin; (inline image) hesperidin; (inline image) rutin; (inline image) kaempferol and (inline image) sinensetin. (b) (inline image) Control; (inline image) naringenin; (inline image) apigenin; (inline image) hesperidin; (inline image) neohesperidin; (inline image) rutin; (inline image) quercetin; (inline image) kaempferol; (inline image) neoeriocitrin; (inline image) sinensetin and (inline image) naringin.

Inhibition of biofilm formation in Vibrio harveyi and Escherichia coli O157:H7

Biofilm formation in Vibrio spp. and E. coli is regulated by quorum sensing particularly by AI-2-mediated cell signalling (Croxatto et al. 2002; Hammer and Bassler 2003; Gonzalez Barrios et al. 2006). Therefore, we hypothesized that compounds inhibiting AI-2-mediated bioluminescence may influence the biofilm formation. The biofilm formation by V. harveyi and E. coli O157:H7 in the presence or absence of flavonoids was investigated. All the flavonoids significantly (P < 0·01) inhibited V. harveyi BB120 biofilm formation, whereas except hesperidin, all the flavonoids demonstrated a significant (P < 0·01) inhibition of E. coli O157:H7 biofilm formation in a dose-dependent fashion (Fig. 4). Quercetin (Fig. 4f) and naringenin (Fig. 4a) were potent inhibitors of biofilm formation by V. harveyi BB120 and E. coli O157:H7. Interestingly, sinensetin depicted strong inhibition of E. coli O157:H7 and V. harveyi biofilm formation (Fig. 4j). These results also support the hypothesis that quorum sensing inhibitory flavonoids are likely to inhibit biofilm formation.

Figure 4.

 Inhibition of biofilm formation by Escherichia coli O157:H7 (grey bars) and Vibrio harveyi BB120 (black bars) by (a) naringenin, (b) naringin, (c) neoeriocitrin, (d) neohesperedin, (e) hesperidin, (f) quercetin, (g) rutin, (h) kaempferol, (i) apigenin and (j) sinensetin. The bars represent average of 9 data points and SD.

Response of Vibrio harveyi TTSS to naringenin

Because naringenin emerged as potential candidate for cell–cell signalling inhibition, further investigation was focused on expression of V. harveyi TTSS in the presence of naringenin. To understand the behaviour of TTSS, the expression of three genes located on three different locus (Henke and Bassler 2004a) over a period of 8 h was investigated. Because DMSO was used as the solvent to dissolve flavonoids, control experiment consisted of treatment of V. harveyi with DMSO. Therefore, expression of the three genes in the presence of DMSO (Fig. 5a) was studied. Interestingly, vopD was upregulated by 151-fold by 2 h over the initiation point (t = 0), which was gradually reduced to 23-fold by 8 h. Similar trend was observed for vcrD and vscO. The expression of vscO and vcrD increased up to 4 h, thereafter, the expression level of both genes was reduced from their high points. This observation was consistent with the results reported by Henke and Bassler (2004a).

Figure 5.

 Expression of vopD, vcrD and vscO over a period of 8 h in the presence of (a) DMSO, (b) naringenin and (d) hesperidin. The fold change was calculated over the time = 0 h, value for which was taken as 1·0. (c) The relative change in expression of vopD, vcrD and vscO upon exposure of 100 μg ml−1 naringenin (c) and hesperidin (e) over control is presented. (inline image) vopD; (inline image) vcrD and (inline image) vscO.

Further, we were interested in the effect of naringenin on the expression pattern of these three genes. The expression pattern of vopD, vscO and vcrD, in the presence of naringenin, depicted a similar trend as control i.e. induction during early stages of growth while repression at later phases. Although, expression of three genes depicted similar trend, their induction levels differed greatly. At 8 h, vopD, vscO and vcrD were suppressed as much as 37-, 43- and 2·5-fold, respectively, over t = 0 (Fig. 5b).

In addition, the expression of vscO, vopD and vcrD in response to naringenin over control was determined. The three genes were downregulated by naringenin treatment compared to the control. Interestingly, vopD was very strongly suppressed, and vcrD and vscO were moderately downregulated (Fig. 5c). Because the three genes are located on different loci, naringenin treatment may possibly down-regulate the loci involved in the production of TTSS.

As a negative control, expression of vopD, vcrD and vscO was investigated upon exposure of 100 μg ml−1 hesperidin. Expression of vopD, vscO and vcrD demonstrated similar pattern as control i.e. all three genes were induced during 2–6 h, and expression was decreased from their high level by 8 h (Fig. 5d). However, relative change over control demonstrated that vopD and vscO were up-regulated 2·8- and 2·9-fold, respectively, at t = 0. Expression of vopD was decreased at t = 2 but increased thereafter upto 6 h (5·7-fold) then decreased by 8 h (Fig. 5e). vscO demonstrated similar trend except that it was induced by 19-fold at t = 8. vcrD was induced upto 4 h and then decreased (Fig. 5e).


Inhibition of cell–cell signalling-mediated bioluminescence

In V. harveyi, bioluminescence production is regulated by quorum sensing. Three coincidence detectors, LuxN, LuxPQ and CqsS, detect three autoinducer molecules HAI-1, AI-2 and CAI-1, respectively. Dephosphorylation cascade from three detectors converges on phosphorelay protein LuxU, resulting in regulation of transcriptional regulator LuxR. LuxR, in turn, regulates all the V. harveyi quorum sensing responsive genes known so far (Mok et al. 2003; Henke and Bassler 2004a,b; Bassler and Losick 2006). Our data suggested that quercetin, kaempferol and naringenin are very effective antagonists of HAI-1- and AI-2-mediated cell–cell signalling. However, quercetin and kaempferol suppressed the growth rate of V. harveyi (Fig. 3a). Taken together, it is likely that the effect of quercetin and kaempferol at higher concentrations is because of their growth-inhibiting ability rather than quorum sensing. However, at lower concentrations, activity of quercetin and kaempferol may be because of inhibition of quorum sensing. These data must be interpreted with a caution because mode of action of these flavonoids is not known. In contrast to quercetin and apigenin, naringenin did not inhibit the growth rate of V. harveyi but strongly inhibited the HAI-1- and AI-2-induced bioluminescence. These observations suggest a possible quorum sensing inhibitory property for naringenin without affecting the bacterial growth. Because HAI-1- and AI-2-mediated signalling activate same transcriptional regulator luxR (Waters and Bassler 2006) in V. harveyi, inhibition of both HAI-1- and AI-2-mediated cell–cell signalling is likely an indication of a nonspecific antagonistic activity. As naringenin inhibited both HAI-1- and AI-2-mediated bioluminescence (Fig. 2a), it is possible that naringenin is a nonspecific antagonist of cell–cell signalling, which targets the signal transduction pathway at LuxU or downstream of LuxU.

Furthermore, naringin and neohesperidin were found to be effective against HAI-1-mediated cell signalling, whereas sinensetin was more active against AI-2. Differential inhibition of HAI-1- or AI-2-mediated bioluminescence by neohesperidin, naringin and sinensetin seems to suggest specific interactions. In addition, it is unlikely that inhibition of cell–cell signalling by sinensetin, naringin and neohesperidin may be because of growth inhibition, as these flavonoids did not affected the growth of V. harveyi (Fig. 3). Interestingly, apigenin, which affected the V. harveyi growth rate moderately, was weaker antagonist of AI-2-mediated cell–cell signalling compared to naringenin. However, mode of action of apigenin has not been elucidated. A probable cytostatic effect may explain the observed results.

Inhibition of Vibrio harveyi and Escherichia coli O157:H7 biofilm

Biofilm formation is highly regulated process and controlled by several factors including autoinducer-mediated cell–cell signalling. As quorum sensing also controls formation of biofilms, study of biofilm formation in the presence of flavonoids was hypothesized to provide further insight into the inhibitory property of flavonoids. As previous studies have demonstrated that AI-2 regulates biofilm formation through MsqR (Gonzalez Barrios et al. 2006) in E. coli, it was likely that flavonoids inhibiting the AI-2-mediated cell–cell signalling will inhibit also biofilm formation. As our data suggested that some of the flavonoids interfere with autoinducer-mediated cell–cell signalling, we were further interested whether these flavonoids can also influence biofilm formation. The results indicated that naringenin was a potent antagonist of V. harveyi BB120 and E. coli O157:H7 biofilm formation. Taken together, results of cell–cell signalling inhibition and biofilm inhibition suggest that the ability of naringenin to modulate the biofilm may stem from its antagonistic activity for autoinducer-mediated cell–cell signalling. On the other hand, inhibition of biofilm at higher concentrations by quercetin may be attributed to its growth inhibitory property (Figs 3a,b and 4f). Similar to V. harveyi., quercetin and kaempferol may act through inhibition of quorum sensing. Interestingly, sinensetin was found to be very potent antagonist of V. harveyi and E. coli O157:H7 biofilm. It is pertinent to note that E. coli possess luxS/AI-2 quorum sensing system but does not produce AI-1 molecule (Walters and Sperandio 2006). It is possible that the antagonistic activity of sinensetin against biofilm formation is because of inhibition of AI-2-mediated cell–cell signalling. On the contrary, apigenin did not inhibit biofilm formation of V. harveyi BB120 or E. coli O157:H7 as effectively as quercetin or naringenin. Consistent with published literature, apigenin reduced the growth rate of V. harveyi (Cushnie and Lamb 2005). It is possible that apigenin possess only growth inhibitory property similar to several growth inhibitory agents, which demonstrate poor antagonistic activity against biofilms (Lewis 2001). Furthermore, the bioluminescence (Fig. 2b–e,g) and biofilm inhibitory activities (Fig. 3b–e,g) of glycosylated flavonoids did not correlated well. Natural compounds are reported to inhibit biofilm formation by various mechanisms without affecting the growth rate (Ren et al. 2005; Duarte et al. 2006). It is possible that the various flavonoids may have differential mode of action for inhibition of biofilm formation.

Suppression of Vibrio harveyi TTSS by naringenin

In V. harveyi, three operons encoding a TTSS were identified (Mok et al. 2003; Henke and Bassler 2004a). TTS system in V. harveyi is suggested to be regulated by cell–cell signalling, and both HAI-1 and AI-2 is required for significant differential expression of type III secretion system (Mok et al. 2003; Henke and Bassler 2004a). Based on phenotypic assays, naringenin emerged as most potent cell–cell signalling inhibitor; therefore, its effect on the expression of TTS system was studied. Three genes, vopD, vscO and vcrD, encoded by three TTSS operons (Henke and Bassler 2004a) were selected to investigate the effect of naringenin on three operons. Naringenin suppressed all the three genes, suggesting suppression of corresponding loci. In contrast, hesperidin, which did not demonstrated appreciable activity in bioluminescence assays and was moderately effective against V. harveyi biofilm, induced the three genes. Quorum sensing inhibitor furanones isolated from the sea algae Delisea pulchera, cinnamaldehyde and their derivatives were shown to inhibit the virulence of V. harveyi BB120 (Defoirdt et al. 2005, 2006; Brackman et al. 2008). In the current study, the naringenin disrupted the cell–cell signalling, biofilm formation and suppressed the virulence genes. Our results were similar to the published reports related to two quorum sensing inhibitors furanone and cinnamaldehyde (Defoirdt et al. 2007; Brackman et al. 2008). Based on these observations, naringenin seems to be a natural quorum sensing disruptor molecule.

Structure-activity relationship

Flavonoids present in Citrus spp. belong to three of six major classes – flavanone, flavone and flavonols (Calabro et al. 2004). Different Citrus spp. are characterized by distinct flavonoids profiles; however, flavanones are the dominant class of flavonoids present in many citrus species (Gattuso et al. 2007). Further, the bioactivity of flavonoids in vitro depends upon the arrangement of functional groups on the nucleus. In the current study, ten flavonoids, representing flavanone (naringenin, naringin, neoeriocitrin, neohesperedin and hesperidin), flavonol (kaempferol, quercetin and rutin), flavone (apigenin) and polymethoxy flavone (sinensetin) were investigated.

In vitro biological activities of flavonoids were dependent upon the type of sugar moiety and the position. Addition of sugar moiety may be detrimental to the biological activity. Reduction in potency of aglycones by the addition of sugar moiety at seventh position (Fig. 1a,b) (naringin) as well as at third position (rutin) in the three assays was observed. Moreover, neohesperidose was found to be more effective than rutinose.

The spatial arrangement and number of functional groups affect the bioactivity of flavonoids. In particular, presence of a double bond between second and third position enhances the antioxidant and anticancer activities of flavonoids (Fotsis et al. 1997; Heim et al. 2002). In contrast, addition of double bond between second and third position was detrimental in our assays, which is evident from the reduced activity of apigenin compared to naringenin. However, addition of hydroxyl groups was found to increase the activity, with third position being more responsive. Whereas, modification at 4′ position did not affect the activity dramatically. In case of aglycones, addition of hydroxyl group at 5′ position was found to increase the antimicrobial action.


In the current study, we provided the evidence that citrus flavonoids may influence the bacterial cell–cell signalling and biofilm formation. Furthermore, several flavonoids modulated the biofilm formation by E. coli O157:H7, a pathogen with no known effective treatment at present. Based on our results, we propose that naringenin is possibly a nonspecific quorum sensing inhibitor. Moreover, naringenin suppressed the biofilm formation in V. harveyi and E. coli O157:H7 as well as TTSS in V. harveyi. These properties may have their implication in human health as naringenin was found to modulate AI-2-mediated signalling and E. coli O157:H7 biofilm. It is pertinent to note that intestinal microflora produce AI-2, which may be involved in interspecies signalling among different bacterial species (Sperandio et al. 2003). Citrus fruits are very widely consumed and are rich source of dietary flavonoids including naringenin. Therefore, consumption of citrus fruits and juices may influence the cell–cell signalling by gut microflora. In addition, these flavonoids may also serve as the lead compounds for the antipathogenic drug discovery, a critical requirement to counter the effects of pathogenic bacteria.


This project is based upon the work supported by the USDA CSREES IFAFS #2001-52102-02294 and USDA-CSREES # 2006-34402-17121 ‘Designing Foods for Health’ through the Vegetable & Fruit Improvement Center.