Secreted indole serves as a signal for expression of type III secretion system translocators in enterohaemorrhagic Escherichia coli O157 : H7

Many enteric bacteria including Escherichia coli produce indole from tryptophan, a reaction that is catalysed by tryptophanase (TnaA). Indole is exported by the AcrEF-TolC multidrug efflux system, and accumulates in culture media during the stationary phase (DeMoss & Moser, 1969; Kawamura-Sato et al., 1999; Wang et al., 2001). TnaA is strongly induced under alkaline conditions (Blankenhorn et al., 1999) and is regulated by catabolite repression (Deeley & Yanofsky, 1982). Therefore, enterobacteria may be exposed to high concentrations of indole when they infect alkaline regions in the body such as the pancreatic duct. To elucidate the behaviour of enteric bacteria at their site of infection and in their habitat, it is therefore important to understand the regulatory mechanism and the roles of indole signalling.

We previously reported that indole induces multidrug resistance as a result of its ability to stimulate transcription of E. coli drug exporter genes (mdtEF, acrD, mdtABC, acrEF, cusABC, emrKY and yceL) through multiple regulatory pathways [BaeS-BaeR and CpxA-CpxR (the bacterial two-component signal transduction pathway) and RpoS-Hfq[GadY]-GadX (a two-component signal transduction system-independent regulatory cascade)] during stationary phase (Hirakawa et al., 2005; Kobayashi et al., 2006). Other studies have shown that indole controls biofilm formation and the expression of genes (gabT, astD and tnaAB) for amino acid metabolism in E. coli K-12 (Di Martino et al., 2003; Lee et al., 2007b; Wang et al., 2001).

Enterohaemorrhagic E. coli (EHEC) causes severe infectious diseases, non-bloody diarrhoea, haemorrhagic colitis and haemolytic uraemic syndrome, which can lead to the death of the patient (Karmali, 1989; Nataro & Kaper, 1998). EHEC adheres to epithelial cells, triggering the formation of hallmark lesions called attaching and effacing (A/E) lesions at an early stage of infection (Nataro & Kaper, 1998). A/E lesions are characterized by the localized destruction of brush border microvilli, intimate attachment of the organism to the host cell membrane and formation of an underlying actin-rich structure in the host cell. The capacity to form A/E lesions is conferred by a type III secretion system (T3SS) and the proteins that it secretes (translocators and effectors), which are gene products of the locus of enterocyte effacement (LEE), a 36 kbp chromosomal pathogenicity island carrying 41 ORFs clustered in five major operons (LEE1 to LEE5) (McDaniel & Kaper, 1997). To counter EHEC pathogenesis, it is therefore important to identify the signals that induce secretion and/or expression of the secreted proteins and to elucidate the mechanism of regulation of LEE expression and type III secretion-related protein secretion under conditions similar to those at sites of infection.

When EHEC adhere to epithelial cells, secreted Tir (an LEE5 gene product) is translocated into the host cell membranes, where it serves as the receptor for intimin, an extracellular bacterial adhesin. EHEC contact host cells via the interaction of intimin and Tir receptor. EspA and EspB (LEE4 gene products) are translocators required for the targeting of Tir. Therefore, these proteins are essential for T3SS-dependent pathogenicity of EHEC.

The recent sequencing of the complete EHEC O157 : H7 genome has revealed an ORF homologous to tnaA that encodes tryptophanase in E. coli K-12. To investigate roles of indole in the T3SS-dependent pathogenicity of EHEC, we examined effects of tnaA deletion and of adding indole to a tnaA deletion mutant on the secretion and production of EspA, EspB and Tir and on A/E lesion production. We found that indole enhances secretion and production of EspA/B and also production of type III-dependent A/E lesions.

Bacterial strains, host cells and plasmids.
EHEC O157 : H7 Sakai (RIMD 0509952) (Hayashi et al., 2001), its derivative strains, and the plasmids used in this study are described in Table 1. Bacteria were grown in LB broth. Gene-deletion mutants of EHEC except for the hfq single-deletion mutant were constructed by a previously described pKO3-based gene-replacement method (Link et al., 1997). Basically, all the genes targeted for inactivation were replaced by a 33 bp linker sequence that retained the start and stop codons. The No and Co primers described in Table 2 anneal at about 500 bp upstream of the start codon and about 500 bp downstream of the stop codon, respectively. To verify the gene-deletion mutations, we performed PCR with No and Co primers, and the length of the respective PCR fragments was compared in the wild-type and the gene-deletion mutant. For construction of the hfq single-deletion mutant, the chloramphenicol-resistance cassette was inserted into the internal region of the hfq gene. A DNA fragment including the 300 bp upstream and 48 bp of coding region of hfq was cloned upstream of the chloramphenicol resistance gene in the suicide vector pYAK1 (Kodama et al., 2002). After homologous recombination, the inactivation of hfq was established. HeLa cells (RIKEN) were grown in Dulbecco's modified Eagle's medium (DMEM, Sigma) containing 10 % fetal bovine serum (FBS, Sigma) and 100 mg gentamicin l^–1 (Sigma) at 37 °C under 5 % CO₂ in air. Before infection, the HeLa cells were washed with PBS before the FBS and gentamicin-free DMEM were added.

Table 1. Bacterial strains and plasmids used in this study

Table 2. Oligonucleotides used for plasmid construction, and verification of gene replacement Bold type indicates the restriction enzyme sites.

A tnaA expression plasmid was constructed as follows. The tnaA gene including a sequence 220 bp upstream of the start codon was cloned into pTrc99A by using the primers listed in Table 2. This plasmid was named pTrc99tnaA. We expected that TnaA would be expressed under control of the native promoter and IPTG-inducible trc promoter. In the presence of excess IPTG, the growth rate was reduced, but addition of 10 µM IPTG did not affect the growth. Therefore, TnaA expression studies were done on cells grown with 10 µM IPTG. In the β-galactosidase assay, ΔlacZI and derivative strains lacking the chromosomal lacZ gene were used to eliminate the effects of endogenous β-galactosidase.

For construction of a hfq expression plasmid, the hfq gene including native promoter was amplified by PCR with hfq-F (gccggatcccactgttagtggg) and hfq-R (gccaagcttacagcccgaaaccttattcgg) primers. The fragment was ligated into pUC19 vector.

Indole production assay.
The extracellular indole concentration was determined by a previously described HPLC method (Kobayashi et al., 2006). EHEC and its derivative strains were grown at 37 °C in LB broth and pelleted by centrifugation at 20 000 g. The resulting supernatants were extracted twice with ethyl acetate and the extracts were loaded on to a Symmetry C₁₈ column (5 µm, 4.6x150 mm: Waters) attached to an L2130 HPLC system (Hitachi). The samples were eluted with acetonitrile/H₂O (1 : 1, v/v) at a flow rate of 0.8 ml min^–1; the indole peak was detected by its absorbance at 276 nm and its identity was confirmed by correspondence to the elution time for pure indole. The indole concentration was calculated from the ratio of the area of the detected peak to that of a standard peak.

Detection of EspA and EspB.
Cells were grown at 37 °C for 8 h (stationary phase) and separated by centrifugation and filtration. Secreted proteins were precipitated from the supernatants with 10 % TCA and dissolved in SDS sample buffer. Cell pellets were dissolved in SDS sample buffer and the solutions were boiled. The samples were analysed by 10 % SDS-PAGE. EspA and EspB were detected by Western blotting with an EspA-specific polyclonal antibody and EspB-specific antiserum as described previously (Cantarelli et al., 2007; Kodama et al., 2002). Protein bands of EspA and EspB were visualized on a LAS-3000 Luminescent Image Analyser (Fujifilm) or X-ray film, and they were quantified by Image J software (National Institutes of Health).

Detection of A/E lesions by fluorescence microscopy.
HeLa cells were grown on glass coverslips and infected for 4 h with 1 µl (3.2–3.4x10⁶ bacterial cells) EHEC from an 8 h shaking culture (stationary phase) in LB broth. After infection, the cells were washed five times with PBS, fixed with 3 % paraformaldehyde for 15 min, and permeabilized with 0.1 % Triton X-100 in PBS. F-actin was stained with rhodamine/phalloidin (1 U ml^–1, Molecular Probes) and the HeLa cell nuclei were visualized with Hoechst 33258 (1 µg ml^–1, Molecular Probes). The mounted samples were observed by fluorescence microscopy, and the number of A/E/ lesions formed on 100 randomly selected cells was counted. HeLa cells were counted in three microscopic fields; ± values indicate the range of A/E lesion numbers in three microscopic fields

Construction of LEE gene reporter plasmids.
The reporter plasmids were constructed as follows. DNA fragments containing the promoter region [513 bp upstream of the start codon for the first gene (ler of LEE1 and sepZ of LEE2) of the LEE1 and LEE2 operons and 506, 516 and 395 bp upstream of the start codon for the first gene (Ecs4570 of LEE3, sepL of LEE4 and tir of LEE5) of the LEE3, LEE4 and LEE5 operons, respectively] were amplified by PCR by using the primers listed in Table 2. These DNA fragments were cloned in front of the promoterless lacZ reporter gene in the single-copy pNN387 vector (Elledge & Davis, 1989). The resulting plasmids were introduced into EHEC and its derivative strains for β-galactosidase assays.

Reporter gene assays.
To determine the effects of indole and TnaA on the transcription of various reporter constructs, each bacterial strain was grown for 8 h (stationary phase) or 2 h (exponential phase) at 37 °C in LB broth containing indole and chloramphenicol. β-Galactosidase activity was assayed in cell lysates using ONPG as the substrate (Miller, 1992).

Indole production by EHEC O157 : H7 Sakai and the tnaA deletion mutant
To examine indole production by EHEC O157 : H7 Sakai and the tnaA deletion mutant (Ecs4645), extracellular concentrations were measured as described in Methods. When wild-type cells were grown for 8 h (stationary phase) in LB medium, the extracellular concentration of indole reached a maximum value of 465 µM and did not increase on further incubation (Fig. 1). In contrast, the tnaA deletion mutant produced no detectable indole. Growth of the ΔtnaA strain was the same as that of the wild-type (Fig. 1). We also examined the indole productivity of EHEC grown in tissue cell culture medium (DMEM). However, wild-type EHEC hardly produced indole (extracellular concentration was less than 10 µM; data not shown). Therefore, to investigate the effect of tnaA deletion and indole addition on EHEC pathogenicity, EHEC and its tnaA mutant were cultured in LB medium for 8 h in this study.

(15K): Fig. 1. Indole production (bars) and growth (triangles) of wild-type EHEC (open symbols) and the tnaA deletion mutant (solid symbols) cultured in LB medium. The extracellular indole concentrations were measured by HPLC, as described in Methods, after 2, 4, 8, and 24 h of culture. The EHEC cell density was determined as OD600. At least three independent experiments were performed in each case.

Effects of indole on the secretion and production of T3SS-dependent EspA and EspB
When EHEC infect epithelial cells, Tir secreted by EHEC is translocated to the plasma membrane of the host cell by EspA and EspB. Therefore, these proteins play central roles in T3SS-mediated EHEC pathogenicity. To examine the relationship between EHEC pathogenicity and indole, we assessed the effects of the tnaA deletion and indole addition on the secretion and production of EspA, EspB and Tir by immunoblotting with anti-EspA antibody, anti-EspB antiserum and anti-Tir antibody. The secretion of EspA and EspB was greatly decreased in the tnaA deletion mutant (Fig. 2) whereas that of Tir was hardly affected (data not shown). The tnaA expression plasmid pTrc99tnaA restored most of the EspA and EspB secretion in the ΔtnaA strain (Fig. 2). To determine whether the reduction of EspA/B proteins caused by tnaA deletion is due to the loss of indole, ΔtnaA was cultured in the presence and absence of indole. The secretion of EspA and EspB was increased to the wild-type level by addition of 125 or 250 µM indole (Fig. 2). Similarly, the amount of intracellular EspA/B was also decreased by the tnaA deletion, whereas it was restored by the addition of 125 or 250 µM indole (Fig. 2). The effect of tnaA deletion and indole addition on EspA/B secretion and production in DMEM was also examined. As in the case of LB medium, indole addition of 125 µM increased the amount of extracellular EspA/B by a factor of 1.5 (data not shown). However, the tnaA deletion did not decrease the amount of EspA/B because EHEC cultured in DMEM produces indole poorly. In contrast to the case of extracellular EspA/B, the amount of intracellular EspA/B hardly increased in response to indole addition while the intracellular EspA/B level in DMEM was higher than that in LB (data not shown). Therefore, the effect of indole may be masked in DMEM.

(59K): Fig. 2. Secretion and production of EspA and EspB in wild-type and ΔtnaA EHEC 8 h stationary-phase cultures. EspA and EspB were detected by Western blotting as described in Methods. Wild-type, ΔtnaA (in the presence or absence of indole), ΔtnaA harbouring the pTrc99A vector (mock) (containing 0.01 mM IPTG) and ΔtnaA harbouring the pTrc99tnaA plasmid (containing 0.01 mM IPTG) were grown for 8 h and secreted and intracellular EspA/B were measured. As a control, intracellular DnaK was detected by means of a DnaK-specific antibody (Calbiochem). The respective protein bands were visualized on a LAS-3000 Luminescent Image Analyser (Fujifilm), and quantified by using Image J software; values indicate the ratio against the parent strain.

To determine whether indole stimulates only the production of EspA/B, or both the production and secretion of these proteins, the effect of indole on a T3SS mutant was investigated. The T3SS mutant used, O157ΔescN, lacks the gene encoding the ATPase of the T3SS. Thus O157ΔescN cannot secrete translocators or effectors (Cantarelli et al., 2007). In the escN deletion mutant, intracellular levels of EspA/B were decreased by the tnaA deletion, and increased by the addition of indole (Fig. 3). However, the influence of the tnaA deletion and indole addition in O157ΔescN was less than that in the parent strain, indicating that indole stimulates not only EspA/B production but also secretion (Fig. 3).

(39K): Fig. 3. Secretion and production of EspA and EspB in ΔescN and ΔescNΔtnaA EHEC strains cultured in LB medium for 8 h (stationary phase). ΔescN and ΔescNΔtnaA (in the presence or absence of indole) were cultured and extracellular and intracellular EspA and EspB were detected. As a negative control, intracellular DnaK was detected by means of a DnaK-specific antibody (Calbiochem). The respective protein bands were visualized and quantified as described for Fig. 2.

Effect of indole on LEE promoter activity in EHEC
Type III secretion-related genes are located in the locus of enterocyte effacement (LEE), a pathogenicity island that comprises five major operons (LEE1 to LEE5). Among them, EspA and EspB are encoded on the LEE4 operon. To examine whether indole induces the expression of LEE4 genes, we measured the promoter activity by using the reporter in the wild-type strain and the tnaA deletion mutant in the presence or absence of indole. The promoter activity of LEE4 in the tnaA deletion mutant was lower than that in the wild-type strain, and was restored by the addition of 125 µM indole (Fig. 4a). The promoter activity of LEE4 was increased up to fourfold by 2 mM indole. We also measured the promoter activity of other LEE genes. Activity of the LEE1 promoter was reduced by the tnaA deletion; however, activity was not restored by the addition of indole at concentrations of less than 500 µM. In the presence of 2 mM indole, the activity was increased twofold (data not shown). The activity of LEE3 was also decreased by the tnaA deletion, and increased by the addition of 500 µM indole. However, these effects were modest (data not shown).The activities of the LEE2 and LEE5 promoters were not affected by either tnaA deletion or the addition of indole (data not shown).

(19K): Fig. 4. (a, b) Effect of indole and tnaA gene deletion on LEE4 promoter activity. (a) Wild-type (ΔlacZI) and ΔtnaA (ΔlacZIΔtnaA) EHEC were grown to stationary phase in LB medium with (solid bars) or without (open bars) indole. (b) Wild-type (ΔlacZI) and ΔtnaA (ΔlacZIΔtnaA) EHEC cells were grown to exponential phase with (solid bars) or without (open bars) indole. (c) Effects of deletion of ler, baeR, cpxR, gadX and hfq on indole-induced expression of LEE4. The respective mutant strains were grown to stationary phase with (solid bars) or without (open bars) 2 mM indole. The β-galactosidase activities of the lacZ fusion of the LEE4 promoter were measured. At least three independent experiments were performed in each case.

We also examined whether the LEE4 promoter is stimulated by indole in the wild-type background. TnaA is induced in stationary phase. Therefore, the wild-type and ΔtnaA strains were cultured to exponential phase (for 2 h) in the presence or absence of indole to avoid the effect of intrinsic indole produced by wild-type (Fig. 1), and the activities of the LEE4 promoter were measured. As expected, deletion of tnaA did not reduce activity of the LEE4 promoter, and addition of 500 µM or 2 mM indole efficiently enhanced it (Fig. 4b). We believe the primary effect of indole on LEE gene transcription is at the LEE4 promoter.

Next, we investigated the mechanism of regulation of LEE4 expression by indole. We previously reported that the induction of drug exporter genes by indole depends on BaeSR/CpxAR (a two-component signal transduction system in E. coli), GadX (an acid-induced regulatory protein) and Hfq (sRNA chaperone protein) (Kobayashi et al., 2006; Hirakawa et al., 2005). It has also been reported that Ler is a positive regulator of LEE genes in EHEC (Elliott et al., 2000). Therefore, BaeSR, CpxAR, GadX, Hfq and Ler are possible indole signal-transducing mediators in the induction of LEE4 expression. Appropriate deletion mutants were constructed in the ΔtnaA background, and the effects of indole on LEE4 promoter activity were examined. As shown in Fig. 4(c), LEE4 expression was higher in the hfq deletion mutant (ΔlacZIΔtnaAΔhfq) than in the hfq intact strain (ΔlacZIΔtnaA), and it was not significantly induced by indole (1.1-fold induction). The levels of induction by indole in the other mutants were almost the same as in the parent strain (Fig. 4c). To examine the effect of hfq deletion on EspA/B (gene products of LEE4) production and the sensitivity to indole, levels of secreted and intracellular EspA/B were determined when these mutant strains were cultured in the presence and absence of exogenous indole. The hfq deletion mutant (ΔlacZIΔtnaAΔhfq) produced large amounts of EspA/B even in the absence of exogenous indole. However, levels of neither secreted EspA/B nor intracellular EspA/B were increased by the addition of indole (Fig. 5). These observations are consistent with the results of the promoter activity measurements. The results indicate that Hfq represses the expression of LEE4 while it is responsible for the induction of LEE4 by indole.

(56K): Fig. 5. Secretion and production of EspA and EspB in ΔlacZI, ΔlacZIΔtnaA and ΔlacZIΔtnaAΔhfq EHEC. Strains were cultured in LB medium for 8 h (stationary phase) in the presence or absence of indole, and levels of secreted and intracellular EspA and EspB were detected. As a negative control, intracellular DnaK was detected by means of a DnaK-specific antibody (Calbiochem). The respective protein bands were visualized and quantified as described for Fig. 2.

Since the effect of hfq deletion on EspA/B secretion and production had been examined in the tnaA deletion background, we tested whether a single hfq mutant also increases EspA/B secretion and production. A large amount of intracellular and extracellular EspA and EspB was detected in the Δhfq : Cm strain carrying pUC19 vector. However, when the hfq expression plasmid (pUC19hfq) was introduced, the amount of EspA and EspB was decreased to about the wild-type level (Fig. 6).

(43K): Fig. 6. Secretion and production of EspA and EspB in wild-type and hfq single-mutant (Δhfq : Cmr) EHEC strains cultured in LB medium for 8 h (stationary phase). Wild-type and Δhfq : Cmr harbouring the pUC19 vector (mock), and Δhfq : Cmr harbouring the pUC19hfq (hfq expression) plasmid were grown for 8 h and secreted and intracellular EspA/B were measured. The respective protein bands were visualized on X-ray film, and quantified by using Image J software; values indicate ratio against the parent strain.

Effects of indole on the formation of A/E lesions in HeLa cells
Because the secretion and production of EspA/B are stimulated by indole, T3SS-dependent EHEC pathogenicity, which is characterized by accumulation of actin in the host cells (A/E lesions), may also be stimulated by indole. Therefore, the effects of tnaA deletion and the addition of indole on the formation of A/E lesions in HeLa cells were examined. HeLa cells were infected for 4 h with stationary-phase wild-type EHEC or with indole-treated or untreated ΔtnaA mutant. The infected cells were fixed and stained with rhodamine-labelled phalloidin and Hoechst 33258 to identify actin and nuclei, respectively. The A/E lesions, which are characterized by accumulated F-actin, were then counted. When the HeLa cells were infected with wild-type EHEC, 14±3 A/E lesions per 100 HeLa cells were observed 4 h post-infection, whereas A/E lesions were rarely observed in uninfected cells (<1 A/E lesion per 100 HeLa cells). When HeLa cells were infected with ΔtnaA EHEC, the number of A/E lesions produced was about half the wild-type value (6±2 A/E lesions per 100 HeLa cells). However, when ΔtnaA EHEC was cultured for 8 h (stationary phase) with 125 µM indole before infection, the number of A/E lesions reached about 18±3 per 100 HeLa cells, indicating that indole restores and promotes the induction of A/E lesions by ΔtnaA EHEC in HeLa cells. Homologues of E. coli K-12 tnaA are found in many bacteria, and indole-producing bacteria are found in the human intestinal tract (DeMoss & Moser, 1969). It was previously reported that tnaA deletion in EPEC abrogates the ability to paralyse or kill Caenorhabditis elegans (Anyanful et al., 2005). Although exogenous indole induces LEE1 expression, the reduction of virulence in the tnaA mutant was not restored by indole. On the other hand, it was recently reported that indole represses several EHEC virulence-related phenotypes (motility, biofilm formation and attachment to HeLa cells) (Bansal et al., 2007). However, the relationship between indole and type III secretion, and the formation of A/E lesions, has not been investigated previously.

We show here that indole increases the production and secretion of EspA/B, which are T3SS-mediated translocators. This results in an increase in the formation of A/E lesions in HeLa cells. Indole also elicits a modest induction of LEE4 transcription. We found that deletion of tnaA reduced the secretion and production of EspA/B and the formation of A/E lesions in HeLa cells (Fig. 2). All of these effects were overcome by addition of 125 µM indole to O157 ΔtnaA cultures. This is a physiologically relevant indole concentration: in wild-type EHEC culture medium in the stationary phase, the indole concentration is about 465 µM (Fig. 1). Therefore, the effect of tnaA deletion can be explained by the loss of indole (Fig. 2).

Experiments using a T3SS mutant, ΔescN, showed that indole stimulates the production of EspA and EspB without secretion and the deletion of tnaA decreases cellular EspA and EspB levels (Fig. 3). However, the influence of indole addition and the tnaA deletion in ΔescN was less than that in the parent strain. We also showed that the promoter activity of espA and espB (located in the LEE4 operon) was moderately decreased by the deletion of tnaA, and induced by indole (Fig. 4a). The change of promoter activity was less than that of EspA and EspB protein levels. Thus, indole participates in three kinds of regulatory process of EspA/B. Among them, indole primarily controls the secretion of EspA/B. Indole also appears to control EspA and EspB translation and at the level of transcription.

We also found that deletion of hfq abolishes the influence of indole on the induction of LEE4, and fully induces it even when indole is absent. These results suggest that Hfq tightly represses the expression of LEE4. Hfq acts as a regulator in complex with sRNAs (Gottesman, 2004), which determine the specificity of regulation. In addition, we note that the effect of hfq deletion on the increase of EspA/B production is greater than that on the increase of the promoter activity (Figs 4c, 5 and 6). We interpret this to mean that the LEE4 operon including espA and espB may be transcriptionally and post-transcriptionally regulated by Hfq-sRNA. It will be of interest to identify the hypothetical sRNA for regulation of EspA and EspB.

It was previously reported that motility-related genes and LEE operons are inversely regulated by GrlA and GrlR (Iyoda et al., 2006). Therefore, GrlA and GrlR could be considered as candidates mediating the induction of LEE genes and the repression of motility by indole. However, this seems unlikely because the LEE gene induction by indole is independent of Ler (Fig. 4c) whereas the regulation of LEE promoters by GrlA and GrlR depends on Ler (Iyoda et al., 2006).

In EHEC, effector proteins are delivered into host cells through EspA filaments and pores in the plasma membrane formed by EspB. Intimin, an outer-membrane protein of EHEC, then binds to Tir on the host-cell plasma membrane (Garmendia et al., 2005; Kenny et al., 1997). Although indole obviously enhances the secretion of EspA and EspB, the production of Tir and intimin (the eae gene product) is not affected (data not shown). This suggests that indole probably assists the efficient delivery of effector proteins, promoting A/E lesion formation.

If indole is produced in stationary phase in LB medium, it is anticipated that the indole-regulated LEE4 promoter should be induced in stationary phase. The activity of this promoter in exponential phase is clearly lower than in stationary phase. However, LEE4 promoter activity of a tnaA deletion mutant in stationary phase and LEE4 in exponential phase is reasonably high (Fig. 4a, b). Therefore, we believe that indole partly contributes to the stimulation of LEE gene expression in the stationary phase.

There are many recent reports that intercellular signal molecules contribute to the pathogenicity of bacteria (Bassler, 2002). In Gram-negative bacteria, three kinds of quorum-sensing molecules (or autoinducers) (AI-1, AI-2 and AI-3) have been reported (Fuqua et al., 2001; Sperandio et al., 2001, 2003). In addition, there is evidence that indole also acts as an intercellular signal molecule during the stationary phase in E. coli and other enteric species (Kobayashi et al., 2006; Lee et al., 2007a; Wang et al., 2001). Here we show that indole enhances expression of specific virulence factors in EHEC. Previously, the relationship between intercellular signal transduction and the production of type III secretion-related proteins was described in EHEC. A LuxR-like quorum-sensing regulator, SdiA, AI-3 and related regulatory factors (luxS, qseA and qseBC) affect the production of type III secretion-related proteins (Kanamaru et al., 2000; Sharp & Sperandio, 2007; Sperandio et al., 1999, 2003). However, the addition of indole did not affect the transcription of the related genes (sdiA, luxS, qseA and qseBC) (data not shown) although indole activates the transcription of sdiA, and results in reduction of biofilm formation in E. coli K-12 (Lee et al., 2007b). In addition, indole is produced later in the stationary phase than AI-1, AI-2 and AI-3, which are active in the early stationary phase. Therefore, the indole-sensing system probably differs from these intercellular signal transduction systems.

As mentioned above, it was previously reported that indole represses several virulence-related phenotypes in EHEC (motility, biofilm formation and attachment to HeLa cells) (Bansal et al., 2007). Addition of 500–600 µM indole decreased these phenotypes whereas we showed that type III secretion-related protein production and virulence phenotypes are stimulated by indole concentrations of 125 µM or more. These results may indicate that indole has dual roles in the virulence of EHEC. Judging from these observations, the virulence of EHEC seems to be tightly regulated by the concentration of indole. Indole concentration in the enteric site may change intermittently with the amount of indole-producing enteric bacteria and enteric environmental conditions. According to an in vivo study with mice, tryptophan in the food markedly affects tryptophanase, and hence presumably indole production, by enteric bacteria (Botsford & Demoss, 1972).

To counteract bacterial infections, it is important to understand the behaviour of bacteria at their site of infection and in their habitat. Although indole is secreted by many bacteria and is abundant in the intestinal tract, its significance in bacterial physiology has remained unclear. We therefore believe that the characterization of the indole signal and the identification of an indole regulon have great interest. Moreover, interestingly, serotonin, melatonin and indole-3-acetic acid, which are typical signal molecules secreted by animal and plant cells, have an indole-like chemical structure. Therefore, it will be intriguing to investigate how these indole-like signals act on bacteria, or how indole affects the signal transduction in eukaryotic hosts.

We wish to thank E. Peter Greenberg for constructive comments, George M. Church for the plasmid pKO3, and Ronald W. Davis for the plasmid pNN387. We are grateful for the technical assistance in HPLC analysis provided by Nobuo Kato and Hajime Nitta of the Institute of Scientific and Industrial Research, Osaka University. H. H. was supported by a research fellowship from the Japan Society for the Promotion of Science for Young Scientists. This research was supported by a Grant-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and CREST, Japan Science and Technology Agency, Japan.

Edited by: B. Kenny