OUP user menu

Inhibitory effects of broccoli extract on Escherichia coli O157:H7 quorum sensing and in vivo virulence

Kang-Mu Lee, Jeesun Lim, Sunyoung Nam, Mi Young Yoon, Yong-Kuk Kwon, Byeong Yeal Jung, YongJin Park, Sungsu Park, Sang Sun Yoon
DOI: http://dx.doi.org/10.1111/j.1574-6968.2011.02311.x 67-74 First published online: 1 August 2011


Broccoli extract (BE) has numerous beneficial effects on human health including anticancer activity. Quorum sensing (QS), mediated by self-produced autoinducer (AI) molecules, is a key process for the production of virulence determinants in pathogenic bacteria. BE suppressed AI-2 synthesis and AI-2-mediated bacterial motility in a dose-dependent manner in Escherichia coli O157:H7. In addition, expression of the ler gene that regulates AI-3 QS system was also diminished in response to treatment with BE. Furthermore, in an in vivo efficacy test using Caenorhabditis elegans as a host organism, C. elegans fed on E. coli O157:H7 in the presence of BE survived longer than those fed solely on the pathogenic bacteria. Quantitative real-time PCR analysis indicated that quercetin was the most active among the tested broccoli-derived compounds in downregulating virulence gene expression, while treatment with myricetin significantly suppressed the expression of the eae gene involved in type III secretion system. These data suggest that BE and its flavonoid constituents can inhibit expression of QS-associated genes, thereby downregulating the virulence attributes of E. coli O157:H7 both in vitro and in vivo. This study clearly elucidates BE's QS-inhibitory activity and suggests that BE has the potential to be developed as an anti-infective agent.

  • broccoli extract
  • quorum sensing
  • Escherichia coli O157:H7
  • Caenorhabditis elegans


Escherichia coli O157:H7, a causative agent for hemorrhagic colitis and hemolytic uremic syndrome (HUS), modulates the expression of its virulence-associated genes via quorum sensing (QS) signaling pathway (Sperandio, 2002). Autoinducer-2 (AI-2), a furanosyl borate diester (Chen, 2002) and AI-3, which has an unknown structure, are two major QS signals in E. coli O157:H7. AI-2 QS mediates both inter- and intraspecies bacterial communication, while AI-3 crosstalks with the mammalian hormone norepinephrine to coordinate bacteria–host interaction (Sperandio, 2003). In E. coli O157:H7, biofilm formation and motility were reported to be controlled by the AI-2 QS system (Sperandio, 2002; Gonzalez Barrios, 2006). Escherichia coli O157:H7 harbors QS-regulated virulence genes on a pathogenicity island termed the locus of enterocyte effacement (LEE) (Surette & Bassler, 1998) that is organized mainly into the five polycistronic operons LEE1LEE5 (Kaper, 2004). The first gene in LEE1, LEE-encoded regulator (ler), produces the principal transcriptional activator of the LEE genes (Elliott, 2000) and its expression was reported to be positively regulated by both AI-3 and norepinephrine (Sperandio, 2003; Jelcic, 2008).

In patients with E. coli O157:H7 infection, antibiotic use is generally limited because bacterial cells lysed by antibiotic treatment release an excessive quantity of Shiga toxin, thereby aggravating the patient's state and resulting in HUS (Wong, 2000). To avoid this risk, an antimicrobial treatment that involves attenuation of bacterial virulence by inhibiting QS has been proposed (Ren, 2004). Halogenated furanone compounds as QS inhibitors were isolated from marine macroalga, Delisea pulchra (Givskov, 1996). Many of the synthesized furanone derivatives have also been identified as QS inhibitors both in vitro (Martinelli, 2004) and in vivo (Wu, 2004). However, most of the characterized QS inhibitors have not yet been qualified as chemotherapeutic agents because they are composed of halogens that exert toxic effects in humans. Thus, more efforts should be made to develop safer QS inhibitors from natural products.

As a soluble fiber, broccoli (Brassica oleracea) contains a large amount of vitamin C and multiple nutrients with potent anticancer properties (Vasanthi, 2009). However, the effect of broccoli against infection by pathogenic bacteria has never been reported. In this study, we demonstrate the inhibitory effects of broccoli extract (BE) on bacterial QS using E. coli O157:H7 as a model organism. The in vivo effects of the BE against E. coli O157:H7 infection were also elucidated in a Caenorhabditis elegans killing assay. Finally, we tested three different flavonoid compounds (quercetin, kaempferol and myricetin) reported to be present in BE (He, 2008; Schmidt, 2010) in order to gain better insight into the active inhibitory compound in BE.

Materials and methods

Bacterial strains, chemicals and culture conditions

An E. coli O157:H7 strain ATCC 43894 producing Shiga toxins I and II, an avirulent E. coli OP50 strain and Chromobacterium violaceum CV026 were grown in Luria–Bertani broth (LB, 10g tryptone, 5g NaCl, 5g yeast extract L−1) at 37°C. Vibrio harveyi BB170, an AI-2 reporter strain, was grown at 30°C with agitation (175r.p.m.) in the AB medium (Fong, 2001). The AB medium consisted of 10mM potassium phosphate (pH 7.0), 0.3M NaCl, 0.05M MgSO4, 0.2% Casamino acids (Difco), 2% glycerol, 1mM l-arginine, 1μgmL−1 of thiamine, and 0.01μgmL−1 of riboflavin. Quercetin, kaempferol and myricetin were purchased from Sigma-Aldridge (St. Louis, MO).

Preparation of BE

The broccoli was purchased at a local market in Seoul, Korea. The broccoli was shade dried and milled to a fine powder. Powdered samples of 50g were extracted using 1000mL of distilled water at 4°C for 12h. The extract was freeze-dried and the dried pellet weighed ∼7g. The pellet was then dissolved again in 100mL of distilled water, sterilized by filtration through a 0.22μm membrane filter and stored at −20°C for further experiments.

AI-2 assay, violacein assay, swarming motility assay and quantitative real-time PCR (qRT-PCR) analysis

For the AI-2 analysis, bacterial supernatants were assayed as described previously (Surette & Bassler, 1998). In brief, the bacteria were grown in LB broth containing 0.5% (w/v) glucose with varying concentrations of BE (from 0% to 5%). AI-2 production was detected via a V. harveyi AI-2 bioassay using culture supernatants harvested at 2h postinoculation. The AI-2 level was expressed as a value relative to the AI-2 value of the supernatant from the culture of E. coli O157:H7 grown without BE. The level of violacein produced by CV026 was assessed as described previously (McClean, 1997). A swarming motility assay was performed as described elsewhere (Gonzalez Barrios, 2006). Briefly, 20μL of E. coli O157:H7 cultures grown overnight were mixed with the same volume of BE solutions to yield final BE concentrations of 0%, 0.25%, 0.5%, 2.5% and 5%. Then, LB agar plates were spot inoculated with 5μL of each mixture. After incubation for 11h at 30°C, the soft agar (0.3%) plates that showed bacterial growth halos were scanned for image analysis. qRT-PCR analysis was performed as described previously (Yoon, 2011). The primer sequences are listed in Table 1 and a transcriptional level of rrsD gene encoding a ribosomal protein was used for normalization.

View this table:
Table 1

Oligonucleotides used for qRT-PCR

GenesSequence of PCR Primers (5′–3′)

Promoter activity assay

A DNA fragment containing the promoter of ler in E. coli O157:H7 was amplified using specific oligonucleotides, PlerF (AGCGCGAGCTCTTAGAGATACTGGCTTTC AGG, SacI recognition site underlined) and PlerR (AGGCCGGATCCTTTAATATTTT AAGCTATTAGCGAC, BamHI recognition site underlined), and then digested with SacI and BamHI, and cloned into the SacI and BamHI sites of pAD123, yielding transcriptional fusion with gfp. The Pler–GFP fusion plasmid, pLER-GFP, was used to measure the promoter activity of ler. Escherichia coli O157:H7 was transformed with pLER-GFP or pAD123 (control) by electroporation. The transformed E. coli strains were inoculated into LB broth and grown overnight at 37°C. The cultures were diluted to 1:100 in Dulbecco's modified Eagle's medium (DMEM) containing norepinephrine (50μM) with or without BE (2.5%, v/v) and then incubated at 37°C for 6h. Green fluorescence intensity of each culture was measured using a Victor X4 multilabel counter (Perkin Elmer Life and Analytical Sciences, Waltham, MA).

Caenorhabditis elegans survival assays

The germ line-defective and temperature-sensitive C. elegans glp-4 mutant strain (Beanan & Strome, 1992) was obtained from the Caenorhabditis Genetic Center at the University of Minnesota and maintained as described previously (Hope, 1998). All E. coli strains were grown overnight in LB broth at 37°C with aeration. Twenty microliters of cultures were mixed with or without 0.5% BE. The mixtures were then spread onto nematode growth media agar plates (Hope, 1998). The plates were dried at 25°C and immediately utilized for the assays. Twenty nematodes previously synchronized on the L4 stage were transferred to each plate and incubated at 25°C. After every 24h, live worms were scored. When the worms did not respond to being touched by a platinum wire pick, they were considered dead.

Statistical analysis

Data are expressed as mean±SD. An unpaired Student's t-test was used to analyze the data. To compare differences among more than three groups, one way anova was used. A P-value of <0.05 was considered statistically significant. All the experiments were repeated for reproducibility.


BE treatment suppresses the production of AI-2, while it increases bacterial growth

AI-2-mediated QS plays a major role in the virulence of E. coli O157:H7 (Sperandio, 2001; Sircili, 2004). To investigate the specific effect of the BE on QS, we measured the level of AI-2 secreted by E. coli O157:H7 in response to the treatment with BE. When assayed using V. harveyi AI-2 reporter strain BB170, a decreasing level of AI-2 was detected in culture supernatants of E. coli O157:H7 grown with increasing concentrations of BE. Figure 1a shows a dose-dependent decrease in AI-2 level upon treatment with BE. It is of note that AI-2 level was almost undetectable in the presence of 5% BE. AI-2 level at each treatment normalized to that obtained from growth with no BE (Fig. 1a). We then tested C. violaceum strain CV026, which produces violacein, a violet pigment, as a result of QS through its autoinducer N-hexanoyl homoserine lactone (McClean, 1997). Violacein production in the presence of BE was also gradually decreased in a dose-dependent manner (Fig. 1b), suggesting that BE is also capable of inhibiting QS of C. violaceum CV026.

Figure 1

Effects of BE on the production of AI-2 and bacterial growth. (a) Effect of BE on AI-2 production in Escherichia coli O157:H7. The Vibrio harveyi BB170 strain was used for the assay as described in the Materials and methods. The AI-2 level was expressed as a relative value for the AI-2 value of the supernatant from the culture of E. coli O157:H7 grown without BE. Data are presented as mean±SD (n=3). *P<0.01 vs. treatment with 0% BE. (b) Effect of BE on violacein production in the CV026 strain. The violacein level was expressed as a relative value to that of the control treatment (0% BE). Data are presented as mean±SD (n=3). The differences in the mean values among the treatment groups are statistically significant (P<0.05, anova). (c) Escherichia coli O157:H7 growth was monitored by measuring A600nm. LB media supplemented with 5% BE was used for bacterial growth. Data are presented as mean±SD (n=3). *P<0.05 vs. OD600nm values of growth with 0% BE. **P<0.01 vs. OD600nm values of growth with 0% BE.

To rule out the possibility that reduced production of AI-2 is a consequence of decreased bacterial growth, we examined whether or not BE exhibited any adverse effects on bacterial growth. Figure 1c compares the growth curves of E. coli O157:H7 during 8h cultures in LB without or with 5% BE. In our experiments, stationary phase was achieved after ∼6h of culture. Growth of E. coli O157:H7 was elevated by the addition of BE (Fig. 1c). The bacterial culture reached OD600nm of ∼5.0 after 6h of growth in plain LB, whereas bacterial cell density reached OD600nm of ∼5.7 in LB media amended with BE. Taken together, these results demonstrate that suppressed AI-2 production was not due to any secondary effects associated with retarded bacterial growth and occurred rather efficiently even at higher cell density.

Inhibition of swarming motility by BE

It has been reported that swarming motility is dependent on AI-2 signaling in E. coli O157:H7 (Sperandio, 2002). To test whether the reduced AI-2 synthesis by BE treatment is reflected in bacterial motility, a swarming motility assay was performed. As shown in Fig. 2, a gradual decrease in bacterial motility was clearly observed in the presence of increasing concentrations of BE. This result further verifies that BE specifically targets AI-2-mediated bacterial virulence pathways in E. coli O157:H7.

Figure 2

Effect of the BE on the motility of Escherichia coli O157:H7. The relative motility of E. coli O157:H7 in the presence of varying concentrations of BE.

BE downregulated norepinephrine-induced ler transcription

To elucidate the effect of BE on an AI-3-mediated QS system, we examined whether the activation of ler promoter by norepinephrine was also compromised by addition of BE. To address this question, we created a green fluorescent protein (GFP) reporter strain, in which the gfp gene was transcribed by the ler promoter. As shown in Fig. 3, green fluorescence intensity was increased ∼1.37 fold by the addition of norepinephrine (second vs. third bar). The addition of BE, however, decreased the norepinephrine-stimulated production of GFP significantly (fourth vs. third bar). This result suggests that BE can prevent the transcription of ler, regulated by AI-3-mediated QS system, from being activated and therefore may block a complex signaling cascade that regulates the expression of genes encoding proteins necessary for full virulence of E. coli O157:H7.

Figure 3

Effect of the BE on ler gene expression in Escherichia coli O157:H7. Green fluorescent intensity of E. coli O157:H7 harboring pLER-GFP upon treatments indicated at the bottom was presented as fold induction over control treatment. Green fluorescent intensity of E. coli O157:H7 harboring pAD123 (empty vector) was used as a control. Data are presented as mean±SD (n=3). *P<0.01 vs. NE treatment; **P<0.001 vs. BE treatment.

Increased survival of C. elegans fed on E. coli O157:H7 in the presence of BE

Next, we tried to determine whether BE could attenuate the virulence of E. coli O157:H7 in vivo using C. elegans as a host. Caenorhabditis elegans is used as a simple and economic invertebrate animal model for the study of mechanisms of microbial pathogenesis (Nicholas & Hodgkin, 2004; Sifri, 2005). In particular, it was reported that C. elegans is a good model organism to evaluate the virulence of E. coli O157:H7 and the antibacterial efficacy of many types of chemical compounds (Breger, 2007; Lee, 2008). As shown in Fig. 4, there were no significant differences in the survival rate of C. elegans for 2 days, but the survival rate of the nematodes fed on E. coli O157:H7 in the presence of 0.5% (v/v) of BE were significantly higher than those fed only on the pathogen for 3 days or more (Fig. 4). Notably, the survival rates of C. elegans fed on E. coli O157:H7 with 0% and 0.5% of BE after 8 days were 21.5% and 50%, respectively (Fig. 4). However, the survival rate of the nematodes fed on E. coli OP50, an avirulent strain routinely used as a nutrient source for C. elegans, was not affected by the presence of 0.5% BE (Fig. 4). These results suggest that BE can considerably protect the nematodes against a pathogenic attack by E. coli O157:H7, and thus, BE treatment can be developed as an agent to attenuate bacterial virulence in vivo.

Figure 4

Effects of BE on the lifespan of Caenorhabditis elegans. Caenorhabditis elegans glp-4 mutant strains were placed on NGM agar plates spread with aliquots of bacterial cultures (Escherichia coli O157:H7 or OP40) with either 0% or 0.5% BE. The survival rates were scored daily and expressed as percentages of survival. Three independent experiments each with 20 nematodes were conducted and the data were presented as mean±SD. *P<0.01 vs. O157:H7 plus BE (0.5%).

Modulation of virulence gene expression by flavonoid compounds included in the BE

We then examined the effects of BE on the expression of virulence-associated genes by qRT-PCR. We analyzed the transcript levels of luxS and pfS, because these two genes are critically involved in AI-2 synthesis (Gonzalez Barrios, 2006). We also tested flhD and eae, which are involved in flagella regulation and type III secretion, respectively (Hughes, 2009). As shown in Fig. 5a, a statistically significant decrease in the mRNA levels of luxS, pfs and flhD genes was observed in response to treatment with 0.5% BE, strongly suggesting that BE regulates the virulence of E. coli O157:H7 by modulating the transcription of virulence genes. Recently, it was reported that citrus flavonoids suppress an array of bacterial virulence mechanisms (Vikram, 2010). Because BE also contains flavonoids such as quercetin, kaempferol and myricetin (He, 2008; Schmidt, 2010), we sought to gain better insight into the active compound(s) that may cause the BE-induced virulence attenuation in E. coli O157:H7. To address this issue, we examined the effects of each of three flavonoid compounds (i.e. quercetin, kaempferol and myricetin) on the modulation of virulence gene expression by qRT-PCR. Each compound was used for treatment at the final concentration of 50μgmL−1 because a previous report clearly demonstrated that compounds at this concentration did not exert any adverse effects on bacterial growth (Vikram, 2010). As shown in Fig. 5b, transcript levels of all tested genes were decreased in response to treatment with quercetin or kaempferol, with quercetin being more effective than kaempferol. In contrast, heterogeneous transcriptional modulation was observed in bacteria treated with myricetin. Our qRT-PCR analysis indicates that expression of luxS and pfs genes was most affected by quercetin, while transcription of these two genes was not significantly influenced by myricetin. In addition, transcription of the eae gene was significantly suppressed by myricetin, but only mildly affected by kaempferol (Fig. 5b).

Figure 5

Effects of BE and three flavonoid compounds on the expression of virulence-associated genes in Escherichia coli O157:H7. (a) The relative expression levels of luxS, pfs, flhD and eae in response to treatment with 0.5% vs. 0% BE. The mRNA levels of each gene were adjusted to that of rrsD mRNA, which was used as the internal control. Three independent experiments were performed and the mean±SD values are displayed in each bar. A relative expression value of 1.0 indicates no change in response to BE treatment. *P<0.05 vs. treatment with 0% BE. (b) The relative expression levels of luxS, pfs, flhD and eae in response to the growth with indicated flavonoids (50μgmL−1). The experimental conditions were identical to those described for Fig. 5a. *P<0.05 vs. control treatment with dimethyl sulfoxide used to dissolve the compounds.


We have already entered an era in which antibiotic-resistant bacterial pathogens pose a huge threat to human health. Therefore, alternative approaches to inhibiting bacterial infection, besides antibiotic treatment, should be pursued to provide safer infection control. Because the production of virulence factors is dependent on QS in most human pathogens, QS has been a major target for alleviating bacterial virulence. To date, a large number of natural plants have been tested for their ability to antagonize bacterial QS. Extracts derived from marine alga, D. pulchra, interfered with the activation of QS-mediated gene expression in E. coli (Manefield, 1999). Vanilla extract (Choo, 2006) and Tremella fuciformis extract (Zhu & Sun, 2008) were both reported to inhibit violacein production in CV026. Moreover, extracts of six different south Florida plants decreased the production of QS-controlled virulence factors in Pseudomonas aeruginosa, an opportunistic human pathogen of clinical significance (Adonizio, 2008).

Being a rich source of isothiocyanates, an agent that can inhibit carcinogenesis (Conaway, 2002), broccoli has been widely tested for its anticancer activity (Mas, 2007). However, whether BE can exert an inhibitory effect on QS-mediated bacterial virulence has never been elucidated. Results presented in the present study suggest that BE treatment can potentially suppress a wide range of QS regulatory systems in enterohemorrhagic E. coli. Addition of 5% BE almost completely repressed the synthesis of AI-2, while exhibiting no negative effect on bacterial growth. This suggests that BE specifically interferes with the regulation of AI-2 synthesis and its downstream pathways, not bacterial growth per se. The suppression of AI-2 synthesis in E. coli O157:H7 was further corroborated by the finding that (1) AI-2-controlled motility was decreased accordingly and (2) transcript levels of the luxS and pfs encoding enzymes that regulate AI-2 synthesis were decreased by broccoli-derived flavonoids. Furthermore, we also demonstrated that BE repressed transcription of the ler gene, encoding a master regulator of LEE genes. Because LEE genes are regulated through the AI-3/norepinephrine QS system (Sperandio, 2003), this suggests that BE can also target the AI-3 specific QS mechanism.

QS-mediated bacterial virulence was successfully tested in an in vivo infection model using C. elegans as a host organism. It was demonstrated that a QS-deficient mutant of P. aeruginosa killed fewer nematodes than its parental strain did (Rasmussen, 2005). It was also shown that E. coli O157:H7 in the presence of exogenous AI-2 molecules killed more nematodes (Kim, 2007). Our results clearly indicated that (1) C. elegans fed on a nonpathogenic E. coli strain (OP50) lived longer than C. elegans fed on E. coli O157:H7 and (2) the addition of BE attenuated the virulence potential of E. coli O157:H7 towards the C. elegans. Therefore, our results suggest that BE can effectively protect the nematodes against bacterial infection by inhibiting bacterial QS.

The discovery that QS is inhibited by BE led us to identify the active compounds contained in BE. We first looked for the effect of flavonoid compounds reported to be present in large quantities in broccoli (He, 2008; Schmidt, 2010). The data described in Fig. 5 suggest that different flavonoid compounds may target different subsets of genes involved in virulence and thus, BE-induced virulence attenuation is likely the combined effect of various flavonoid compounds. Although other active compounds may be present beyond the three flavonoid compounds, we expect that the data presented herein will form the basis of further investigation to elucidate BE's mode of QS inhibition.

In conclusion, this report provides renewed interest in using BE as a food extract that can potentially inhibit both bacterial QS and infectivity. We anticipate that this strategy will provide an effective approach to controlling bacterial infection without imposing pressure towards selection for antibiotic resistance.


This work was supported by the National Research Foundation (NRF) grant funded by the Korea government (MEST) (No. 2009-0087951) to S.S.Y. and the National Research Foundation (NRF) grants funded by the Korean government (MEST) (SRC program No. 2011-001334 and Public welfare & Safety research program No. 2010-0020775) to S.P.


  • Editor: Ian Henderson


View Abstract