Host-encoded functions that regulate the transfer operon (tra) in the virulence plasmid of Salmonella enterica (pSLT) were identified with a genetic screen. Mutations that decreased tra operon expression mapped in the lrp gene, which encodes the leucine-responsive regulatory protein (Lrp). Reduced tra operon expression in an Lrp− background is caused by lowered transcription of the traJ gene, which encodes a transcriptional activator of the tra operon. Gel retardation assays indicated that Lrp binds a DNA region upstream of the traJ promoter. Deletion of the Lrp binding site resulted in lowered and Lrp-independent traJ transcription. Conjugal transfer of pSLT decreased 50-fold in a Lrp− background. When a FinO− derivative of pSLT was used, conjugal transfer from an Lrp− donor decreased 1000-fold. Mutations that derepressed tra operon expression mapped in dam, the gene encoding Dam methyltransferase. Expression of the tra operon and conjugal transfer remain repressed in an Lrp− Dam− background. These observations support the model that Lrp acts as a conjugation activator by promoting traJ transcription, whereas Dam methylation acts as a conjugation repressor by activating FinP RNA synthesis. This dual control of conjugal transfer may also operate in other F-like plasmids such as F and R100.
This study shows that a bacterial global regulator, the leucine-responsive regulatory protein (Lrp), is required for traJ transcription and, hence, for tra operon expression. We also provide evidence that Lrp and Dam methylation play opposite roles in the regulation of tra operon expression: Lrp is an activator of conjugal transfer, whereas Dam methylation is a repressor. Although the significance of these findings remains to be understood, the nature of the bacterial functions involved is noteworthy. Lrp is a master regulatory protein that co-ordinates the expression of a large number of bacterial operons in response to nutrient availability (D’Ari et al., 1993; Calvo and Matthews, 1994). Dam methylation is also known to participate in mechanisms of global control (Heithoff et al., 1999; Reisenauer et al., 1999). Hence, an attractive idea is that Lrp and Dam methylation might regulate mating in response to environmental or physiological stimuli. Because of the metabolic and energetic burden placed on the host by the plasmid-encoded conjugation apparatus, tight control of mating can be expected to have a high adaptive value (Zatyka and Thomas, 1998).
Our study has been carried out with pSLT, the virulence plasmid of Salmonella enterica strain LT2 (Spratt et al., 1973). pSLT is self-transmissible (Ahmer et al., 1999) and carries a wild-type (FinO+ FinP+) system of fertility control (Smith et al., 1973). Several lines of evidence suggest that the control of conjugal plasmid transfer by Lrp and Dam methylation described below may also operate in F, R100 and other F-like plasmids.
A genetic screen for regulators of tra operon expression
Tn10dCm insertions that altered the expression of the tra operon of plasmid pSLT were sought in isogenic Dam+ and Dam− strains of S. enterica. Use of a Dam− strain was justified by the following rationale: as Dam methylation represses tra operon expression (Torreblanca et al., 1999), additional regulators might be identified more easily in a Dam− background. For instance, activators of conjugation might be identified as suppressors of tra operon derepression in a Dam− host. The strains used (SV3003 and SV3080) carry a MudJ-generated lac fusion in traB, the sixth gene of the tra operon of plasmid pSLT (Torreblanca and Casadesús, 1996). This fusion is Lac− in a Dam+ background and Lac+ in a Dam− background (Torreblanca and Casadesús, 1996). To avoid unwanted insertions in pSLT, Tn10dCm pools were prepared in a pSLT-cured strain, SV3081. SV3003 and SV3080 were then transduced with the pools. Cmr transductants were selected on plates containing chloramphenicol and Xgal. Two classes of isolates were sought among the transductants: (i) Cmr isolates that formed Lac+ (blue) colonies in a Dam+ background (SV3003) identified potential tra operon repressors (class I); and (ii) Cmr isolates that formed Lac− (white) colonies in a Dam− background (SV3080) identified potential tra operon activators (class II).
Examination of about 35 000 independent Tn10dCm inserts provided us with isolates of both classes, which were subjected to reconstruction analysis by P22 HT transduction (Torreblanca and Casadesús, 1996). One boundary of each Tn10dCm insertion was then sequenced to identify the gene affected by the mutation.
Class I isolates (10/10) carried Tn10dCm insertions in dam (data not shown), thereby confirming that DNA adenine methylation represses tra operon expression (Torreblanca and Casadesús, 1996; Torreblanca et al., 1999). Activation of tra operon expression by mutations other than dam was not observed.
Class II isolates (2/2) carried Tn10dCm insertions in lrp, the gene for the leucine-responsive regulatory protein (Anderson et al., 1976). Use of the collection allele lrp-41::Tn5 confirmed that lack of Lrp prevents tra operon derepression in a Dam− background (data not shown). In the absence of Dam methylation, a null lrp mutation caused a 15-fold decrease in tra operon expression (see data for the mutation lrp-42::Tn10dCm in Fig. 1). The similarity in expression of the tra operon in Lrp+ Dam+ and Lrp− Dam+ hosts (Fig. 1) may not be significant, given the low β-galactosidase activities obtained (<10 Miller units). In fact, lrp mutations do decrease tra operon expression in a FinO− background, where tra expression is higher (data not shown).
Effect of lrp mutations on the expression of tra, finP and traJ transcriptional units
To identify the promoter regulated by Lrp in the transfer regulon of pSLT, we investigated the effect of lrp mutations on the expression of traY, the first gene of the tra operon, and on the expression of the regulatory genes traJ and finP. The assays were carried out with trans-criptional lac fusions, carried on the multicopy plasmids pIZ903, pIZ880 and pIZ898, all derived from the fusion vector pIC552 (Torreblanca et al., 1999). Plasmid pIZ903 carries a 610 bp SspI–EcoRV fragment that generates a traY::lac fusion. Plasmid pIZ898 carries an EcoRI–HinfI fragment of pSLT, 360 bp long, that generates a traJ::lac fusion. Plasmid pIZ880 carries a 295 bp HinfI–NaeI fragment of pSLT that generates a finP::lac fusion. Lrp+ and Lrp− strains harbouring these plasmid-borne fusions, all in a pSLT− background, were grown in liquid minimal medium. When the cultures reached the late exponential phase, β-galactosidase activities were measured. The activity of the traJ::lac fusion showed a significant decrease in an Lrp− background, whereas the traY::lac and finP::lac fusions remained largely unaltered (Fig. 2). This result suggested that Lrp activates the transcription of traJ. The concomitant observation that the tra operon remains repressed in an Lrp+ TraJ− background (Fig. 2; data not shown) is also relevant, as it suggests that Lrp activates the tra operon indirectly by enhancing traJ transcription.
Binding of Lrp protein to the traJ upstream activating sequence: gel retardation assays
Computer analysis of DNA sequences upstream of the traJ promoter of plasmid pSLT revealed a region homologous to the consensus sequence for Lrp binding (Calvo and Matthews, 1994). The nucleotide sequences used were from the EMBL database, and their accession numbers are as follows: pSLT, AJ011572 (Torreblanca et al., 1999); F, U01159 (Penfold et al., 1996); R100, M13054 (Fee and Dempsey, 1986). This site is conserved in other F-like plasmids such as F and R100 (Fig. 3). To investigate whether the Lrp protein was able to bind this DNA region, gel retardation assays were carried out. A fragment of pIZ860, which carries the putative upstream activating sequence (UAS) of traJ as well as the traJ promoter and part of the traJ coding region (Torreblanca et al., 1999), was polymerase chain reaction (PCR) amplified using the universal and reverse primers of pBluescript. The amplified fragment was purified and digested with restriction enzymes. The resulting fragments were incubated for 20 min at 22°C with various concentrations of a crude extract from an Lrp-overproducing strain. As a control, a crude extract of a non-overproducing, isogenic strain was used. The mixtures were then separated by electrophoresis in a polyacrylamide gel. Two representative experiments are shown in Fig. 4. When the fragment under study was digested with HincII, retardation of the 313 bp fragment was observed (left). Upon EcoRV + DraI digestion, retardation of the 84 bp fragment was observed (right). Because the 313 bp and the 84 bp fragments both contain the region homologous to the Lrp binding site, these observations provided evidence that Lrp binds to the UAS of traJ.
Binding of Lrp upstream of traJ was confirmed by gel retardation assays using pure Lrp protein. For this purpose, a 487 bp fragment of plasmid pIZ860 was PCR amplified using the universal primer of Bluescript and a traJ primer designed ad hoc. The amplified DNA fragment was digested with HindIII to eliminate a 91 bp region of pBluescript. The resulting 396 bp fragment, which contains traJ DNA from –383 to +8, was purified, end-labelled and mixed with aliquots of pure Lrp protein. Binding reactions were allowed to proceed for 20 min at room temperature. Electrophoretic separation was then carried out in a non-denaturing polyacrylamide gel. A representative experiment is shown in Fig. 5. Retardation of the DNA fragment under study is clearly observed, thereby confirming that Lrp binds the traJ UAS.
Deletion analysis of the traJ UAS: effect on traJ expression
To ascertain whether the Lrp binding site located upstream of traJ does or does not act as a transcription-activating sequence, we determined whether deletion of this site affected traJ expression. For this purpose, plasmids pIZ1537 and pIZ1538 were constructed. Both carry a traJ::lac transcriptional fusion and differ only in the DNA stretch located upstream of the traJ promoter: pIZ1537 carries the Lrp binding site, but pIZ1538 does not (Fig. 6). Analysis of β-galactosidase activity in Lrp+ and Lrp− backgrounds indicated that Lrp-mediated activation of traJ transcription requires the presence of the Lrp binding site located upstream of traJ (Fig. 6). We thus propose that binding of Lrp to the UAS upstream of traJ activates traJ transcription.
Effect of lrp mutations on conjugal plasmid transfer
Together, the results described in the former sections suggest that Lrp might act as an activator of conjugal plasmid transfer. To test this hypothesis, matings were performed to compare pSLT plasmid transfer from isogenic Lrp+ and Lrp− donors (SV4201 and SV4305 respectively). The recipient was SV3081, a pSLT-cured strain. Use of a pSLT derivative carrying a Kmr marker in the unrelated gene spvA (Hensel et al., 1995) permitted the selection of transconjugants on minimal plates supplemented with kanamycin. Use of minimal medium counterselected the auxotrophic donors. The frequency of pSLT plasmid transfer was calculated per donor bacterium. A 50-fold decrease in the frequency of conjugal transfer was observed when an Lrp− donor was used (Table 1).
Table 1. Effect of lrp and dam mutations on pSLT plasmid transfer.
Plasmid transfer frequencies are given per donor bacterium; all are medians of >12 independent experiments. Donors of pSLT were SV4201 (Lrp+ Dam+), SV 4305 (Lrp−), SV4202 (Dam−) and SV4306 (Lrp− Dam−). Donors of pSLT ΔfinO were SV4522 (Lrp+ Dam+), SV 4524 (Lrp−), SV4523 (Dam−) and SV4525 (Lrp− Dam−). The recipient was SV3081 in all crosses.
4 × 10−6
8 × 10−8
5 × 10−5
8 × 10−8
1 × 10−3
9 × 10−7
2 × 10−2
3 × 10−6
Matings were also carried out with a derepressed (FinO−) derivative of plasmid pSLT constructed for this study. Donors were isogenic Lrp+ and Lrp− strains (SV4522 and SV4524 respectively). The recipient strain and the procedure for the selection of transconjugants were as above. As expected, the FinO− derivative of pSLT was transferred at higher frequency than the wild-type plasmid (Table 1). However, when an Lrp− donor was used, the frequency of pSLT FinO− plasmid transfer decreased >103-fold. This extremely high difference provides further evidence that Lrp acts as a conjugation activator.
Because Dam methylation is known to repress con-jugal transfer of F-like plasmids (Torreblanca and Casadesús, 1996), we also examined pSLT transfer from Dam− Lrp+ and Dam− Lrp− donors (SV4202 and SV4306 respectively). The recipient strain and the mating design were as above. A dam mutation increased pSLT plasmid transfer >10-fold (Table 1), as reported previously for F (Torreblanca et al., 1999). In contrast, the presence of both dam and lrp mutations in the donor strain restored the repression of pSLT transfer to a level similar to that of an Lrp− mutant (Table 1). When a FinO− derivative of pSLT was used, a dam mutation increased plasmid transfer around 20-fold. As expected, transfer of pSLT ΔfinO from a Dam− Lrp− donor restored repression. In this case, however, repression of plasmid transfer was incomplete (Table 1). This result can be accommodated in the regulatory circuitry of F-like plasmids (Firth et al., 1996; Zatyka and Thomas, 1998) as an additive effect of finO and dam mutations on traJ expression.
Here, we identify the leucine-responsive regulatory protein (Lrp) as a positive regulator of tra operon expression in the virulence plasmid of Salmonella enterica. Lrp does not act at the tra operon directly but at the regulatory gene traJ, as indicated by the following observations: (i) lrp null mutations cause a decrease in traJ transcription (Fig. 2); (ii) Lrp protein binds upstream of the traJ promoter (Figs 4 and 5); (iii) deletion of the Lrp binding site renders the traJ promoter Lrp independent, and lowers its activity (Fig. 6); and (iv) lrp mutations cause a 50-fold decrease in conjugal plasmid transfer (Table 1).
When a derepressed (FinO−) pSLT derivative is used, the decrease in conjugal plasmid transfer caused by an lrp mutation is >1000-fold. This magnification of the difference in the absence of fertility inhibition further supports the model that Lrp acts as an activator of pSLT plasmid transfer: a FinO− plasmid is no longer derepressed when an lrp mutation reduces traJ transcription.
The addition of leucine to the medium did not affect the frequencies of pSLT transfer from any of the donors used (data not shown). It is well known that some Lrp-regulated operons respond to the presence of exogenous leucine, but others do not (Ernsting et al., 1993; Calvo and Matthews, 1994). Hence, the traJ gene of pSLT may belong to the latter class.
The screen has also identified mutations in the dam gene as the only class that derepresses tra operon expression. In a Dam− background, expression of the tra operon increases 15-fold (Fig. 1) and results in high frequencies of plasmid transfer (Table 1). In F-like plasmids that harbour a wild-type system for fertility inhibition, conjugal transfer is tightly repressed and occurs at frequencies below 10−5 per donor cell (Smith et al., 1973; Firth et al., 1996). A key element for such a repression is Dam methylation, which is required to sustain high levels of FinP RNA synthesis (Torreblanca et al., 1999). In fact, dam mutations increase transfer even in the F episome (Torreblanca et al., 1999), whose conjugation system is derepressed by a finO mutation (Cheah and Skurray, 1986).
Strains harbouring both dam and lrp mutations express the tra operon at very low levels (Fig. 1); in other words, lrp mutations reduce the derepression of tra caused by dam mutations. The results from mating experiments are analogous: pSLT transfer is derepressed in an Lrp+ Dam− donor, but not in an Lrp− Dam− donor (Table 1). Again, the use of a derepressed plasmid magnifies these effects: transfer of pSLT ΔfinO decreases 1000-fold from an Lrp− Dam+ donor and 300-fold from an Lrp− Dam− donor. All these observations can be accommodated in the regulatory circuitry of F-like plasmids (Frost et al., 1994; Firth et al., 1996). In an Lrp+ Dam− background, reduced FinP RNA synthesis permits increased translation of traJ mRNA, resulting in tra operon derepression (Torreblanca et al., 1999). In an Lrp− background, a dam mutation is unable to derepress the tra operon because of inefficient traJ transcription.
The control mechanisms discussed above are not unique to the Salmonella virulence plasmid. Regulation of FinP RNA synthesis by Dam methylation also occurs in F and R100 (Torreblanca et al., 1999; E. M. Camacho and J. Casadesús, unpublished). The Lrp binding site located upstream of traJ is highly conserved in pSLT, F and R100 (Fig. 3), and evidence that Lrp regulates conjugal transfer of F and R100 has been obtained (E. M. Camacho and J. Casadesús, unpublished data). Hence, regulation of conjugal transfer by Lrp and Dam methylation may be a common trait among F-like plasmids.
Lrp and Dam methylation control of tra operon expression is a feature shared by fimbrial operons such as the pap operon of E. coli uropathogenic strains (Blyn et al., 1989; Braaten et al., 1992), the pef operon of S. enterica (Nicholson and Low, 2000) and others (van der Woude and Low, 1994; van der Woude et al., 1996). However, the involvement of the same regulatory elements does not imply an analogous design. In the pap operon, Dam methylation and Lrp both target the same DNA region, upstream of the papBA promoter (Nou et al., 1993). In pSLT, Dam methylation and Lrp regulate the expression of different genes (finP and traJ respectively), and their targets do not overlap. Different design may be viewed as a consequence of different regulatory strategies: in the pap operon, Lrp and DNA adenine methylation (together with other effectors) provide a switch for phase variation (van der Woude et al., 1996), whereas in the tra operon of F-like plasmids they appear to make a repressor/activator pair. This double control may tentatively be viewed as a cross-check mechanism that controls TraJ synthesis at the transcriptional level via Lrp and at the post-transcriptional level via FinP.
The involvement of global regulators such as Dam methylation and Lrp in the regulation of TraJ synthesis suggests that mating may be controlled by global regulatory mechanisms of the bacterial cell. Lrp is known to co-ordinate cell metabolism in response to the availability of nutrients in the external environment (D’Ari et al., 1993; Calvo and Matthews, 1994). Because pilus building and DNA transfer are both energetically expensive (Curtiss and Stallions, 1967; Zatyka and Thomas, 1998), the assessment of metabolic conditions via Lrp control might provide an obvious advantage. In turn, Dam methylation can sense DNA replication (Roberts et al., 1985), DNA damage (Schlagman et al., 1986) and other processes that alter DNA topology (Maas, 2001). Information on the DNA status inside the cell can likewise be relevant to decide whether conjugal transfer should be undertaken or not. Dam methylation might also detect other kinds of signals, as the methylation state of specific GATC sites in the genome is controlled by growth conditions (van der Woude et al., 1998), and transcription of specific sets of genes can be affected by Dam methylation patterns (Heithoff et al., 1999). In addition, previous studies have shown that a redox sensor of the bacterial cell, ArcA, regulates tra operon expression in F-like plasmids (Strohmaier et al., 1998; Taki et al., 1998). One may thus hypothesize that host-encoded functions such as Lrp, Dam methylation and ArcA might be part of signal transduction pathways that regulate bacterial conjugation in response to cellular or environmental cues. If future research confirms this view, we might be able to identify factors that influence the rates of conjugal plasmid transfer in natural environments. Such knowledge might prompt measures to reduce the dissemination of plasmids and inspire pharmacological treatments to prevent plasmid spread.
Bacterial strains and plasmids
The strains of S. enterica used in this study are listed in Table 2. All belong to the serovar Typhimurium and derive from strain LT2. The allele lrp-41::Tn5 was provided by D. R. Hillyard, Department of Pathology, University of Utah, Salt Lake City, USA. To introduce a dominant, selectable marker in pSLT, the insertion spvA::Kmr (Hensel et al., 1995) was transferred to strain LT2 by P22 HT transduction (Schmieger, 1972). The spvA::Kmr allele was provided by D. Holden, Department of Infectious Diseases, Imperial College School of Medicine, London. Plasmid pIZ860 carries an EcoRI–NaeI fragment of pSLT that contains the UAS of traJ, the traJ promoter and part of the traJ coding region (Torreblanca et al., 1999). pIZ880, pIZ898 and pIZ903 are pIC552 derivatives that carry finP::lac, traJ::lac and traY::lac transcriptional fusions respectively (Torreblanca et al., 1999). pIZ1537 and pIZ1538 both derive from pIZ898. In pIZ1537, the DNA region upstream of the EcoRV site has been deleted in the traJ UAS; the deletion does not remove the Lrp binding site (Fig. 6). In pIZ1538, a deletion extending 5′ from the DraI site has eliminated the Lrp binding site (Fig. 6). pJWD1, provided by D. A. Low, University of California, Santa Barbara, CA, USA., is a pTrc99A derivative that carries the E. coli lrp gene and permits Lrp overproduction (Ernsting et al., 1993). pIZ1526 is a pJWD1 derivative from which the lrp gene has been removed. Construction of pIZ1526 was achieved by NcoI + SalI digestion of pJWD1, end filling with Klenow DNA polymerase and religation.
E medium (Vogel and Bonner, 1956) was used as minimal medium. The rich medium was Difco nutrient broth (8 g l−1) with added NaCl (5 g l−1). Solid media contained agar at 1.5% final concentration. Auxotrophic requirements, antibiotics and Xgal were used at the final concentrations described elsewhere (Torreblanca et al., 1999).
Mutagenesis with Tn10dCm
The pSLT-cured strain SV3081 was mutagenized with Tn10dCm (Torreblanca and Casadesús, 1996). Pools of 5000 colonies, each carrying an independent Tn10dCm insert, were then prepared and lysed with phage P22 HT. The lysates were used to transduce strains SV3003 and SV3080, selecting chloramphenicol-resistant transductants on E plates supplemented with Xgal.
Levels of β-galactosidase activity were assayed as described by Miller (1972).
Cultures of the donor and the recipient were prepared in minimal medium. For matings, late exponential cultures were used. Cells were harvested by centrifugation and washed with 10 mM MgSO4. Aliquots of both strains, 500 μl each, were then mixed and centrifuged for 30 s at 13 000 r.p.m. The pellet was resuspended in 50 μl of 10 mM MgSO4 and sucked onto a Millipore filter of 0.45 μm pore size. The filters were placed on E plates without glucose and incubated at 37°C for 3–4 h. After mating, the mixtures were diluted in 10 mM MgSO4 and spread on selective plates. As controls, 0.1 ml of both the donor and the recipient cultures were also spread on selective plates. Conjugation frequencies were calculated per donor bacterium.
Plasmid DNA was obtained by alkaline lysis (Stephen et al., 1990). Restriction enzymes were purchased from New England Biolabs and Boehringer Mannheim. Deoxyribo-nucleotides and Klenow DNA polymerase were purchased from Promega Biotech. T4 polynucleotide ligase was from Boehringer Mannheim; ligation was achieved by incubation at 16°C for >12 h. For blunt end ligation, a low-ATP buffer was used (New England Biolabs). Preparation of electrocom-petent cells and electrotransformation has been described elsewhere (Torreblanca et al., 1999).
Chromosomal DNA was prepared from 1.5 ml overnight cultures in LB. Cells were harvested by centrifugation and resuspended in 1.5 ml of 10 mM Tris-HCl, pH 8.0, 25 mM EDTA, pH 8.0. A volume of 0.55 ml of lysozyme solution (10 mg ml−1 in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, pH 8.0) was added. The mixture was incubated for 20 min at 37°C. Proteinase K (100 μg ml−1) was then added and incubated for 1 h at 55°C. After three or four extractions with phenol and chloroform–isoamyl alcohol (24:1), DNA was precipitated with ammonium acetate and absolute ethanol and, finally, suspended in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA, pH 8.0. DNA was sheared further with a 23G needle and cleaned in a Sephadex G50 column. The primer used for sequencing of the boundaries of Tn10 insertions was 5′-CTAATGACAAG ATGTGT-3′ (Way and Kleckner, 1984). The conditions for sequencing were as follows: (i) incubation of 2 μg of chromosomal DNA at 58°C for 30 min; (ii) addition of 13 pmol of primer, 16 μl of BigDye Terminator mix, 1 μl of Thermo-Fidelase I (Fidelity Systems) and H2O up to 40 μl; (iii) PCR amplification: 95°C, 5 min (95°C, 30 s/53°C, 20 s/60°C, 4 min) (55 cycles), 60°C, 5 min, 4°C, hold; (iv) DNA precipitation with Centrisep columns (Applied Biosystems–Perkin-Elmer).
Preparation of crude cell extracts
Overnight LB cultures of strains MC4100/pJWD1 (Lrp overproducer) and MC4100/pIZ1526 (Lrp non-overproducer) were diluted 1:100 in fresh LB. When the cultures reached an OD600 of 0.6–0.7, 0.5 mM IPTG was added to induce transcription of the plasmid-borne lrp gene. After incubation for 2 h at 37°C, the cells were harvested, and the pellet was resuspended in 5 ml of lysis buffer (50 mM Na2HPO4, 10 mM β-mercaptoethanol, 10 mM EGTA, 0.1 M NaCl and 10% glycerol). The preparation was sonicated eight times for 30 s at minimal intensity and centrifuged at 15 000 r.p.m. for 10 min. The supernatant was then collected.
Gel retardation assays with crude cell extracts
The binding buffer contained 150 mM Na2HPO4, 0.4 M NaCl, 0.1 M MgCl2 and 0.5 μg ml−1 bovine serum albumin (BSA). Samples were prepared by mixing 2 μl of cell extract (diluted as necessary), 2 μl of binding buffer, 2.8 μl of pure glycerol and 3.2 μl of DNA. The samples were incubated for 20 min at 22°C before loading a polyacrylamide gel. After the run, the gels were stained with ethidium bromide.
Gel retardation assays with pure Lrp protein
Lrp protein, purified as described by Ernsting et al. (1993), was obtained from D. A. Low, University of California, Santa Barbara, USA. Lrp was judged to be ≥ 95% pure by SDS– PAGE and Coomassie blue staining.
For gel retardation analysis, a 391 bp DNA fragment carrying the traJ UAS was end labelled with Klenow DNA polymerase in the presence of [α-32P]-dCTP. DNA-binding reactions were prepared to obtain a final volume of 20 μl. Each reaction contained 0.4 pmol of labelled DNA, 4 μl of Lrp protein diluted in binding buffer and 0.5 μg of competitor DNA [poly-(dI–dC)]. The final composition of the binding buffer was 20 mM Tris-HCl, pH 8.0, 75 mM NaCl, 5 mM MgCl2, 1 mM dithiothreitol (DTT), 12.5% glycerol, 0.1 mg ml−1 BSA and 25 μg ml−1 poly-(dI–dC). Binding was allowed to proceed for 20 min at room temperature. Five microlitres of loading buffer was then added. Samples were subjected to electrophoretic separation in a non-denaturing, 5% polyacrylamide gel prepared in 1× TBE. Electrophoresis was carried out at 200 V for 2–3 h. After drying, gels were analysed with a Fujifilm FLA-3000 betascope.
Construction of a FinO− derivative of plasmid pSLT
The finO locus of pSLT was knocked out by adapting to S. enterica the method of Datsenko and Wanner (2000). Primers were designed to eliminate a 540 bp DNA stretch from the finO ORF (McClelland et al., 2001). The resulting mutation, ΔfinO1::Kmr, was transduced with P22 to strain LT2. The kanamycin resistance cassette was then elimi-nated by recombination with plasmid pCP20 (Datsenko and Wanner, 2000). PCR amplification using primers from both sides of the finO locus identified kanamycin-sensitive isolates that carried a deletion in the finO locus of pSLT. To introduce a dominant, selectable marker in the pSLT ΔfinO1 plasmid, a spvA::Kmr insertion (Hensel et al., 1995) was transduced with P22 HT. The resulting plasmid, pSLT ΔfinO1 spvA::Kmr, was then transferred by conjugation to suitable backgrounds to obtain S. enterica strains SV4522, SV4523, SV4524 and SV4525, all isogenic. Each carries a his mutation that permits counterselection when the strain is used as donor in a mating experiment. In addition, the strains carry lrp and dam mutations, alone or combined.
This study was supported by grant PM97-0148-CO2-02 from the Ministry of Science and Technology of Spain. We are deeply grateful to David Low for advice and discussions. Pure Lrp protein for gel retardation experiments was also a generous gift from David Low. We also thank David Holden and David Hillyard for providing strains, and Marjan van der Woude and Dick d’Ari for critical reading of the manuscript.