Direct and indirect transcriptional activation of virulence genes by an AraC-like protein, PerA from enteropathogenic Escherichia coli


  • Megan E. Porter,

    Corresponding author
    1. Zoonotic and Animal Pathogens Research Laboratory, Medical Microbiology, University of Edinburgh, Teviot Place, Edinburgh, EH8 9AG, UK.
      E-mail; Tel. (+44) 131 650 4522; Fax (+44) 131 650 6531.
    Search for more papers by this author
  • Paul Mitchell,

    1. Zoonotic and Animal Pathogens Research Laboratory, Medical Microbiology, University of Edinburgh, Teviot Place, Edinburgh, EH8 9AG, UK.
    Search for more papers by this author
  • Andrew J. Roe,

    1. Zoonotic and Animal Pathogens Research Laboratory, Medical Microbiology, University of Edinburgh, Teviot Place, Edinburgh, EH8 9AG, UK.
    Search for more papers by this author
  • Andrew Free,

    1. Institute of Cell and Molecular Biology, University of Edinburgh, The King's Buildings, Mayfield Road, Edinburgh, EH9 3JR, UK.
    Search for more papers by this author
  • David G. E. Smith,

    1. Zoonotic and Animal Pathogens Research Laboratory, Medical Microbiology, University of Edinburgh, Teviot Place, Edinburgh, EH8 9AG, UK.
    Search for more papers by this author
    • Present address: Moredun Research Institute, Pentlands Science Park, Bush Loan, Peniciuk, Midlothian, EH26 0PZ, UK.

  • David L. Gally

    1. Zoonotic and Animal Pathogens Research Laboratory, Medical Microbiology, University of Edinburgh, Teviot Place, Edinburgh, EH8 9AG, UK.
    Search for more papers by this author

E-mail; Tel. (+44) 131 650 4522; Fax (+44) 131 650 6531.


The plasmid-encoded Per regulatory locus of enteropathogenic Escherichia coli (EPEC) is generally considered to consist of three genes, perA, perB and perC. PerA, a member of the AraC-like family of transcriptional regulators, is known to be an activator of its own promoter (autoactivation) as well as of the plasmid-located bfp operon encoding bundle-forming pili, but its role in activation of the chromosomal locus of enterocyte effacement (LEE) pathogenicity island, which confers the property of intimate adherence on EPEC, requires clarification. Here, we show that PerA is also required for activation of the master regulatory LEE operon, LEE1, but that this activation is indirect, being achieved via autoactivation of the per promoter which ensures sufficient production of the PerC protein to activate LEE1. In contrast, PerA-dependent activation of the per and bfp promoters is direct and does not require the other Per proteins, but is modulated by the nucleoid-associated protein H-NS. The closely related VirF regulator from Shigella flexneri cannot substitute for PerA to activate these promoters, despite being able to bind their upstream regions in vitro. PerA can bind the per and bfp promoter fragments to form multiple complexes, while VirF forms only a single complex. Site-directed mutagenesis of the PerA protein suggests that, like VirF, it may use both of its carboxy-terminal helix—turn–helix motifs for DNA interaction, and may also make direct contacts with RNA polymerase. In addition, we have isolated mutations in the poorly characterized amino-terminal domain of PerA which affect its ability to activate gene expression.


Enteropathogenic Escherichia coli (EPEC) is a major cause of infant diarrhoea in developing countries, as well as in day-care centres and paediatric wards in the developed world, and certain EPEC strains have been linked to diarrhoeal disease in adults (Kaper et al., 2004). EPEC generate attaching and effacing (A/E) lesions of the intestinal epithelium via a chromosomal pathogenicity island (PAI), the locus of enterocyte effacement (LEE; Elliott et al. 1998). Most EPEC strains also possess a ∼60 MDa plasmid called the EPEC adherence factor plasmid (pEAF; Kaper et al. 2004) which is essential for in vitro adherence to HEp-2 cells and for full virulence in volunteer studies (Bieber et al., 1998) and carries a type IV or ‘bundle-forming’ pilus encoded by the 14 genes of the bfp operon. Bfp expression mediates interbacterial adherence during colonization of epithelial cell surfaces, the ‘localized adherence’ (LA) phenotype (Hicks et al., 1998). However, pEAF has also been implicated in the control of the A/E phenotype via a modulation of LEE expression. This regulation is mediated by the pEAF-encoded per locus, which consists of three genes, perA, perB and perC, also called bfpT, bfpV and bfpW (Gómez-Duarte and Kaper, 1995; Tobe et al., 1996; Mellies et al., 1999). The perABC genes also activate the bfp operon (Tobe et al., 1996), thus mediating coordinate regulation of the A/E and LA phenotypes. The 31.8 kDa PerA protein shows C-terminal homology to the AraC family of transcriptional regulators, a large family of proteins which share a conserved DNA-binding domain containing two potential helix—turn–helix (HTH) motifs (Gallegos et al., 1997; Martin and Rosner, 2001). In contrast PerB and PerC, 14.8 kDa and 10.5 kDa, respectively, show no significant similarity to other known prokaryotic proteins (Gómez-Duarte and Kaper, 1995).

The PerABC proteins have been shown to activate the key regulatory operon of the LEE, LEE1, encoding the H-NS-like regulator Ler which then activates the LEE2, LEE3, LEE5, and to some extent LEE4, operons (Mellies et al., 1999; Sánchez-SanMartín et al., 2001). Hence, this constitutes a virulence regulatory cascade, although the roles of the individual Per proteins in this cascade are unclear. Bustamante et al. (2001) suggested that only PerC was involved in activation of the LEE via an indirect effect working to overcome an unknown negative factor encoded elsewhere on the pEAF plasmid. However, this work studied LEE2 and LEE3 promoter fusions in E. coli K-12, where Per is predicted to have no effect in the absence of Ler. In contrast, the effect of PerA as a transcriptional activator of promoters located on pEAF is relatively well-defined. Thus, PerA activates the promoter upstream of bfpA, the first gene of the bfp operon (Tobe et al., 1996), as well as mediating autoactivation of its own promoter (Martínez-Laguna et al., 1999), and both regulatory pathways require AT-rich regions upstream of the respective promoters which constitute potential PerA binding sites (Bustamante et al., 1998; Martínez-Laguna et al., 1999). Like many AraC-like regulatory proteins (Gallegos et al., 1997) PerA has proved difficult to purify because of its insolubility, but a maltose-binding protein (MBP)–PerA fusion protein has been shown to interact with the AT-rich regions upstream of the bfpA and perA promoters (Ibarra et al., 2003).

PerA is most closely related to the virulence-regulating subfamily of AraC-like regulators, which unlike most of the family, show sequence similarity throughout the N-terminal domain of the protein as well as the C-terminal DNA-binding domain (Gallegos et al., 1997; Dorman and Porter, 1998). The closest homologue of PerA is the VirF protein of Shigella flexneri, which is the primary activator of the virulence gene regulatory cascade in that organism (Dorman and Porter, 1998). PerA and VirF are also closely related to the Rns activator from enterotoxigenic E. coli (ETEC), and Rns and VirF can cross-substitute for each other to regulate their target promoters (Porter et al., 1998; Munson et al., 2001). However, PerA cannot substitute for Rns or vice versa (Munson et al., 2001); substitution of PerA by VirF has not been tested. Both VirF (Tobe et al., 1993; Porter and Dorman, 1994; 1997a) and Rns (Murphree et al., 1997) activate promoters that are repressed by the nucleoid-associated protein H-NS, and in the case of VirF a second nucleoid-associated protein, integration host factor (IHF), is also involved in modulating the target promoters (Porter and Dorman, 1997b). The involvement of these nucleoid-associated proteins in PerA-dependent activation has not been defined, although it has been suggested that IHF might be required for bfp transcription (Ibarra et al., 2003). DNA binding by AraC-like proteins has been studied by cocrystal structures of the MarA (Rhee et al., 1998) and Rob (Kwon et al., 2000) proteins, leading to the proposal of two different modes of DNA binding involving either one or both of the HTH motifs (Martin and Rosner, 2001). Mutagenesis of S. flexneri VirF (Porter and Dorman, 2002) suggests that it uses both HTH motifs for specific DNA contacts.

The current body of data on per-dependent regulation suggests three possible models for the role of PerA: (i) PerA requires the accessory proteins PerB and PerC to achieve maximal activation from all its target promoters, as originally suggested (Gómez-Duarte and Kaper, 1995; Tobe et al., 1996); (ii) PerA can work alone as a classical AraC-like protein or together with the accessory factors to activate gene expression depending on the target promoter; and (iii) PerA functions solely as a classical AraC-like protein, and PerB and PerC have either no role or an entirely independent role in activating EPEC virulence gene expression. Here, we have investigated these possibilities and show that the activation of bfp and per requires PerA alone, while both PerA and PerC are required to activate LEE1. However, the involvement of PerA in activating LEE1 is indirect, being required only to ensure sufficient transcription of PerC from the perABC promoter. The roles of the H-NS and IHF proteins in PerA-dependent per and bfp activation have been investigated, and PerA has been compared to the related VirF protein via DNA-binding studies and site-directed mutagenesis.


LEE1 requires PerA and PerC for full expression, whereas per and bfp require only PerA

In order to study directly the effect of the three Per proteins on expression from the per, bfp, and LEE1 promoters without interference from other regulatory mechanisms operating in EPEC or from other factors on the pEAF plasmid, we constructed single-copy translational fusions of each promoter to lacZ and inserted them into the chromosome of the E. coli K-12 strain MG1655 at the location of the native lac genes (see Experimental procedures). We then cloned the intact perABC locus from EPEC strain E2348/69 into the low-copy cloning vector pTH19kr on a 2.8 kb DNA fragment, and transformed this construct into the strains carrying the per–lacZ, bfp–lacZ and LEE1–lacZ fusions. We had initially made repeated attempts to clone the per locus into the medium-copy vector pACYC184, but the clones that were retrieved all had nonsense or potentially inactivating missense mutations in either the per promoter or the perA or perC genes, suggesting that over-expression of these proteins, when present in multicopy, is poorly tolerated, as has been observed by others (Okeke et al., 2001). Such mutability could account for some of the inconsistent results previously obtained with perABC clones. Using the low-copy clone pTHperABC as a starting point, we then constructed in-frame deletions in each of the per genes (see Experimental procedures) to generate clones with non-polar mutations in perA, perB, or perC, which retained the ability to express the respective non-mutated genes. These mutations were also combined to create clones expressing a single Per protein, and all clones were sequenced in their entirety to confirm that no other point mutations were present. The deletion clones were also transformed into the single-copy lacZ fusion strains.

The expression from each of the lacZ chromosomal fusions was assayed in strains transformed with either the perABC clone pTHperABC or the cloning vector pTH19kr. β-Galactosidase was measured in cultures grown to mid-logarithmic phase (OD600∼0.6) in Dulbecco's Modification of Eagle's medium (DMEM), conditions which were found to maximize secretion of the EPEC secreted proteins in strain E2348/69 and expression of the lacZ fusions in strain MG1655 (our unpublished data). Background expression from the per–lacZ and bfp–lacZ fusions in the absence of PerABC was very low (∼50 U and ∼20 U respectively), but these fusions were strongly activated by the perABC clone (∼2400 U for per and ∼20 000 U for bfp; Fig. 1A and B). The LEE1–lacZ fusion gave a higher basal level of expression (∼500 U), but was also activated by the Per proteins (Fig. 1C), although the fold activation was lower (30-fold) than at the bfp promoter (1000-fold). The plasmids with deletions in either perB, perC or in both genes together were found to activate per-lacZ and bfp-lacZ as well as the intact perABC plasmid, while the perA deletion construct mediated no activation of either per or bfp (Fig. 1A and B). Thus at these two promoters, PerA acts as a classical AraC-like activator protein, and does not require PerB or PerC as an accessory factor. However, at LEE1 a different pattern of activation was seen. While the perB mutation did not affect the LEE1–lacZ expression level, deletion of perA or perC, either singly or in combination, eliminated the transcriptional activation mediated by the intact locus (Fig. 1C). Therefore, both the PerA and PerC proteins are necessary together to activate LEE1. Hence, these studies show definitively that per plays an activating role at LEE1, which requires PerC as well as the AraC-like protein PerA. LEE1 activation does not require any other EPEC-specific factors, and is distinct from the PerA-dependent activation mechanism at per and bfp.

Figure 1.

Effects of cloning vector pTH19kr, full-length perABC clone pTHperABC (perABC+), and derivatives of pTHperABC with in-frame deletions in perA, perB, perC, perA + perB, perA + perC or perBC on chromosomally located lacZ fusion expression in E. coli K-12. A. per–lacZ. B. bfp–lacZ. C. LEE1–lacZ.

PerC can activate LEE1 transcription independently of PerA when expressed from a heterologous promoter

There are two possible explanations for the dual PerA- and PerC-dependency of LEE1 transcription. The first is the originally proposed model for Per activation, that the PerA and PerC proteins cooperate at the LEE1 promoter to activate expression, presumably via the DNA-binding activity of the AraC-like partner, PerA, and a potential protein–protein interaction with PerC, which allows the two proteins to activate expression cooperatively. The second possibility is that the effect of PerA at LEE1 is entirely indirect. As shown in Fig. 1A, PerA is a strong activator of the promoter upstream of the perA gene. If this is the sole promoter transcribing the entire perABC operon, perC expression will be dependent on PerA, and the observed PerA-dependency of LEE1 may therefore simply reflect a lack of PerC protein in the absence of PerA. To test the PerA-dependency of perC transcription, we isolated RNA from strains carrying the intact perABC plasmid pTHperABC, each of the single-deletion plasmids, and a plasmid carrying all three deletions combined (pTHΔperABC), and performed Northern blots with probes against the internal regions of the perA, perB and perC genes (see Experimental procedures; Fig. 2A). As expected, perA transcript was detectable in the RNA from the intact perABC clone and from the ΔperB and ΔperC clones. The level of perA transcript was in fact elevated in the presence of the downstream deletions. In contrast, perB- and perC-specific transcripts were detectable at significant levels in the intact clone, but were absent when either perA or the gene being probed for was deleted. Again, the perB and perC transcript levels were elevated in the ΔperC and ΔperB clones, respectively, compared to the intact perABC clone, possibly because of increased efficiency of transcriptional readthrough or altered stability of the deleted transcript. These results suggest that there are no alternative non-PerA-dependent promoters within the perABC operon which can transcribe perB and perC in the absence of PerA, and we therefore predict that the expression of PerC from a heterologous promoter should eliminate the PerA-dependency of LEE1 expression. Indeed, expression of PerC from the arabinose-inducible araBAD promoter in the plasmid pBADPerC, in the presence or absence of functional PerA (and PerB), mediated the arabinose-dependent induction of LEE1-lacZ expression (Fig. 2B). This expression was independent of the presence of the perA and perB genes on pTHperABΔC, and the cloning vector pBAD33 mediated no such activation. In contrast, per–lacZ expression was dependent on pTHperABΔC, but unaffected by arabinose-dependent induction of PerC (Fig. 2C). Therefore, LEE1 expression does not require PerA per se as expression of PerC from a heterologous promoter in the absence of PerA still activates LEE1. PerA should hence be considered an indirect activator of LEE1 via its autoactivation effect on the per promoter, while PerC is most likely a direct activator.

Figure 2.

A. Northern blot analysis of perA-, perB- and perC-specific transcripts from the cloned perABC locus (pTHperABC) in E. coli K-12, and from clones with in-frame deletions in perA, perB, perC or in all three genes (Δper).
B and C. Effect of expression of PerC from a heterologous arabinose-inducible promoter (araBAD) in pBADPerC, or the vector control pBAD33, on chromosomally located lacZ fusion expression in E. coli K-12 in the presence of cloned perAB (pTHperABΔC) or vector control (pTH19kr). Concentrations of arabinose added to the medium are indicated by solid bars (0%), heavy shading (0.002%), medium shading (0.02%), light shading (0.2%) or open bars (1.0%). B. LEE1–lacZ. C. per–lacZ.

PerA-dependent activation of per and bfp is modulated by the nucleoid-associated protein H-NS

The above results indicate that PerA is a classical AraC-like virulence regulator, which, like most other family members, activates gene expression without the requirement for specific accessory factors. Determining which other characteristics PerA shares with this family of proteins should enable a better understanding of how PerA activates transcription. Among the virulence regulator subfamily of AraC-like proteins, S. flexneri VirF and ETEC Rns are closely related, and VirF is the closest relative of PerA. As both VirF and Rns overcome repression by the nucleoid-associated protein H-NS, it might be expected that PerA could function similarly. VirF-activated promoters are also modulated by the IHF protein, and we therefore assessed the roles of these two nucleoid-associated proteins in activation by PerA. As the positive autoregulatory loop at the perA promoter can lead to complications in interpreting data on PerA activation, we used an inducible PerA derivative transcribed from the araBAD promoter to study the effects of hns and ihf mutations on PerA-dependent activation of bfp–lacZ and per–lacZ. In the absence of PerA (i.e. in the presence of the vector control pBAD33), an hns null mutation caused significant (fivefold) derepression of the bfp promoter (Fig. 3A), while a mutation of the α-subunit of IHF had little effect on expression. When the pBADPerA plasmid was present and induced with 1% arabinose, however, bfp expression was activated to the same level regardless of the presence or absence of H-NS or IHF in the cell (Fig. 3A). This indicates that PerA not only efficiently overcomes H-NS-dependent repression at the bfp promoter, but also has a direct activating effect independent of H-NS. While a twofold derepression of basal expression in the absence of H-NS was also seen at the per promoter (Fig. 3B), in this case, the level of activated expression achieved by PerA was enhanced in the absence of IHF but reduced in the absence of H-NS. Such a positive effect of H-NS on gene expression is very unusual (Dorman, 2004), while IHF also more commonly acts as a transcriptional activator than as a repressor. Therefore, it is possible that these effects are indirect (see Discussion).

Figure 3.

Activation of (A) bfp–lacZ and (B) per–lacZ expression by the control vector pBAD33 (solid bars) or the PerA-expressing plasmid pBADPerA (shaded bars) in E. coli K-12 wild-type, hns-206::Apr, ihfA82::Tn10 and hns-206::AprihfA82::Tn10 strains grown in the presence of 1% arabinose. The activities of the bfp–lacZ and per–lacZ fusions are plotted on a logarithmic scale to enable the regulatory effects in both the absence and presence of PerA expression to be seen clearly.

VirF fails to activate the per and bfp promoters, and forms a single protein–DNA complex instead of the twin complexes formed by PerA

VirF and Rns are closely related and have been shown to be able to activate each other's target promoters (Porter et al., 1998; Munson et al., 2001). While Rns is unable to substitute for PerA, the possibility that the more closely related VirF protein might activate PerA-dependent promoters has not been studied. To investigate this possibility, we studied DNA binding and activation by PerA and VirF at the per and bfp promoters. Expression of VirF from the multicopy vector pACYC184 in the per-lacZ and bfp-lacZ fusion strains mediated no significant activation at either promoter (Table 1). This could be because, unlike PerA, VirF autorepresses rather than autoactivates its own promoter (Porter and Dorman, 1997a), so that even a multicopy clone will express VirF at lower levels than a low-copy clone of perABC will express PerA. Such a low level of VirF might be insufficient to activate the per and bfp promoters to a significant extent. Therefore, we expressed VirF in the same fusion strains from the araBAD promoter using plasmid pBADVirF. However, after induction with 1% arabinose, pBADVirF mediated no activation of per and bfp, while pBADPerA activated transcription efficiently (Table 1). Therefore, despite its sequence similarity to PerA, VirF is unable to substitute for PerA at these promoters. To determine whether this was due to differential DNA-binding abilities or not, we sought to purify the PerA and VirF proteins. AraC-like regulators have proved notoriously difficult to purify in the past because of their insolubility which has been attributed to the conserved DNA-binding domain (Gallegos et al., 1997), and this problem has often been overcome by fusing the N-terminal domain to a large protein molecule such as maltose-binding protein (MBP). Indeed, an MBP–PerA fusion protein has been purified and shown to bind with moderate affinity (a kD in the µM range) to the per and bfp promoters (Ibarra et al., 2003). However, given that this seems very weak binding for a gene-specific regulatory protein, we were concerned that such a large (40 kDa) protein fused to PerA might be reducing the measurable DNA-binding affinity and/or interfering with properties such as multimer formation. Therefore, we employed an alternative strategy using a double, N-terminal epitope tag consisting of a hexahistidine tag followed by an S-peptide to attempt purification of PerA and VirF. After expression in E. coli strain BL21 (λDE3) and purification over nickel-agarose (see Experimental procedures), the proteins were dialyzed against a buffer previously used in purification of the Salmonella typhimurium AraC-like regulator HilD (Schechter and Lee, 2001). This procedure yielded a small but useable amount of soluble PerA (0.08 mg l−1) and significantly more VirF (0.5 mg l−1; Fig. 4A). Some breakdown of each protein was observed despite the presence of protease inhibitors throughout the purification, as has previously been observed for MBP–PerA (Ibarra et al., 2003). We confirmed that the tagged PerA protein was able to activate transcription as efficiently as the native protein by expressing it from the arabinose-inducible araBAD promoter in the per-lacZ fusion strain (Fig. 4B).

Table 1.  Comparison of PerA and VirF effects on per-lacZ and bfp-lacZ expression.
FusionPlasmidaβ-galactosidase (U)b
  • a

    . Plasmid construct encoding PerA or VirF (with arabinose concentration used for induction shown for arabinose-inducible clones).

  • b

    . Assays were carried out in duplicate; mean values, in Miller units, are shown. Standard deviations are in parentheses.

per–lacZpTH19kr63.8 (0.33)
pTHperABC2420 (55.0)
pACYC18440.9 (0.63)
pACVirF97.7 (2.65)
pBADPerA (0% arabinose)44.7 (0.14)
pBADPerA (1% arabinose)1229 (4.24)
pBADVirF (0% arabinose)37.0 (0.21)
pBADVirF (1% arabinose)35.7 (0.42)
bfp–lacZpTH19kr17.9 (0.34)
pTHperABC50 499 (1497)
pACYC18413.8 (0)
pACVirF16.0 (0.15)
pBADPerA (0% arabinose)16.3 (0.71)
pBADPerA (1% arabinose)13 695 (58.0)
pBADVirF (0% arabinose)16.0 (0.21)
pBADVirF (1% arabinose)15.7 (0.35)
Figure 4.

A. Purification of His6-S-tagged PerA and VirF via Ni2+-agarose column chromatography. Coomassie-stained gels of samples of total soluble proteins, the unbound column fraction in the presence of 5 mM imidazole, wash fractions in the presence of 35 mM imidazole and the fraction eluted with 0.5 M imidazole are shown. The intact PerA (36.3 kDa) and VirF (35.2 kDa) proteins are indicated by arrows, as are breakdown products of each protein which were routinely observed. Positions of molecular weight markers (kDa) are shown.
B. Activation of per–lacZ expression in E. coli K-12 by the His6-S-tagged PerA protein expressed from the pBADH6SPerA plasmid in the presence of 1.0% arabinose, compared to untagged PerA expressed from pBADPerA and the vector control.
C. Binding of His6-S–PerA to the upstream regions of the per (−141 to −4 with respect to the ATG) and bfp (−176 to −28 with respect to the ATG) promoters in an electrophoretic mobility shift assay. Concentrations of purified His6-S–PerA in nM are shown, and the unbound DNA fragments and PerA–DNA complexes are indicated by arrows.
D. Binding of His6-S–VirF to the same per and bfp promoter fragments as in (C). Concentrations of purified His6-S–VirF in nM are shown, and the unbound DNA fragments and VirF–DNA complexes are indicated by arrows.

To test the DNA-binding ability of our purified epitope-tagged PerA and VirF, we amplified and labelled DNA fragments from the per and bfp promoters containing the sites to which the MBP–PerA protein fusion had been shown to bind (Experimental procedures), and used them as targets in mobility shift assays. When PerA was titrated onto either of these fragments, protein–DNA complexes were seen to form readily in the presence of non-specific competitor DNA (Fig. 4C). Complexes were seen with concentrations of PerA in the binding reaction as low as 13 nM, which is much lower than the 300 nM concentration necessary to observe a complex between the MBP–PerA fusion and its binding sites (Ibarra et al., 2003). Moreover, at higher concentrations of PerA, secondary complexes of a higher molecular weight were observed, suggestive of a second binding site for PerA on the DNA. Therefore, it appears that the epitope-tagged PerA binds much more tightly to the same DNA regions than does the MBP–PerA fusion, and also has an additional ability to form a larger nucleoprotein complex, either via multimerization or through an independent binding event. It is likely that the DNA-binding affinity and the ability to form this second complex are negatively affected by the bulky MBP protein fused to PerA in the previously studied construct. Although it could not activate per or bfp transcription, VirF could bind to the per and bfp DNA fragments at similarly low protein concentrations to PerA (Fig. 4D). However, only a single protein–DNA complex was detectable within the concentration range used (up to 100 nM), although VirF can readily form higher-order complexes on DNA fragments from both the S. flexneri virB and virF promoters, which are natural in vivo targets of the protein (our unpublished data). Therefore, the lack of activation of per or bfp by VirF correlates with its inability to form higher-order protein–DNA complexes on these promoters, which PerA can do readily, suggesting that these secondary complexes are important for transcriptional activation.

PerA proteins mutated in both HTH motifs in a potential RNA polymerase-contacting residue and in regions of the N-terminal domain are deficient in activation

In an attempt to further our understanding of the molecular mechanism of PerA-dependent activation, we carried out site-directed mutagenesis at selected residues throughout the PerA protein (Table 2; Fig. 5A). To confirm the importance of the twin HTH motifs for DNA binding, we mutated a key residue from each which, from the MarA-DNA cocrystal (Rhee et al., 1998), is predicted to form specific contacts to DNA but is not involved in the integrity of the HTH structure. These residues, K200 (HTH1) and Y246 (HTH2) are equivalent to K193 and Y239 in VirF, which are essential for transcriptional activation (Porter and Dorman, 2002), and mutations to isoleucine or phenylalanine, respectively, were chosen to remove charge–charge or hydrogen-bonding interaction capacity while minimizing structural disruption. In contrast, residue E234 in HTH2 is predicted to be oriented away from the DNA based on the MarA structure, but is the equivalent of D241 in the rhamnose-dependent AraC-like activator RhaS, which has been suggested to mediate contacts with the RNA polymerase σ70 subunit (Bhende and Egan, 2000). We abolished this potential contact by making an E234A mutation; during construction of this mutation, a second mutation at an adjacent residue, S236C, was also isolated. We also made alanine substitutions in two acidic side chains in the α-1 helix which lies above the HTH core and may mediate relative positioning of the N- and C-terminal domains: D168, which points away from the C-terminal domain in MarA, and E175, which is integral to the HTH core. Given the lack of information about the N-terminal domain, the N-terminal mutations were targeted at residues conserved in VirF and Rns (Y26, D100 and K108) or those likely to be involved in putative structural elements of the N-terminal domain as determined by a secondary structure prediction (A20, G37, W68; Fig. 5A). Randomly acquired mutations I85M and L94P in the N-terminus, and A221T and R226G within helix α-4 of the C-terminal domain were initially isolated during attempts to clone perABC into a medium-copy vector, and were subsequently subcloned into the low-copy perABC clone; a less conservative mutation of residue 85 to arginine was also constructed subsequently (Table 2).

Table 2.  Characteristics and phenotypes of PerA point mutations studied here.
  • a

    . Mutations are indicated by single letter amino acid codes.

  • b

    . Source of the mutation; randomly acquired or by site-directed mutagenesis (SDM).

  • c

    . Phenotype of the mutant protein as assayed using the perA–lacZ fusion (Fig. 5).

  • d

    . Location of the mutated residue relative to predicted secondary structures in the N-terminal domain or relative to the structural features of the C-terminal domain as defined for MarA and Rob (Fig. 5).

  • e

    . Conservation of the mutated residue among related AraC family proteins. Sim. = conservative substitution.

A20PSDMFull activityN-term., helix
Y26FSDMFull activityN-term.VirF/Rns
G37WSDMNo activityN-term., β-loop-β
W68GSDMSlight activityN-term.VirF/Rns sim.
I85MRandomFull activityN-term.
I85RSDMFull activityN-term.
L94PRandomNo activityN-term., helixVirF sim.
D100ASDMFull activityN-term.VirF/Rns sim.
K108ESDMFull activityN-term., helixVirF/Rns
D168ASDMFull activityHelix 1AraC-like sim.
E175ASDMSlight activityHelix 1AraC-like
K200ISDMNo activityHelix 3 (HTH1) base interactingAraC-like
A221TRandomNo activityHelix 4AraC-like
R226GRandomV. slight activityHelix 4
E234ASDMPartial activityHelix 5 (HTH2) RNA pol. interacting?VirF/Rns sim.
E234AS236CSDM/RandomNo activityHelix 5 (HTH2)AraC-like sim.
Y246FSDMNo activityHelix 6 (HTH2) base-interactingAraC-like
Figure 5.

A. Secondary structure of the PerA protein and locations of residues mutated in this study. The boundary between the N- and C-terminal domains is indicated by an oblique line. Predicted secondary structure motifs in the N-terminal region of PerA are shown by open boxes (α = α-helix; β = β-strand), with predicted loop regions indicated by ‘L’. C-terminal helices α-1–α-7 identified in the crystal structure of MarA bound to DNA (Rhee et al., 1998) are shown in black (helices involved in the HTH motifs) or gray (other helices). The turns of the HTH motifs are indicated by ‘T’.
B. Activation of the per–lacZ fusion in E. coli K-12 by wild-type pTHperABC (indicated by ‘wt’) or derivatives of the same plasmid containing point mutations in the perA gene as indicated. The vector control pTH19kr is also shown.
C. Abilities of the PerA mutant derivaties showing impaired activation ability in (B) to activate per–lacZ and bfp–lacZ expression when induced from the araBAD promoter in the presence of 1.0% arabinose. The wild-type PerA construct pBADPerA and the vector control pBAD33 are shown for comparison. Levels of activated bfp–lacZ expression (right-hand scale, open bars) are approximately 10 times higher than those of activated per–lacZ expression (left-hand scale, solid bars).

We initially assayed all the mutant proteins in the context of a low-copy perABC clone activating the per–lacZ fusion (Fig. 5B). This screen indicated that several of the mutations, particularly within the N-terminal domain (A20P, Y26F, I85R, I85M, D100A and K108E) were essentially unaffected in activation function. However, other N-terminal mutants (G37W, W68G and L94P) had no activation function, indicating that they affected residues and/or secondary structural elements crucial for PerA folding or activity. Targeting the key DNA-interacting side chains from either of the two C-terminal HTH motifs (K200I and Y246F) inactivated PerA, suggesting that like VirF, it requires both its HTH motifs to interact with its DNA targets correctly. Three out of four mutations in the non-DNA-binding helices of the C-terminal domain (E175A, A221T and R226G, but not D168A) also led to inactive proteins, probably because they affect the structural integrity of the domain and hence DNA binding. Interestingly, the E234A mutation, equivalent to D241A previously studied in RhaS, reduced but did not eliminate PerA activity; introduction of the adjacent S236C mutation in conjunction with E234A reduced activity to background levels (Fig. 5B).

As our previous results suggested that the powerful perA autoactivation mechanism might enhance any negative effects of the various mutations when the proteins were expressed from their native promoter, we re-cloned the mutant proteins with compromised function (G37W, W68G, L94P, E175A, K200I, A221T, R226G, E234A and Y246F) downstream of the araBAD promoter and re-assessed their ability to activate the per–lacZ fusion (Fig. 5C). This revealed that the W68G, E175A and, to a lesser extent, R226G proteins still retained some activation function, whereas the E234A mutant retained > 50% activity in this assay. As expected, the DNA-binding residue mutants K200I and Y246F were still completely inactive, as were the G37W and L94P N-terminal mutants and A221T from helix α-4, which acts as a bridge between the two HTH motifs. Using the same plasmid constructs, we also studied the abilities of these proteins to activate the bfp–lacZ fusion (Fig. 5C). Again, the E234A mutant showed significant residual activity, although the other mutants were generally less active than at the per promoter, and some (e.g. W68G) showed significantly less activity. This suggests that activation by PerA differs slightly between the two promoters, possibly connected to the different locations of the PerA-binding sites relative to the core promoter, spanning positions −75 to −47 at the per promoter but situated further upstream (−83 to −55) at the bfp promoter (Ibarra et al., 2003). As we had successfully shown high-affinity DNA binding by wild-type PerA at the per and bfp promoters, we attempted to purify the various mutated derivatives of the protein to compare their DNA-binding abilities. However, all of the proteins showed markedly reduced solubility compared to the wild-type, and although we could detect them all by Western blotting, we were unable to obtain stable preparations of them at high enough concentrations to perform mobility shift assays. The Western analysis did reveal that the W68G and L94P mutants exhibited slightly altered mobility in SDS-PAGE, consistent with a significant structural change in the protein as a result of these mutations (data not shown).


The data presented here show clearly that the perABC locus of EPEC mediates its effects on expression of the pEAF-encoded bundle-forming pilus operon and the chromosomally encoded LEE pathogenicity island via distinct mechanisms. Activation of bfp, like autoactivation of the per promoter, requires PerA alone, and deletion mutations in the perB or perC loci have no effect at either of these promoters. In contrast, LEE activation, which functions at the master regulatory promoter of LEE1, requires both PerA and PerC at the genetic level, but the effect of PerA is indirect and is required only to ensure sufficient production of PerC via the autoactivation mechanism. Thus, PerA is a classical AraC-like protein which activates transcription without accessory factors but, by virtue of the genetic organization of the per locus, mediates LEE1 activation via PerC. The observation that PerC alone expressed from an inducible promoter can activate an LEE1–lacZ fusion in E. coli K-12 shows that the previous suggestion that PerC-dependent LEE activation might be indirect via the relief of a pEAF-associated inhibitory effect (Bustamante et al., 2001) is incorrect. Rather, our data allow us to refine the two-step LEE regulatory cascade defined previously (Mellies et al., 1999) into a three-step cascade: PerA activates its own promoter thus producing PerC; PerC activates LEE1 thus producing Ler; and Ler then activates the other principal promoters of the LEE (Fig. 6). Other factors which regulate the perABC promoter, such as GadX (Shin et al., 2001), will also feed into this cascade, as will factors such as the newly identified LEE1 regulators GrlA and GrlR (Deng et al., 2004).

Figure 6.

Roles of the PerA and PerC proteins in the activation of EPEC virulence genes. Promoters are shown by angled arrows, and the perABC transcript identified here by a dashed line. Direct activation of the pEAF-located per (and bfp) promoters by PerA, corresponding to the first stage of the Per-regulated cascade, is indicated in the upper part of the figure. The second and third stages of the cascade, activation of the chromosomally located LEE1 promoter by PerC and activation of the LEE2-5 operons by Ler (Friedberg et al., 1999; Mellies et al., 1999) are shown in the lower part. H-NS has multiple negative inputs at LEE2-5 (Bustamante et al., 2001; Haack et al., 2003), LEE1 (Umanski et al., 2002), per and bfp (this work).

Our data suggest no role for the PerB protein in either per/bfp or LEE regulation. Although the original studies on the per locus (Gómez-Duarte and Kaper, 1995; Tobe et al., 1996) suggested a positive role for PerB in the presence of the other Per proteins in expression of intimin and Bfp, these studies used multicopy plasmids or highly induced transcription levels, respectively, which may have affected the results. Our experiments with low-copy per clones show that, although PerA-dependent transcription of the perB region is readily detectable, no effect of an in-frame deletion of the perB gene is seen in the systems we have studied. Despite the recent plethora of bacterial genomes sequenced, there are no homologues of PerB found outside EPEC strains, suggesting that it is not a protein of conserved function. It may be that the perB reading frame is maintained in EPEC purely because interruptions in it (unlike the in-frame deletion used here) might affect the PerA-dependent expression of the downstream perC gene, and hence LEE expression. However, the perB (and perC) gene sequences are less variable among different EPEC isolates than is perA (Okeke et al., 2001). An alternative hypothesis is that PerB performs an as-yet-undefined function in EPEC, and that, like PerC, it relies on being transcribed from the autoactivated perABC promoter for this function. We have also found that expression of a single-copy trcA–lacZ fusion in E. coli K-12 is unaffected by any of the Per proteins either singly or in combination (our unpublished data), suggesting that this chromosomal gene, originally isolated as a PerA-binding promoter fragment (Tobe et al., 1999), may not in fact be part of the per regulon.

Our studies of PerA-dependent activation demonstrate that, like VirF and Rns at their target promoters (Tobe et al., 1993; Porter and Dorman, 1994; 1997a; Murphree et al., 1997), PerA overcomes a negative effect of the nucleoid-associated protein H-NS at the per and bfp promoters (Fig. 6). H-NS is a global regulatory protein which represses a wide variety of promoters, particularly those involved in environmental responsiveness and/or pathogenesis, via its ability to interact with intrinsically curved DNA (Dorman, 2004). Curvature analysis of the bfp and per promoters (data not shown) shows that both are predicted to have a region of curvature centred on the mapped PerA binding sites; H-NS may compete with PerA for these intrinsically curved binding sites, leading to the observed pattern of regulation. Full PerA-dependent autoactivation of the per promoter is however, reduced in an hns mutant; the most likely explanation for this is that H-NS also acts indirectly by repressing a repressor of per, such as GadX (Shin et al., 2001). A second nucleoid-associated protein, IHF, which is a positive regulator of both the VirF-activated promoters in S. flexneri (Porter and Dorman, 1997b) and the LEE1 promoter in EPEC (Friedberg et al., 1999), is found here to have only slight effects at the bfp and per promoters. This contrasts with a previous suggestion (Ibarra et al., 2003) that IHF is required for bfp transcription, but we have verified that the same ihf mutation negatively affects expression of our LEE1–lacZ fusion as expected (our unpublished data). Our results suggest that IHF does not have a significant input at the PerA level of the EPEC virulence regulatory cascade; its weak negative effect on activated per transcription, like the positive effect of H-NS, may be indirect via a factor such as GadX.

Work reported here on the purification and analysis of a peptide-tagged PerA derivative indicates that PerA has a much higher affinity for its target promoters than previously suspected. His6-S-tagged PerA can bind the per and bfp promoters at concentrations in the 10–50 nM range, compared to 300–1500 nM previously reported for the MBP–PerA fusion protein (Ibarra et al., 2003). Unless a large percentage of this MBP–PerA fusion preparation has lost its DNA-binding activity, this suggests that the bulky MBP moiety, which is larger than PerA itself, destabilizes the PerA–DNA complex significantly, at least under conditions of gel electrophoresis. MBP–PerA also fails to form a second protein–DNA complex on the bfp and per promoter fragments despite complete binding of the DNA fragment (Ibarra et al., 2003), while His6-S–PerA readily forms a second complex at subsaturating concentrations. This could indicate independent binding to a weaker, secondary site or oligomerization not seen with MBP–PerA. Oligomerization of PerA in solution has not been detected in vivo using LexA–PerA fusion proteins (Ibarra et al., 2003) or in vitro using His6-S-tagged PerA and zero-length chemical cross-linking (our unpublished data). Although technical problems may account for these negative results, it is possible that PerA may only oligomerize when bound to its cognate DNA target, as shown for the E. coli AraC-like protein MelR (Bourgerie et al., 1997) and suggested by the genetics of S. flexneri VirF mutant derivatives (Porter and Dorman, 2002). Consistent with this interpretation, coexpression of the predicted DNA-binding mutants of PerA with wild-type protein has no dominant negative effect (our unpublished data). Indeed, only the E234A mutant, predicted to bind DNA normally but to be defective in activation, has a slight dominant negative phenotype. Interestingly, we have shown that VirF forms only single complexes on the PerA–DNA targets, although it readily forms multiple complexes on its specific target promoters, virB and virF. This correlates with the inability of VirF to activate the per and bfp promoters when expressed either in multicopy or from a strong inducible promoter, and suggests that the ability for multiple protein molecules to bind to a specific DNA target may be crucial for these AraC-like virulence regulators to activate transcription.

By site-directed mutagenesis and analysis of randomly acquired mutants of PerA selected during multicopy cloning attempts, we have studied PerA protein function at the molecular level. The clearest conclusions concern the C-terminal HTH motifs comprising the DNA-binding domain. Studies on the structures of AraC-like proteins in complex with their DNA-binding sites suggest that they can use either the first HTH motif alone (Rob; Kwon et al. 2000) or both motifs (MarA; Rhee et al. 1998) to make specific contacts with DNA, while it has been suggested that AraC itself can use either one or both motifs depending on the binding site (Hendrickson and Schleif, 1985; Niland et al., 1996). In the MarA crystal structure, the side chain of amino acid T93 in the recognition helix of HTH2 makes water-mediated hydrogen bonds with two DNA bases in the binding site (Rhee et al., 1998); although a MarA T93A mutant is unaffected in transcriptional activation (Gillette et al., 2000), elimination of the hydrogen-bonding potential of the equivalent residue in HTH2 of VirF via a Y239F mutation suggests that VirF requires both HTH motifs for DNA binding (Porter and Dorman, 2002). Here, a similar mutation at the equivalent residue in PerA (Y246F) completely abolishes PerA-dependent activation of the per and bfp promoters, suggesting that PerA also uses the twin-HTH mode of DNA binding. Likewise, mutation of a key base-interacting side chain in HTH1 (K200, equivalent to R46 in MarA and K193 in VirF) to isoleucine also eliminates PerA activity. Amino acid E234 of PerA, corresponding to Y81 in MarA is, by contrast, predicted to be surface-exposed on the opposite side of HTH2 directed away from the DNA target, and modelling studies and mutational analysis of its equivalent in the RhaS protein, D241, suggest that it makes a direct contact with the R559 side chain of the RNA polymerase σ70 subunit, which mediates transcriptional activation (Bhende and Egan, 2000). We have shown that an E234A mutant of PerA is compromised but still partially functional in transcriptional regulation, suggesting that E234 may contribute to similar RNA polymerase contacts in PerA. Mutations of residues within the α-4 helix, which links the two HTH motifs, A221T and R226G, were selected during attempts to clone the per locus in multicopy and lead to inactive proteins, presumably because they disrupt the overall structure of the twin-HTH domain. By contrast, only one of two mutations within the α-1 helix which precedes HTH1 (E175A, but not D168A) affects PerA function, probably because D168 is predicted to be oriented away from the rest of the domain whereas E175 is integral to the HTH core (Rhee et al., 1998).

A secondary structure prediction for the PerA N-terminus suggests that it consists of several α-helical regions interspersed with loops and with a β-strand-rich region between amino acids 31 and 75 (Fig. 5A). We used this prediction together with the alignment of the N-terminal domain with those of the related VirF and Rns proteins to design mutations to test the functional importance of different regions within this domain of PerA. Mutations within the central part of the domain (amino acids 30–95), which includes the β-strand rich region, had the greatest effect on PerA function, with G37W, W68G and L94P mutants being severely compromised (although W68G retained partial activity at the per promoter only). G37W in particular is predicted to disrupt a short loop or turn between the first two β-strand sequences, suggesting that this predicted structure may exist in the folded N-terminal domain. An inactivating mutation of VirF, Y23S, has previously been isolated in the same region (Porter and Dorman, 2002). Although mutations targeting residues in the flanking predicted helical regions (A20P, K108E) or other residues close to these regions, which are well-conserved in VirF and Rns (Y26F, D100A) did not affect PerA function, the randomly acquired L94P mutation, also within a predicted α-helix, did inactivate the protein. Further analysis of our PerA N-terminal mutants, together with those in the C-terminal domain, was limited by our inability to obtain sufficient quantities of these proteins to test their in vitro properties. A better understanding of how the N-terminal domain contributes to transcriptional activation will require more extensive mutagenesis and, ideally, a structural study.

Experimental procedures

Bacterial strains and plasmids

The bacterial strains and plasmids used in this study are listed in Table 3, and the oligonucleotide primers used in their construction are shown in Table 4. The EPEC perABC locus plus flanking DNA was amplified as a 2.8 kb fragment from E2348/69 plasmid DNA using primers per1 + perex, and cloned into the low-copy vector pTH19kr after digestion with BamHI + HindIII to generate pTHperABC. An in-frame deletion of perA (amino acids 178–267) was made initially in a smaller 1.9 kb clone of perABC in pTH19kr by outward polymerase chain reaction using primers perAdel1 + perAdel2, followed by digestion with DpnI (to remove template DNA) plus EcoRV, and re-ligation. The deleted locus was then subcloned as a SacI-BglII fragment to replace the intact locus in full-length pTHperABC, giving pTHperΔA(BC+). In-frame deletions of perB (amino acids 18–113) and perC (amino acids 11–70) were made similarly using the full-length pTHperABC template and primers delperB3 + delperB4 and delperC3 + delperC4, respectively, followed by DpnI + StuI digestion and re-ligation to give pTHperΔB(AC+) and pTHperΔC(AB+). The double deletion derivative pTHperΔBC(A+) was made by the same procedure but using the delperB3 + delperC4 primer pair. The perB, perC and perBC deletions were combined with the perA deletion by repeating the same deletion procedures on the pTHperΔA(BC+) template, resulting in pTHperΔAB(C+), pTHperΔAC(B+) and pTHperΔABC. To construct the arabinose-inducible PerA and PerC clones, the perA and perC open reading frames (ORFs) were amplified from pTHperABC using the primers perAfara + perArara and perCfara + per2, respectively, digested with PstI + HindIII or KpnI + HindIII, respectively, and cloned downstream of the araBAD promoter in similarly digested pBAD33 to give pBADPerA and pBADPerC. A tagged PerA derivative for expression and purification (pET30PerA) was made by amplifying the perA ORF from pTHperABC using perAex1 + perAex2, digesting with NcoI + BamHI and cloning into pET30b; this entire tagged ORF was also amplified with PETBADF + PETBADR and cloned into KpnI-HindIII-digested pBAD33 to give an arabinose-inducible construct (pBADH6SPerA) to test the functionality of the tagged protein. The virF gene plus flanking DNA was amplified as a 1.8 kb fragment from the total S. flexneri BS184 DNA using primers virF3 + virF4, digested with BamHI + XbaI and cloned into pACYC184 to yield pACVirF. This was then used as a template to generate an arabinose-inducible clone, pBADVirF, using primers virFfara + virF2, digesting with PstI + HindIII and ligating into pBAD33, and a tagged expression clone, pET30VirF, using primers virFNT + virFCT, digesting with NcoI + BamHI and ligating into pET30b. All constructs were verified by sequencing the entirety of the inserted DNA.

Table 3. E. coli and S. flexneri strains and plasmids used in this study.
 BL21 (λDE3)/pLysSExpression strain with IPTG-inducible T7 RNA polymeraseNovagen
 BS184S. flexneri 2a mxiC-lacZMaurelli et al. (1984)
 DH5arecA cloning strainOur stocks
 E2348/69Wild-type EPEC O127:H6Levine et al. (1978)
 MG1655 bfp-lacZMG1655 with chromosomal bfp-lacZ fusion at lacThis work
 MG1655 bfp-lacZ hnsbfp-lacZ hns-206::AprThis work
 MG1655 bfp-lacZ ihfbfp-lacZ ihfA82::Tn10This work
 MG1655 bfp-lacZ hns ihfbfp-lacZ hns-206::AprihfA82::Tn10This work
 MG1655 LEE1-lacZMG1655 with chromosomal LEE1-lacZ fusion at lacThis work
 MG1655 per-lacZMG1655 with chromosomal per-lacZ fusion at lacThis work
 MG1655 per-lacZ hnsper-lacZ hns-206::AprThis work
 MG1655 per-lacZ ihfper-lacZ ihfA82::Tn10This work
 MG1655 per-lacZ hns ihfper-lacZ hns-206::AprihfA82::Tn10This work
 MG1655 sacB kanMG1655 with lacZY replaced by sacB kanThis work
 NEC007BL21 (λDE3) ihfA82::Tn10Gally et al. (1996)
 PD32MC4100 hns-206::AprDersch et al. (1993)
 pACYC184P15A replicon, Cmr, TcrNew England Biolabs
 pACvirFvirF plus flanking DNA in pACYC184This work
 pAJR25MG1655 lacI and lacA regions in pIB307This work
 pAJR32pAJR25 with sacB kan cassette inserted between lacIAThis work
 pAJR36pAJR25 with promoterless lacZ gene inserted between lacIAThis work
 pBAD33Arabinose-inducible expression vector, CmrGuzman et al. (1995)
 pBADH6SPerAArabinose-inducible His6-S-tagged PerAThis work
 pBADPerAArabinose-inducible PerAThis work
 pBADPerCArabinose-inducible PerCThis work
 pBADVirFArabinose-inducible VirFThis work
 pET30bT7-controlled His6-S-tagged expression vector, KmrNovagen
 pET30PerAT7-controlled His6-S-tagged PerAThis work
 pET30VirFT7-controlled His6-S-tagged VirFThis work
 pIB307pMAK705-based temperature-sensitive vector, CmrBlomfield et al. (1991)
 pIBbfpA-lacbfp promoter fused in-frame with lacZ in pAJR36This work
 pIBler-lacLEE1 promoter fused in-frame with lacZ in pAJR36This work
 pIBperA-lacper promoter fused in-frame with lacZ in pAJR36This work
 pSPT18SP6/T7 in vitro transcription vector, AprRoche Molecular
 pSPTperAInternal fragment of perA in pSPT18This work
 pSPTperBInternal fragment of perB in pSPT18This work
 pSPTperCInternal fragment of perC in pSPT18This work
 pTH19krLow-copy cloning vector, KmrHashimoto-Gotoh et al. (2000)
 pTHperABCperABC plus flanking DNA in pTH19krThis work
 pTHperΔA(BC+)pTHperABC with in-frame deletion in perAThis work
 pTHperΔB(AC+)pTHperABC with in-frame deletion in perBThis work
 pTHperΔC(AB+)pTHperABC with in-frame deletion in perCThis work
 pTHperΔAB(C+)pTHperABC with in-frame deletions in perA and perBThis work
 pTHperΔAC(B+)pTHperABC with in-frame deletions in perA and perCThis work
 pTHperΔBC(A+)pTHperABC with in-frame deletion in perBCThis work
 pTHperΔABCpTHperABC with in-frame deletions in perA and perBCThis work
Table 4.  Oligonucleotide primers used in this study.
PrimerSequence (5’-3’)aUse
  • a

    . Sequences complementary to the target DNA are shown in upper case, sequences unique to the oligonucleotide primer in lower case. Restriction sites used for cloning are underlined.

per1cgagtggatccGCCAGGCTGCCGTGGCloning of perABC and per–lacZ fusion
perrexcgagtaagcttGTTACAGGCTCTGGCCloning of perABC
perAdel1cgagtgatatcCAACTCAATTACTTTTATAACCCperA in-frame deletion
perAdel2cgagtgatatcACACAAGGAACATTGCCATAAGperA in-frame deletion
delperB3cgagtaggcctCATACATCCATCAAAAACTACTTCTTTCTCperB in-frame deletion
delperB4cgagtaggcctACGATGGAAGGAGGCGGTGATATTCperB in-frame deletion
delperC3cgagtaggcctGTACTTCGCCTTTTTATCTCTTATTTCCperC in-frame deletion
delperC4cgagtaggcctGCATATAAGAAAATGGGGCTTGTAAATGperC in-frame deletion
PerAFaractagcactgcagGAAACAAACGCGCATGAAGGCloning of arabinose-inducible PerA
PerARaractagcaaagcttCGAGTGCTCATTGCAAATAAGCloning of arabinose-inducible PerA
perCFaractagcaggtaccGATACATACACTACTATAAACCloning of arabinose-inducible PerC
per2cgagtaagcttGCTCAGTCCGGTTGGTGCloning of arabinose-inducible PerC
perAex1cgagtccATGgTTACATCTAAAAAAGAAATGCCloning of tagged PerA for expression
perAex2cgagtggatccTTATGGCAATGTTCCCloning of tagged PerA for expression
PETBADFctagcaggtaccGTGAGCGGATAACAATTCCCCloning of arabinose-inducible tagged PerA
PETBADRGGCCGCAAGCTTGTCGACCloning of arabinose-inducible tagged PerA
virF3cgagttctagaAACCCATCTGGCAATAGCCloning of multicopy VirF
virF4cgagtggatccATTATAGTCGGCTAAATGCloning of multicopy VirF
virFfaractagcactgcagCTCTGTAAACACTAAATATAGCloning of arabinose-inducible VirF
virF2cgagtaagcttAACCCATCTGGCAATAGCCloning of arabinose-inducible VirF
virFNTctagcaccATGgTGGATATGGGACATAAAAACCloning of tagged VirF for expression
virFCTctagcaggatccTTAAAATTTTTTATGATATAAGTAAAATTTCTTTGGCloning of tagged VirF for expression
lacI5’gggggagctcCGTTATTTCTTGATGTCTCTGACAmplification of lacI flanking sequence
lacI3’ggggatccGCCTGGGGTGCCTAATGAGTGAGAmplification of lacI flanking sequence
lacA5’ccggatccAATGACCGAAAGAATAAGAGAmplification of lacA flanking sequence
lacA3’aaaactgcagATGTCTTTTGTGACGATACTAmplification of lacA flanking sequence
lacZ5’ccggatccggtaccATGACCATGATTACGGATTCACAmplification of promoterless lacZ gene
lacZ3’ccagatctCCTTACGCGAAATACGGGCAmplification of promoterless lacZ gene
perATlFcgagtggtaccCAATAACGCTAAATTCTCCTCCloning of per–lacZ fusion
bfpTlF1cgagtggtaccCACCATTGCAGATTCAATCAAAGCloning of bfp–lacZ fusion
bfp1CGTAAAGGATCCTTTTTCTGCCloning of bfp–lacZ fusion
LEE1TlFcgagtggtaccCTCTATAAGCTGAATGTATGGCloning of LEE1–lacZ fusion
LEE1AcgagtggatccGTGAAACGGTTCAGCCloning of LEE1–lacZ fusion
perARPacgagtggatccCCAAAAACTGGAAACTAGGCGAConstruction of perA riboprobe
perARPbcgagtggtaccGGTGTTGTGTTGTAATATTCCConstruction of perA riboprobe
perBRPacgagtggatccGAAGTAGTTTTTGATGGATGTATGConstruction of perB riboprobe
perBRPbcgagtggtaccGCCTCCTTCCATCGTTATCAConstruction of perB riboprobe
perCRPacgagtggatccGAGATAAAAAGGCGAAGTACTTGConstruction of perC riboprobe
perCRPbcgagtggtaccCATTTACAAGCCCCATTTTCTConstruction of perC riboprobe
perBSaGTTATAACTGGGGCCTGper bandshift fragment, 5’ end
per1RcgagtaagcttCCACCTTCATGCGCGper bandshift fragment, 3’ end
bfpBSaGTACCGGAAGTCAAATTCbfp bandshift fragment, 5’ end
bfpBSbCGTATTAATAGGTCACGGbfp bandshift fragment, 3’ end
virB1CTATGGAGCTCTCACATCAGvirB bandshift fragment, 5’ end
virB2GACAAAAGTTAAATGCAGTGvirB bandshift fragment, 3’ end
virFU1CACAGATATTGCCTAAGAAAAGvirF bandshift fragment, 5’ end
virFU2CTAAGAAATTTCTGTAACAGTGvirF bandshift fragment, 3’ end
perA168aCTAAGAGCATCGTAGcCAGGGTTATAAAAGTAperA D168A mutagenesis (forward)
perA168bTACTTTTATAACCCTGgCTACGATGCTCTTAGperA D168A mutagenesis (reverse)
perA175aGGGTTATAAAAGTAATTGcGTTGGATATATCCperA E175A mutagenesis (forward)
perA175bGGATATATCCAACgCAATTACTTTTATAACCCperA E175A mutagenesis (reverse)
perA234aGATAAAAACATAGATGcAATATCTTGTTTGGTTGperA E234A mutagenesis (forward)
perA234bCAACCAAACAACATATTgCATCTATGTTTTTATCperA E234A mutagenesis (reverse)

Construction of single-copy per–lacZ, bfp–lacZ and LEE1–lacZ fusions

Single-copy translational fusions of the per, bfp and LEE1 promoters to lacZ were constructed in the chromosome of MG1655 at the lac locus using a temperature-sensitive vector based on pIB307 (Blomfield et al., 1991) and a derivative of MG1655 in which the lacZ and lacY genes are replaced by a sacB kan cassette encoding kanamycin-resistance and sucrose-sensitivity. To create the MG1655 sacB kan strain, flanking regions of the lac operon were amplified with primer pairs lacI 5′ + lacI3′ and lacA5′ + lacA3′ and cloned into pIB307 to produce pAJR25. A sacB-kan cassette was then cloned into the BamHI site of pAJR25 as described previously (Roe et al., 2003), creating pAJR32 containing strain-specific lacI-A regions flanking the sacB kan cassette. This cassette was then exchanged into the chromosomal lac locus of MG1655 as described previously (Roe et al., 2003). Primers lacZ5′ + lacZ3′ were used to amplify the lacZ gene from MG1655, which was then digested with BamHI and BglII and cloned into pAJR25. The resultant plasmid (pAJR36) contains single BamHI and KpnI sites 5′ of the lacZ gene allowing promoters of interest with their initial coding sequence to be cloned in-frame with the reporter gene. A fragment encompassing the first 22 codons of perA, together with 206 bp upstream of the ATG start codon, which includes the mapped per promoter (Martínez-Laguna et al., 1999) and PerA binding site (Ibarra et al., 2003), was amplified from pTHperABC with primers per1 and perATlF, digested with BamHI + KpnI, then cloned in-frame with the lacZ gene in pAJR36 to generate pIBperA-lac. This plasmid was transformed into MG1655 sacB kan at 30°C, then forced to recombine into the chromosome by repeated subculturing in the presence of chloramphenicol at 42°C for 48 h, followed by repeated subculturing at 30°C for 48 h without antibiotic selection to resolve the plasmid co-integrates. Successful recombinants in which the per-lacZ construct had replaced the chromosomal sacB-kan cassette were then selected on agar plates containing 6% sucrose, and checked for kanamycin-sensitivity, chloramphenicol-sensitivity and the absence of plasmid DNA. A bfp promoter fragment containing 270 bp upstream of the ATG and 22 codons of the bfpA ORF, covering the mapped bfp promoter (Bustamante et al., 1998) and PerA binding site (Ibarra et al., 2003) was amplified from E2348/69 plasmid DNA using primers bfp1 and bfpTlF, digested with BamHI and KpnI and cloned into pAJR36 to generate pIBbfpA-lac. Finally, an LEE1 promoter fragment which contains 24 codons of the ler gene plus 368 bp of DNA upstream of the ATG and includes the mapped EPEC LEE1 promoter (Mellies et al., 1999) and all upstream DNA as far as the EPEC-specific ERIC sequence (Elliott et al., 1998) was amplified from E2348/69 chromosomal DNA with primers LEE1A and LEE1TlF, digested with BamHI and KpnI, then cloned into pAJR36 to yield pIBler-lac. The pIBbfpA-lac and pIBler-lac constructs were integrated into the chromosome of MG1655 sac kan as for pIBperA-lac to give strains MG1655 bfp-lacZ and MG1655 LEE1-lacZ.

Growth conditions and enzyme assays

For analysis of Per-dependent transcription, strains were grown overnight in Luria–Bertani (LB) medium and then subinoculated 1:100 into high-glucose DMEM containing Hepes and lacking phenol red (Gibco-Invitrogen Corporation, cat. no. 21063-029). The DMEM cultures were grown to mid-logarithmic phase (OD600∼0.6) and assayed for β-galactosidase activity as described by Miller (1992). Cultures were assayed in duplicate and the assays were repeated at least twice. Standard deviations were less than 10%. antibiotic selection with chloramphenicol (20 µg ml−1) or kanamycin (25 µg ml−1) was as appropriate. When arabinose-inducible derivatives of PerA and PerC were used, arabinose was used at a maximum concentration of 1.0% to overcome catabolite repression of the araBAD promoter by glucose in the medium (Guzman et al., 1995).

Transduction with bacteriophage P1cml

The hns-206::Apr and ihfA82::Tn10 alleles were transduced from the donor strains PD32 and NEC007, respectively, into the MG1655 per-lacZ and MG1655 bfp-lacZ fusion strains either singly or in combination using bacteriophage P1cml as described by Silhavy et al. (1984).

RNA extraction, probe synthesis and Northern blotting

Total cellular RNA was extracted from cultures grown to mid-log phase by cell lysis in boiling RNA extraction buffer (20 mM sodium acetate [pH 5.2], 2% SDS, 0.3 M sucrose) followed by phenol extraction and DNaseI treatment as described previously (Porter and Dorman, 1997a). Internal fragments (∼200–300 bp) of the perA, perB and perC genes were amplified with the primer pairs perARPa + perARPb (spanning codons 179–259 of the perA ORF), perBRPa + perBRPb (codons 9–118 of perB) and perCRPa + perCRPb (codons 4–79 of perC), respectively, digested with BamHI and KpnI and cloned into the in vitro transcription vector pSPT18. The resulting plasmid constructs (pSPTperA, pSPTperB and pSPTperC) were linearized with HindIII and transcribed in vitro with T7 RNA polymerase to generate digoxigenin-UTP-labelled RNA probes using a DIG-RNA labelling kit supplied by Roche Molecular according to the manufacturer's instructions. Samples of total cellular RNA (20 µg) were electrophoresed on MOPS-formaldehyde-agarose gels, transferred to Hybond-N+ positively charged membranes (Amersham), and hybridized with the labelled RNA probes overnight. After stringent washing, bound probes were detected with antidigoxigenin–alkaline phosphatase conjugate and the chemiluminescent substrate CSPD (Roche Molecular) as described previously (Porter and Dorman, 1997a).

Protein purification and mobility shift assays

Two-litre cultures of BL21 (λDE3)/pLysS transformed with pET30PerA, pET30VirF or mutated derivatives of pET30PerA (see below) were grown to mid-log phase (OD600∼0.6) in LB under kanamycin selection (50 µg ml−1), then induced with IPTG at a concentration of 1 mM for 3 h. The cells were pelleted by centrifugation, washed in 160 ml of 50 mM Hepes, 0.1 M NaCl, and resuspended in 60 ml of His-Tag binding buffer (5 mM imidazole, 0.25 M NaCl, 20 mM Tris pH 7.9, 10% glycerol, 0.1% Triton X-100, 10 mM β-mercaptoethanol). After cell lysis by sonication (MSE Soniprep 150; 3–6 pulses of 1 min at 5 µm amplitude interspersed with cooling on ice), cell debris was pelleted by centrifugation at 16 000 rpm for 20 min, and the cleared supernatants were loaded onto columns of Ni2+-NTA agarose (Qiagen) equilibrated with His-Tag binding buffer. The columns were washed twice with 30 ml of His-Tag binding buffer supplemented with 30 mM imidazole to remove weakly bound proteins, and the specifically bound AraC-like proteins were then eluted with His-Tag binding buffer supplemented with 0.5 M imidazole. Eluted proteins were dialyzed against HilD buffer (20 mM Tris pH 7.9 50 mM KCl, 5 mM NaCl, 1 mM EDTA, 1 mM DTT, 20% glycerol, 0.1% NP-40; Schechter and Lee, 2001) overnight, and then stored in aliquots at −80°C. Final soluble protein yields were ∼0.08 mg l−1 of cell culture for PerA and ∼0.5 mg l−1 of cell culture for VirF as assessed using a Bradford assay kit (Bio-Rad). Mutant derivatives of PerA gave much lower yields (<0.01 mg l−1) and rapidly precipitated upon storage at either −80°C or 4°C. Fragments for use in mobility shift assays were amplified with primer pairs: (i) perBSa and per1R (amplifying between −141 and −4 with respect to the perA start codon); (ii) bfpBSa and bfpBSb (amplifying between −176 and −28 with respect to the bfpA start codon); (iii) virB1 and virB2 (amplifying between −194 and −33 with respect to the virB start codon); and (iv) virFU1 and virFU2 (amplifying between −252 and −47 with respect to the virF start codon), then end-labelled with digoxigenin-ddUTP using an end-labelling kit (Roche Molecular) according to the manufacturer's instructions. Approximately 1.2 ng of labelled probe was mixed with purified protein in binding buffer (10 mM Tris pH 7.5, 1 mM EDTA, 5 mM NaCl, 50 mM KCl, 8% glycerol, 0.05 mg ml−1 BSA, 1 mM DTT) in the presence of ∼1000-fold excess (1 µg) of non-specific competitor poly[d(I-C)] and incubated at room temperature for 20 min. Protein–DNA complexes were electrophoresed on native 4% polyacrylamide gels at 100 V, 4°C for 5 h and then transferred to Hybond-N+ membranes (Amersham). The labelled DNA probes were then detected using the alkaline phosphatase/CSPD-based chemiluminescence system (Roche Molecular) as described for Northern blots.

Site-directed mutagenesis of perA

Site-directed mutagenesis of the perA gene was performed using the pTHperABC template, oligonucleotide primers as listed in Table 4, and a Quikchange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. Random mutations in perA, which were isolated during attempts to clone the locus into medium-copy vectors were subcloned into pTHperABC after the release of the mutation on a BamHI-PacI fragment (I85M mutation) or a PacI-ScaI fragment (L94P, A221T and R226G mutations). To generate arabinose-inducible and tagged derivatives of selected mutant proteins, the appropriate mutated templates were amplified with perAfara + perArara and perAex1 + perBex1 primer pairs and cloned into pBAD33 and pET30b, respectively, as described for the wild-type perA gene.

Computational analysis

Protein secondary structure predictions were performed at using the PHD algorithm (Rost and Sander, 1993; 1994). Protein sequences were aligned by the blast program (Altschul et al., 1990) at DNA curvature predictions were performed using (Munteanu et al., 1998) at


We would like to thank the staff of the ICMB sequencing service for DNA sequencing, Padraig Deighan for sending the hns-206::Apr mutant, Seiichi Yasuda of the Cloning Vector Collection, National Institute of Genetics, Japan for providing the pBAD33 and pTH19kr vectors, and members of the ZAP Laboratory for useful discussions. This work was supported by grants nos. 066381/Z/01/Z and 065574/Z/01/Z from the Wellcome Trust.