The Per regulon of enteropathogenic Escherichia coli : identification of a regulatory cascade and a novel transcriptional activator, the locus of enterocyte effacement (LEE)-encoded regulator (Ler)

Authors


James B. Kaper E-mail jkaper@umaryland.edu; Tel. (+1) 410 706 3004; Fax (+1) 410 706 0182.

Abstract

Enteropathogenic Escherichia coli (EPEC) is the prototype organism of a group of pathogenic Gram-negative bacteria that cause attaching and effacing (AE) intestinal lesions. All EPEC genes necessary for the AE phenotype are encoded within a 35.6 kb pathogenicity island termed the locus of enterocyte effacement (LEE). The LEE encodes 41 predicted open reading frames (ORFs), including components of a type III secretion apparatus and secreted molecules involved in the disruption of the host cell cytoskeleton. To initiate our studies on regulation of genes within the LEE, we determined the genetic organization of the LEE, defining transcriptional units and mapping transcriptional start points. We found that components of the type III secretion system are transcribed from three polycistronic operons designated LEE1, LEE2 and LEE3. The secreted Esp molecules are part of a fourth polycistronic operon designated LEE4. Using reporter gene fusion assays, we found that the previously described plasmid-encoded regulator (Per) activated operons LEE1, LEE2 and LEE3, and modestly increased the expression of LEE4 in EPEC. Using single-copy lacZ fusions in K-12-derived strains, we determined that Per only directly activated the LEE1::lacZ fusion, and did not directly activate the other operons. Orf1 of the LEE1 operon activated the expression of single-copy LEE2::lacZ and LEE3::lacZ fusions in trans and modestly increased the expression of LEE4::lacZ in K-12 strains. Orf1 was therefore designated Ler, for LEE-encoded regulator. Thus, the four polycistronic operons of the LEE that encode type III secretion components and secreted molecules are now included in the Per regulon, where Ler participates in this novel regulatory cascade in EPEC.

Introduction

Enteropathogenic Escherichia coli (EPEC) is a leading cause of infantile diarrhoea in developing countries and is the prototype organism of a group of pathogenic bacteria that cause attaching and effacing (AE) intestinal lesions (Nataro and Kaper, 1998). AE lesions are characterized by disruption of the epithelial cytoskeleton and formation of pedestals that protrude out from the host cell surface to cup the bacterium (Moon et al., 1983; Nataro and Kaper, 1998). In addition to EPEC, a variety of other bacterial pathogens are capable of forming AE lesions. They include enterohaemorrhagic E. coli (EHEC) serotype 0157:H7, which cause bloody diarrhoea and haemolytic uraemic syndrome (HUS) (Tzipori et al., 1986), Hafnia alvei, which cause diarrhoea in children (Albert et al., 1992), the mouse pathogen Citrobacter rodentium (Schauer and Falkow, 1993) and REPEC (rabbit enteropathogenic E. coli ), which cause diarrhoea in rabbits (Robins-Browne et al., 1994).

All genes necessary for the AE phenotype in EPEC, the best studied of the AE pathogens, are encoded within a 35.6 kb pathogenicity island termed the locus of enterocyte effacement (LEE) (McDaniel and Kaper, 1997), which contains 41 predicted open reading frames (ORFs; Elliott et al., 1998). The EPEC LEE can be divided into three regions (for a review, see Nataro and Kaper, 1998; Frankel et al., 1998). First, the eae and tir genes encode intimin and Tir respectively; intimin is necessary for intimate attachment to epithelial cells (Jerse et al., 1990), and Tir serves as a receptor for intimin after translocation to the host cell membrane (Kenny et al., 1997a). A second region encodes the type III secretion apparatus (Jarvis et al., 1995), and a third region encodes the Esp molecules, or E. colisecreted proteins, involved in the formation of a translocon for delivering effector molecules to the host cell (Knutton et al., 1998) and disruption of the cytoskeleton in the formation of the AE lesions (Donnenberg et al., 1990; Kenny and Finlay, 1995). The components of the type III secretion apparatus are encoded in three putative operons, and the secreted Esp molecules are encoded in a fourth putative operon (McDaniel and Kaper, 1997; Elliott et al., 1998).

At least five proteins are known to be secreted by the EPEC type III secretion system. These include EspA, EspD and EspB, which are required for cytoskeleton disruption, and Tir (Foubister et al., 1994; Kenny et al., 1996; Lai et al., 1997). EspF was recently shown to be secreted by the type III system. However, a mutation in this locus did not affect host cell signalling or the ability to form AE lesions (McNamara and Donnenberg, 1998). Because of the complexity of the LEE, we expected to observe co-ordinate regulation. Consistent with this idea, the secretion of EspA, EspD, EspB and Tir is increased by the presence of Per (plasmid-encoded regulator), located on the EAF plasmid of EPEC (Kenny and Finlay, 1995).

Studies on the control of EPEC gene expression at the level of transcription have demonstrated that Per activates transcription of the LEE-located eae encoding the adhesin intimin (Gómez-Duarte and Kaper, 1995) and the EAF plasmid-located bfp operon encoding the bundle-forming pilus (Tobe et al., 1996; Bustamente et al., 1998). The per locus, which includes perA, perB and perC, is thought to be transcribed as a single polycistronic message (Gómez-Duarte and Kaper, 1995), and all three genes are necessary for full transcriptional activation of both eae and the bfp operon. PerA is a member of the AraC-like family of transcriptional activators (Gómez-Duarte and Kaper, 1995). PerB shows homology to eukaryotic DNA-binding proteins, and PerC shows no homology to protein sequences currently in the database. Because of the observation that Per increases the secretion of Tir and the Esp molecules and the fact that AraC-like proteins are known to regulate the type III secretion systems of other Gram-negative pathogens (for a review, see Hueck, 1998), we hypothesized that Per also regulates the genes encoding the type III system and the Esp molecules of EPEC at the level of transcription. Therefore, the goals of this study were to (i) determine the operon structure of the genes encoding the type III secretion apparatus and secreted molecules of the EPEC LEE; (ii) identify transcriptional start points of these operons; and (ii) determine whether Per increased the expression of these loci.

Results

Operon structure

The spacing of the LEE genes encoding the EPEC type III secretion apparatus and secreted molecules suggested the existence of at least four polycistronic operons (McDaniel and Kaper, 1997; Elliott et al., 1998). To identify promoter sequences useful for the construction of defined operon fusions to reporter genes, we determined the operon structure of the LEE by reverse transcriptase–polymerase chain reaction (RT–PCR) analysis (Stanier et al., 1997). Whole-cell RNA was isolated from strain E2348/69, and cDNA was synthesized as template. Paired primers were used to amplify specific amplicons of the LEE from cDNA. Because the putative operons are at least 4 kb in length (Fig. 1), PCR was used to amplify multiple amplicons for each putative polycistronic operon. Those genes included in a single amplicon indicated transcriptional coupling, and overlapping amplicons indicated that the genes included in the multiple amplicons were transcribed as a single polycistronic message.

Figure 1.

. Genetic organization of the EPEC LEE. Operons LEE1, LEE2, LEE3 and LEE4 and oligonucleotide primers used for RT–PCR and PCR are noted. Operons and individual genes with identified transcriptional start sites are also noted. A transcriptional start point for eae has been defined previously (Gómez-Duarte and Kaper, 1995). esc (E. colisecretion), esp (E. colisecreted proteins) and ces (chaperone for E. colisecreted proteins) genes are labelled, reflecting the nomenclature of conserved type III secretion components of other Gram-negative pathogens. This figure is a modified version of the genetic map of the EPEC E2348/69 LEE (Elliott et al., 1998). The complete sequence of the LEE is available (accession number AF022236).

As expected from the predicted operon structure, RT–PCR analysis demonstrated that orf1 and escS were transcriptionally coupled (primers K325 and K440) (Fig. 2A). Similar to orf1 and escS, escR and escU were found to be transcriptionally linked using primers K903 and K904 (Fig. 2A). Because escR lies 5′ to escS, we concluded that orfs1 to 5 and escRSTU are transcribed as a single polycistronic message. Continuing this analysis, we demonstrated that sepZ and escC (K510 and K243), and escC and cesD (K518 and K509) were transcriptionally coupled (Fig. 2A). Thus sepZ, rorf8, escJ, rorf6, escC and cesD are transcribed as a single polycistronic mRNA. For the third putative operon, orf12 and escN (K956B and K210) and escN and orf18 (K236 and K169) were transcriptionally coupled (Fig. 2A). Thus, we concluded that orf12, escVN, orf15, orf16, sepQ and orf18 are also transcribed as a single polycistronic message.

Figure 2.

. RT-PCR analysis. A. Putative LEE1, LEE2 and LEE3 operons. B. Putative LEE4 operon. RNA was extracted from E. coli E2348/69 and subjected to RT–PCR or cDNA amplification. A direct PCR step was carried out in parallel on E2348/69 DNA to ensure that specific products were amplified. Lanes labelled ‘−’ are controls for DNA contamination in RNA preparations. After treatment with DNase, RNA samples were combined with reverse transcriptase using the same conditions as for cDNA synthesis. PCR products were not visualized in the ‘−’ control lanes.

Molecules secreted by the type III secretion apparatus of E2348/69, including espA, espB and espD, are encoded within a fourth putative polycistronic operon in the LEE (Fig. 1). By RT–PCR analysis, we demonstrated that orf23 and espA (K909 and K910), espA and espB (K551 and K908) and espD and espF (K647 and K957B) were transcriptionally coupled (Fig. 2B), thus identifying a fourth operon that included orf23, espADB, orf27, escF, orf29 and espF.

Identification of transcriptional start sites

To identify transcriptional start points (tsps) of promoter sequences driving expression of the putative LEE operons described above, we performed primer extension analysis. Oligonucleotide primers were designed such that tsps located between 5 bp and 250 bp upstream of the translational start codon of the first gene of each putative operon could be identified. The tsp of the first operon was at position 3913 (Figs 3A and 4), 175 bp upstream of the translational start of orf1. Although we observed multiple bands that might represent additional tsps, we designated the tsp as position 3913 because no other potential σ70-dependent promoter sequence could be identified; additional bands were most probably products of premature termination of reverse transcriptase. The tsp of the second operon was at position 14769 (Figs 3B and 4), 110 bp 5′ to the translational start of sepZ. Similarly, the minor band seen did not correspond to a reasonable σ70-dependent promoter sequence. The tsp of the third operon was at position 14786 (Figs 3C and 4), 60 bp upstream of the translational start of orf12. A direct repeat sequence in the LEE3 spacer region was also noted (Fig. 4). Lastly, the tsp of the operon encoding the secreted molecules was at position 29165 (Figs 3D and 4), 80 bp upstream of the translational start of orf23.

Figure 3.

. Primer extension analysis. Whole-cell RNA was isolated from E2348/69, JPN15, an EAF plasmid-cured derivative of E2348/69, and JPN15 (pJLM161) containing perABC on a multicopy plasmid. Primer extension reactions were performed using template mRNA isolated from JPN15 (lane 1), E2348/69 (lane 2) and JPN15 (pJLM161) (lane 3) and oligonucleotide primers listed in Table 2 to identify tsps for the four operons. A. LEE1. B. LEE2. C. LEE3. D. LEE4. Sequencing reactions were performed using purified plasmid templates and the same oligonucleotide as in each of the respective primer extension reactions. Transcriptional start points (tsps) are indicated by arrows. See Fig. 4 for identified promoter sequences and their sequence position in the EPEC LEE.

Figure 4.

. Identified promoter sequences of the EPEC LEE. Transcriptional start sites were identified by primer extension analysis, noted as + 1 and underlined. −35 and −10 hexamers are indicated, and the canonical sequence of σ70-dependent promoters is provided (Harley and Reynolds, 1987). The position of +1, the start of transcription for each operon, in the published EPEC E2348/69 LEE sequence (Elliott et al., 1998) is underlined. Note the direct repeat sequence in the spacer of the LEE3 promoter.

In addition to the tsps of the four LEE operons, we identified promoter sequences driving the expression of escD, a component of the type III secretion apparatus (Elliott et al., 1998; Kresse et al., 1998), rorf3 and orf10 by primer extension (Fig. 4). The analysis elucidated unusual operon organization, namely three pairs of divergently transcribed, overlapping promoters: LEE2 and LEE3, escD and LEE4, and rorf3 and orf10.

Terminator analysis

Potential transcriptional terminators for the four polycistronic operons were predicted using the terminator program of the Genetics Computer Group (GCG) (Devereux et al., 1984). This analysis identified a potential terminator sequence downstream of escU at position 9239, the final gene in the first operon (Fig. 1). The second operon contains a potential transcriptional terminator just downstream of its final gene, cesD, at position 10859. For the third operon, we identified a potential terminator sequence at position 21084 downstream of orf18.

Using the GCG terminator program, we identified two potential terminators downstream of the final gene of the LEE4 operon, espF. In addition, a potential terminator was identified within orf27, located downstream of espB. It is certainly possible that orf23espADB are transcribed as a polycistronic message with transcriptional termination occurring in orf27. However, we were not able to identify a potential σ70-dependent promoter sequence 5′ to escF by inspection, and RT-PCR analysis clearly demonstrated that espD and espF were transcriptionally coupled (Fig. 2B; K647 and K957B). Thus, we determined that the transcriptional unit including orf23, espADB, orf27, escF, and orf29 and espF was a fourth polycistronic mRNA.

Regulation of LEE genes by Per in EPEC

Primer extension analysis provided preliminary evidence that the LEE1 and LEE2 operons were positively regulated by Per in EPEC (Fig. 3A and B). Whole-cell RNA isolated from wild-type EPEC strain E2348/69 used for this analysis resulted in bands of greater intensity at the tsps of LEE1 and LEE2 when compared with those generated from RNA of the EAF plasmid-cured derivative JPN15 (Fig. 3A and B, lanes 1 and 2). Providing per on the multicopy plasmid pJLM161 in JPN15 further increased the intensities of the bands (lanes 2 and 3), suggesting dosage-dependent regulation. per in trans did not alter the tsp for any of the operons subjected to primer extension analysis (Fig. 3 and data not shown). As a positive control, it was shown that the plasmid pJLM161 activated expression from the eae::phoA fusion in JPN15.96 (data not shown), consistent with previously reported results (Gómez-Duarte and Kaper, 1995).

Because the sepZ-containing LEE2 operon appeared to be regulated by Per (Fig. 3B), per-containing plasmids pJLM170 and pJLM171 were transformed into the previously constructed, single-copy sepZ::phoA fusion strain 30-5-1(3) (Donnenberg et al., 1990), and alkaline phosphatase activities were monitored to confirm our initial observation. Both the 3.9 kb fragment containing per in pJLM170 (the identical per fragment contained in pJLM161) and the per minimal clone in pJLM171 (containing only perA, perB and perC ) activated expression from the sepZ::phoA fusion in 30-5-1(3) (Fig. 5A). In 30-5-1(3), increasing the level of Per activated the sepZ::phoA fusion from ≈50 to 200 units of alkaline phosphatase activity, an increase of fourfold. As a positive control, both pJLM170 and the per minimal clone in pJLM171 increased transcriptional activity derived from the eae::phoA fusion in JPN15.96 (data not shown). Thus, consistent with the primer extension data, Per activated expression from the LEE2 operon in EPEC.

Figure 5.

. Effect of Per on the expression of LEE operons in EPEC. A. Strain 30-5-1(3) is E2348/69 containing a sepZ::phoA fusion and the EAF plasmid, and thus wild-type per (Donnenberg et al., 1990). sepZ is the first gene of the LEE2 operon (see Fig. 1). Plasmids pJLM170 and pJLM171 contain a 3.9 kb per fragment and a perABC minimal clone, respectively, in pBR322. B. Plasmids containing LEE1::lacZ, LEE3::lacZ and LEE4::lacZ fusions in pJLM164, pJLM172 and pJLM165, respectively, were transformed into the EAF plasmid-cured derivative of E2348/69, JPN15. Either pJLM161 containing per or the plasmid control pACYC184 were then transformed into the strains containing the multicopy lacZ fusions. The JPN15 (pJLM164) fusion strain depicted in the figure also contained the chloramphenicol-resistant pBAD33 as a plasmid control. The strain containing pJLM164 alone was unstable (data not shown).

To determine whether Per also activated expression of LEE1, LEE3 and LEE4 in the native pathogen, EPEC, defined regulatory fragments from these operons were fused to the lacZ reporter gene in plasmid pRS551 (Simons et al., 1987). Multicopy lacZ fusions in pJLM164, pJLM172 and pJLM165, which contained no less than 425 bp upstream of the start of transcription, were transformed into JPN15, an EAF plasmid-cured derivative of E2348/69 (Nataro et al., 1987), JPN15 (pACYC184) and JPN15 (pJLM161). Per activated the expression of the LEE1::lacZ fusion in pJLM164 from ≈6000 to 15 000 Miller units, about 2.5-fold (Fig. 5B). Per activated the expression of the LEE3::lacZ fusion in pJLM172 ≈fivefold, increasing expression from 2000 to 10 000 Miller units. Expression from the multicopy LEE4::lacZ fusion in pJLM165 was increased modestly, less than twofold, by Per. For all three multicopy fusions, we observed a relatively high basal level of β-galactosidase activity in the absence of the activator Per.

Effect of Per on LEE gene expression in E. coli K-12

Next, we wished to determine whether Per affected the expression of the four defined operons of the LEE in the absence of additional EPEC-specific regulatory components. For these experiments, regulatory fragments from the four operons fused to the lacZ reporter gene in pRS551 were transferred onto the chromosome of the E. coli K-12 strain MC4100 in single copy using the specialized transducing phage λRS45 as described previously (Simons et al., 1987). per-containing plasmids pJLM170 and pJLM171 were then transformed into the LEE1::lacZ, LEE2::lacZ, LEE3::lacZ and LEE4::lacZ operon fusion strains JLM164, JLM166, JLM172 and JLM165, respectively, and β-galactosidase activities were determined.

Per expressed from the multicopy plasmid pJLM170 activated expression from the LEE1::lacZ fusion in a K-12 background from 200 to 1200 Miller units in strain JLM164 (Fig. 6), consistent with the multicopy lacZ fusion assays (Fig. 5B). Activation by the minimal per clone was less than that of the larger 3.9 kb fragment containing per, but activation occurred nonetheless. In contrast to the activation of the sepZ::phoA fusion in 30-5-1(3) (Fig. 5A), the LEE2 operon fusion in the E. coli K-12 strain JLM166 was not activated by Per (Fig. 6). A second independent plasmid clone containing the identical LEE2::lacZ fusion to that found in JLM166 was sequenced to ensure that it contained no mutations and was transferred onto the chromosome of strain MC4100. This second, independently constructed LEE2::lacZ fusion was also not activated by Per (data not shown). In addition, neither the LEE3::lacZ nor the LEE4::lacZ single-copy operon fusions in the K-12-derived strains JLM172 and JLM165, respectively, were activated by Per (Fig. 6). For all four single-copy lacZ fusions in the K-12 strains, the parent plasmid pBR322 did not affect expression.

Figure 6.

. Increased expression of a LEE1::lacZ fusion in the presence of Per in E. coli K-12-derived strains. Single-copy chromosomal lacZ fusion strains derived from E. coli MC4100, strains transformed with per-containing plasmids pJLM170 and pJLM171 and those transformed with plasmid controls are indicated.

Identification of a new positive-acting regulator encoded within the LEE

In contrast to the activation of the LEE2 and LEE3 operons by Per in EPEC (Fig. 5A and B), these operons were not activated in K-12-derived strains (Fig. 6), suggesting that an additional factor in EPEC was necessary for Per-mediated activation of LEE2 and LEE3. Because the LEE1::lacZ operon in the K-12 strain JLM164 was activated by Per in the absence of EPEC-specific factors, we hypothesized that a positive-acting factor necessary for the activation of LEE2 and LEE3 in EPEC might be encoded within the LEE1 operon. Through mutational analyses and observed homologies with type III secretion components of other Gram-negative pathogens, functions have been assigned to EscRSTU (McDaniel, 1996; Elliott et al., 1998). The functions of orfs1 to 5 to this point are unknown, although it was known previously that Orf1 shares amino acid homology with the DNA structural protein and transcriptional regulator H-NS (24% identity and 44% similarity to H-NS of Salmonella ; McDaniel, 1996; Elliott et al., 1998).

To test whether Orf1 affected the expression of the operons of the LEE, pCVD456 containing orf1 (McDaniel, 1996) was transformed into the four K-12-derived lacZ operon fusion strains, and β-galactosidase activities were monitored. The presence of pCVD456 strongly activated expression of the LEE2::lacZ and LEE3::lacZ fusions in strains JLM166 and JLM172, respectively, increasing the expression of both operons from ≈100 to 1200 Miller units (Fig. 7). Expression of the LEE4::lacZ fusion in JLM172 was modestly increased, ≈twofold. To confirm that Orf1 was responsible for the increased transcriptional activity, a plasmid containing an orf1 minimal fragment, pSE1100, was transformed into the four fusion-containing strains. Similar to the effect of pCVD456, expression of the LEE2, LEE3 and LEE4 operon fusions was increased by the presence of Orf1. Orf1, however, did not appear to be autoregulated in E. coli K-12, as it did not affect the expression of the LEE1::lacZ fusion in JLM164 (Fig. 7).

Figure 7.

. Orf1 increased the expression of LEE2::lacZ, LEE3::lacZ and LEE4::lacZ fusions in E. coli K-12-derived strains. Single-copy lacZ fusion strains were transformed with pCVD456, a previously described orf1-containing subclone of the EPEC LEE in pMOB (McDaniel, 1996), pSE1100, which contains an orf1 minimal fragment in pBR322, or the pBR322 plasmid control. lacZ fusion strains containing the pMOB plasmid control exhibited β-galactosidase activities similar to those transformed with pBR322 (data not shown).

Discussion

Through a thorough analysis of the genetic organization of the EPEC LEE, including RT–PCR to define transcriptional units, identification of tsps by primer extension and location of predicted transcriptional terminators at the 3′ end of each message, we have identified four polycistronic operons. Genes encoding components of the type III secretion apparatus are encoded in three polycistronic operons designated LEE1, LEE2 and LEE3, and secreted molecules are transcribed from a fourth polycistronic operon designated LEE4. Internal termination sites and additional promoters, however, may also exist within these polycistronic messages.

In this report, we found that, in addition to the previously described activation of eae and bfp, Per activates the expression of four polycistronic operons of the EPEC LEE, thus confirming that Per is a global regulator. Although Per could activate the expression of these four operons in EPEC, when tested in E. coli K-12 strains, Per only activated the LEE1 operon. In K-12 strains, Orf1 increased the expression of the LEE2, LEE3 and LEE4 operons in the absence of additional EPEC-associated factors. Orf1 is therefore a novel regulator of EPEC virulence genes, part of a regulatory cascade activating the expression of genes encoding type III secretion components and modestly increasing the expression of the secreted molecules. We have thus designated orf1 as ler, for LEE-encoded regulator. To our knowledge, this is the first description of a cascade regulating virulence factor expression in EPEC. Data obtained from reporter gene fusions in K-12-derived strains with either Per or Ler in trans suggested that Per acted directly on the LEE1 promoter, while Ler acted directly on the LEE2, LEE3 and LEE4 promoters. Protein–DNA binding assays suggested that PerA interacts directly with bfp and eae promoter fragments (Tobe et al., 1996). However, our results suggest that Per activates eae indirectly or requires additional factors to facilitate activation (unpublished data; Gómez-Duarte and Kaper, 1995). These regulatory observations are summarized in Fig. 8.

Figure 8.

. Summary of positive regulatory cascade controlling the expression of genes of the EPEC LEE. Per-mediated activation of eae (Gómez-Duarte and Kaper, 1995) and bfp (Tobe et al., 1996) has been described previously.

Consistent with our observation that Ler is an important transcriptional regulator of EPEC virulence genes, a non-polar mutation in ler resulted in the inability of EPEC to secrete the Esp molecules or induce cytoskeletal rearrangement in host epithelial cells (S. J. Elliott, unpublished). Although Ler shares amino acid similarity with the DNA-binding protein and transcriptional regulator H-NS, its mechanism of regulation is obviously different, as H-NS acts negatively (Atlung and Ingmer, 1997), while Ler is an activator of transcription. Perhaps the similarities at the amino acid level indicate that, like H-NS, Ler binds DNA.

Regulators of type III secretion systems of Gram-negative pathogens are not typically transcribed as part of polycistronic messages that include secretion apparatus components; more commonly, they are located near these loci. However, the regulators invF of Salmonella typhimurium (Kaniga et al., 1994) and mxiE of Shigella flexneri (Allaoui et al., 1993) are transcribed as part of polycistronic operons. The observation that Ler is part of the LEE1 operon and activates expression of virulence factors within a pathogenicity island suggests that this protein is dedicated to the regulation of genes encoded within the LEE.

We identified a potential terminator downstream of the translational stop codon of orf1(ler). This might suggest that an additional promoter drives the expression of downstream genes. However, transcriptional activity derived from a single-copy orf1′–orf4′::lacZ fusion in a K-12 strain was relatively low compared with the other LEE operon fusions, ≈15 Miller units, and was not activated by Ler, a reasonable prediction if orf2345escRSTU were to be co-ordinately regulated with the LEE2 and LEE3 operons. This result was consistent with our original assignment of ler being the first gene of the polycistronic LEE1 operon.

Dual regulatory control of LEE genes, involving Per and Ler, is reminiscent of the VirF and VirB regulatory cascade that activates the expression of secreted molecules and type III components in Shigella flexneri (Adler et al., 1989; Dorman and Porter, 1998). Here, the two-component response regulator CpxR and H-NS act as negative regulators at virF and virB, respectively, responding to pH and temperature (Maurelli and Sansonetti, 1988; Hromockyj et al., 1992; Nakayama and Watanabe, 1995; 1998). We have determined that multiple regulators are involved in the expression of the major operons of the EPEC LEE, and transcription of these genes is most probably regulated by a number of environmental cues (Puente et al., 1996; Kenny et al., 1997b). The high level of expression of EPEC operon fusions in a K-12 background might indicate that a negative regulator is missing or titrated out in strains containing multicopy constructs. Perhaps the direct repeat sequence identified in the spacer of the LEE3 promoter interacts with a negative-acting factor, repressing the expression of this locus. In addition, the high level of basal expression of the operon encoding the secreted molecules, LEE4, might indicate that a feedback mechanism for controlling expression is not functional in K-12 strains. We are currently investigating which loci are involved in perceiving and mediating environmental signals regulating important virulence loci of the LEE.

Studies in our laboratory on transcriptional control of genes within both the EPEC and the EHEC LEE have greatly increased our understanding of these two related, yet different, pathogens. Like EPEC, the LEE2 and LEE3 operons of EHEC are regulated by EHEC Ler, which is nearly identical to Ler of EPEC in DNA and amino acid homology (V. Sperandio, unpublished). EPEC secretes effector molecules more efficiently than EHEC (our unpublished results), and this observation might be partially explained by the presence of Per. Per is a novel regulatory locus in that, unlike other AraC-like proteins described to this point, not only the AraC-like protein PerA, but also PerB and PerC are required for the full activation of target genes (Gómez-Duarte and Kaper, 1995; Tobe et al., 1996). How these proteins activate transcription is, however, unknown. We have not been able to identify a Per homologue in EHEC (Gómez-Duarte and Kaper, 1995), although the existence of an analogous, but non-homologous, regulator cannot be ruled out.

EHEC colonizes the large intestine, whereas EPEC primarily colonizes the small intestine. Perhaps colonization is facilitated by attachment factors unique to each pathogen or, alternatively, by differences in the regulation of virulence determinants. In a separate report, we have shown that the LEE1 and LEE2 operons of both EPEC E2348/68 and EHEC strain 86-24 are regulated by quorum sensing, most probably by a mechanism that does not involve a homoserine lactone as the signal molecule (V. Sperandio, unpublished). The higher bacterial concentrations in the large intestine compared with those in the small intestine may cause EPEC and EHEC to respond differently to quorum sensing in vivo, and the presence of Per might allow EPEC to compensate for the lower bacterial concentrations.

Experimental procedures

Bacterial strains and growth conditions

Strains and plasmids used in this study are listed in Table 1. All E. coli strains were grown aerobically at 37°C in Luria–Bertani (LB) medium unless otherwise stated. Selective antibiotics were added to the following concentrations: ampicillin, 100 μg ml−1; kanamycin, 50 μg ml−1; chloramphenicol, 20 μg ml−1.

Molecular cloning and sequencing procedures

Plasmid DNA purification, restriction, ligation, transformation and DNA agarose electrophoresis were performed using standard methods (Sambrook et al., 1989). A 3.9 kb BamHI fragment containing per was excised from pJPN14 (Nataro et al., 1987) and cloned into the BamHI site of pACYC184. This plasmid was designated pJLM161 and was used to investigate regulation by Per in the primer extension analysis (Fig. 3) and assays of β-galactosidase activity derived from the multicopy lacZ operon fusions shown in 5Fig. 5B.

To monitor transcriptional activities, defined regulatory fragments from the four identified operons were amplified by PCR using primers containing EcoRI and BamHI restriction sites (Table 2) and directionally cloned into the lacZ fusion vector pRS551 (Simons et al., 1987) after digestion with the same enzymes. Primers are listed in Table 2. The size of the amplified fragments were as follows: LEE1 726 bp, including 505 bp upstream of the tsp; LEE2 888 bp, including 682 bp upstream of the tsp; LEE3 696 bp, including 425 bp upstream of the tsp; LEE4 920 bp, including 486 bp upstream of the tsp. Cloned fragments were confirmed by plasmid isolation and restriction digest, PCR and DNA sequencing. These plasmids were designated pJLM164, pJLM166, pJLM172 and pJLM165 respectively. DNA sequence analysis was performed at the University of Maryland Biopolymer Facility on an ABI automated sequencer using DNA purified by Qiagen midi-columns.

Table 2. . Oligonucleotides used in this study.Thumbnail image of

Plasmid pJLM170 contains the identical 3.9 kb BamHI per fragment found in the pACYC184 derivative pJLM161, and pJLM171 contains a 1.8 kb minimal per clone containing only perA, perB and perC on a BamHI/Bgl II fragment cloned into the BamHI site of pBR322. The pBR322 derivatives, pJLM170 and pJLM171, were used in assays of β-galactosidase activities derived from the lacZ fusions in K-12 strains, as these single-copy fusion constructs were not compatible with pACYC184 and its derivatives.

The 797 bp orf1 minimal clone in pSE1100 was constructed by PCR amplification of EPEC E2348/69 LEE sequences using primers K1372 (5′-CCGGAATTCCTGTAACTCGAATTAAGTAGAGT-3′) and K1420 (5′-CGCGGATCCAGCTCAGTTATCGTTATCATT-3′) containing EcoRI and BamHI restrictions sites (underlined) respectively. This fragment was ligated directionally into pBR322.

Construction of single-copy chromosomal lacZ fusions

In order to study regulation of the four operons of the EPEC LEE, defined regulatory regions of the operons fused to the lacZ reporter gene in pRS551 were transferred to the chromosome of the K-12 strain MC4100 in single copy as described previously (Simons et al., 1987). Briefly, plasmids pJLM164, pJLM166, pJLM172 and pJLM165 containing regulatory regions of LEE1, LEE2, LEE3 and LEE4, respectively, were transformed into the recA+ strain MC4100 for obtaining λRS45 bacteriophage lysates. Overnight cultures were resuspended in 0.01 M MgSO4 and combined with 1 × 107 pfu; phage particles were allowed to adsorb, placed in 2.5 ml of top agar and overlaid on LB plates. After overnight incubation, phage particles were harvested.

For transfer to single copy onto the chromosome of MC4100, phage lysate was combined with an equal volume of the recipient strain in 0.01 M MgSO4, allowed to adsorb, combined with 2 ml of LB supplemented with 2 mg ml−1 maltose, shaken at 37°C for 2 h and plated on LB agar containing kanamycin and X-gal. Transductants were streak purified, screened for ampicillin sensitivity and assayed for β-galactosidase activity.

RNA extraction and RT–PCR technique

Bacterial strains were grown to exponential phase, OD600 0.5–0.7, and whole-cell RNA was isolated using the Trizol reagent (BRL) according to the manufacturer's instructions. RNA was treated with RNase-free DNase to eliminate contaminating DNA, and cDNA was synthesized using the random primers provided in the Preamplification Kit from BRL according to the manufacturer's instructions. Oligonucleotide primers used for the RT–PCR analysis are listed in Table 2, and their locations within the EPEC LEE are illustrated in Fig. 1. PCRs on the RT products, no RT control and DNA templates were performed using standard procedures with Taq DNA polymerase (BRL).

Primer extension analysis

Oligonucleotide primers used for primer extension are listed in Table 2. Each primer (10 pmol) was combined with 50 μCi [γ-32P]-ATP and radiolabelled with T4 polynucleotide kinase (BRL). End-labelled primer (1 μl) was then combined with 35 μg of whole-cell RNA, brought to 12 μl with DEPC-treated water, heated to 70°C and placed on ice. Four microlitres of 5 × First Strand Buffer (BRL), 2 ml of 0.1 M dithiothreitol (DTT) and 1 μl of 10 mM dNTPs were then added. Reaction mixtures in a total volume of 19 μl were incubated at 42°C for 2 min, then 200 U SuperScript II RTase was added and incubation continued for 50 min. After RTase inactivation, RNA was digested by adding RNase to 10 mg ml−1 and incubating at 37°C for 20 min.

DNA extension products were precipitated, pellets washed with 70% ethanol and resuspended in 8 μl of H2O and 4 μl of stop solution (US Biochemicals). DNA products were then separated adjacent to 35S-labelled sequencing reactions (US Biochemicals Sequenase 2.0 kit) generated using Qiagen-purified plasmid templates and the same oligonucleotide primers for the respective primer extension reactions (see Table 2) in an 8% polyacrylamide sequencing gel. Gels were dried on Whatman filter paper, and the results were visualized by autoradiography.

Enzyme assays

Bacterial strains were grown to exponential phase, OD600 of 0.3–0.5, and samples harvested for assay. The alkaline phosphatase assay has been described previously (Kaufman and Taylor, 1994); activities were measured in cell extracts, presented as units per OD of bacteria at 600 nm.

β-Galactosidase activity, expressed in Miller units, was measured as described previously (Miller, 1972). The data are presented as units per OD of bacteria at 600 nm.

Note added in proof

We have recently determined that tir, cesT and eae are transcribed as a single, polycistronic mRNA (Mellies and Kaper, unpublished). A transcriptional start site was identified previously for eae (Gómez-Duarte and Kaper, 1995), and it is certainly possible that transcription initiates from this internal site under certain conditions.

Acknowledgements

The authors are grateful for cloning and PCR performed by Maria Dubois, enzyme assays performed by Tim Moorehead, results on the regulation of eae collected by Sooan Shin, and many stimulating discussions with Steve Hutcheson. Lisa Sadzewicz and Nick Ambulos of The University of Maryland at Baltimore Biopolymer Laboratory are acknowledged for DNA sequencing and analysis. This work was supported by NIH grants AI21657 (J.B.K.) and AI32074 (M.S.D.).

Ancillary