Involvement of differential efficiency of transcription by Eσs and Eσ70 RNA polymerase holoenzymes in growth phase regulation of the Escherichia coli osmE promoter

Authors

  • Patricia Bordes,

    1. Laboratoire de Microbiologie et Génétique Moléculaire, UPR 9007 CNRS, 118 Route de Narbonne, F-31062 Toulouse Cedex, France.
    Search for more papers by this author
  • Francis Repoila,

    1. Laboratoire de Microbiologie et Génétique Moléculaire, UPR 9007 CNRS, 118 Route de Narbonne, F-31062 Toulouse Cedex, France.
    Search for more papers by this author
  • Annie Kolb,

    1. Unité de Physicochimie des Macromolécules Biologiques, URA 1773 CNRS, Institut Pasteur, 25 rue du Dr Roux, F-75724 Paris Cedex 15, France.
    Search for more papers by this author
  • Claude Gutierrez

    Corresponding author
    1. Laboratoire de Microbiologie et Génétique Moléculaire, UPR 9007 CNRS, 118 Route de Narbonne, F-31062 Toulouse Cedex, France.
    • *For correspondence. E-mail clg@ibcg.biotoul.fr; Tel. ( + 33) 561 33 58 72; Fax ( + 33) 561 33 58 86.

    Search for more papers by this author

Abstract

Transcription of the gene osmE of Escherichia coli is inducible by elevated osmotic pressure and during the decelerating phase of growth. osmE expression is directed by a single promoter, osmEp. Decelerating phase induction of osmEp is dependent on the σs (RpoS) factor, whereas its osmotic induction is independent of σs. Purified Eσs and Eσ70 were both able to transcribe osmEpin vitro on supercoiled templates. In the presence of rpoD800, a mutation resulting in a thermosensitive σ70 factor, a shift to non-permissive temperature abolished induction of osmEp after an osmotic shock during exponential phase, but did not affect the decelerating phase induction. Point mutations affecting osmEp activity were isolated. Down-promoter mutations decreased transcription in both the presence and the absence of σs, indicating that the two forms of RNA polymerase holoenzyme recognize very similar sequence determinants on the osmE promoter. Three up-promoter mutations brought osmEp closer to the consensus of Eσ70-dependent promoters. The two variant promoters exhibiting the highest efficiency became essentially independent of σsin vivo. Our data suggest that Eσs transcribes wild-type osmEp with a higher efficiency than Eσ70. A model in which an intrinsic differential recognition contributes to growth phase-dependent regulation is proposed. Generalization of this model to other σs-dependent promoters is discussed.

Introduction

The product of the gene rpoS, an RNA polymerase sigma factor, called RpoS, σs, σ38 or KatF (Mulvey and Loewen, 1989; Lange and Hengge-Aronis, 1991; Tanaka et al., 1993), controls a regulon of 30 or more genes expressed in response to starvation and during the transition to stationary phase (for recent reviews, see Loewen and Hengge-Aronis, 1994; Hengge-Aronis, 1996a). Mutations in rpoS are pleiotropic. They lead to increased sensitivity to UV radiation, elevated temperature, high salt and hydrogen peroxide (Hengge-Aronis et al., 1991; Lange and Hengge-Aronis, 1991; McCann et al., 1991), suggesting that σs should be considered as a master regulator of stress response (Hengge-Aronis, 1996b).

Sequence comparisons between promoters transcribed by Eσs revealed some peculiarities (Espinosa-Urgel et al., 1996) but, altogether, no consensus sequence has been identified that could account for the recognition of all the promoters transcribed by Eσs and would allow the prediction of such a promoter. A number of these promoters are transcribed by the two forms of RNA polymerase, Eσs and Eσ70. Examples have been described in which auxiliary proteins are involved in the selectivity of recognition by Eσs and Eσ70. In the case of the csgBA promoter, Arnqvist et al. (1994) reported that the nucleoid-associated protein H-NS prevents its transcription by Eσ70, whereas Eσs is able to transcribe it even in the presence of H-NS. Altuvia et al. (1994) reported that the dps promoter can be transcribed by Eσ70 during exponential phase, and that such transcription depends on stimulation by the activator OxyR in oxidative conditions. Upon entry into stationary phase, transcription of dps is ensured by Eσs, stimulated by the nucleoid-associated protein IHF and OxyR independent. However, for most promoters transcribed by Eσs and Eσ70, the basis of differential recognition by the two holoenzymes remains to be established.

Transcription of the gene osmE, which encodes an envelope protein of unknown function, is induced at elevated osmotic pressure and during the decelerating phase, before entry into stationary phase. A single promoter, osmEp, is responsible for both the osmotic and the decelerating phase induction of osmE (Gutierrez et al., 1995; Conter et al., 1997). The decelerating phase induction is dependent upon a functional σs, whereas the induction upon an osmotic shock is still observed in the absence of σs (Conter et al., 1997). Therefore, it was hypothesized that osmEp belongs to the class of promoters recognized by both Eσs and Eσ70. In the present work, we present evidence supporting this hypothesis and investigate the basis of osmEp recognition by analysing a collection of point mutations affecting this promoter. Our data suggest that an intrinsic difference in transcription efficiency of the osmE promoter by Eσs and Eσ70 contributes to its growth phase regulation. Based on observations reported in other promoters, we propose that this property could also participate in the regulation of at least a subset of σs-dependent promoters.

Results

s and Eσ70 are both able to transcribe osmEp in vitro

Because osmEp full expression was strongly dependent upon the presence of σs (Conter et al., 1997; see also Fig. 4B), we postulated that this promoter could be transcribed by Eσs. To address this question more directly, we tested whether the two forms of purified RNA polymerase were able to transcribe osmEpin vitro. A 154 bp EcoRI DNA fragment carrying the wild-type promoter was cloned in the unique EcoRI site of the vector pJCD01 (Table 2). The resulting plasmid was extracted from strain DH5α and used, as supercoiled template, to programme in vitro transcription experiments with purified Eσ70 or Eσs. As shown in Fig. 1, a transcript of the length expected from osmEp (≈150 nucleotides) was obtained with the two forms of RNA polymerase (Fig. 1, lanes 1 and 4). The correct assignment of the osmE transcript was confirmed using two mutant promoters (see below). It was very low in transcriptions performed with a down-promoter mutant, osmEp28Δ (Fig. 1, lanes 3 and 6). A quantification of the ratio between the amount of osmE′ RNA and the pJC01-derived RNA I indicated that osmE transcript was obtained in higher amounts with osmEp12T, an up-promoter mutant (Fig. 1, lanes 2 and 5).

Figure 4.

Effect of up-promoter mutations on osmEp expression.

A. Sequence of the osmE promoter region. The substituted nucleotides in the mutants are shown above the sequence.

B. Expression of the wild-type promoter during growth.

C. Effect of osmEp8A mutation.

D. Effect of osmEp12T mutation.

E. Effect of osmEp31C mutation. Overnight cultures in LB0N (open symbols) or LB0N +  0.4 M NaCl (closed symbols) of rpoS+ strains (circles) and their rpoS derivatives (triangles) were diluted 1000-fold in the same preheated medium, and A600 (interrupted lines) and β-galactosidase activities (filled lines) were monitored during growth.

Table 1. Relative efficiency of five σs-dependent promotersa.
  β-Galactosidase activity (Miller units)
Promoter LB0NLB0N + 0.4 M NaClScore
  • a

    . Strains were grown overnight in the indicated media and samples were harvested to assay β-galactosidase activity.

osmE p rpoS  +  34065749
rpoS::Tn1042119 
osmY p rpoS  +  30445435
rpoS::Tn101423 
osmC p2 rpoS  +  17530646
rpoS::Tn102128 
treA p rpoS  +  9126844
rpoS::Tn10714 
fic p rpoS  +  1049833
rpoS::Tn1034 
Figure 1.

RNA polymerase holoenzymes Eσ70 and Eσs are both able to transcribe osmEpin vitro. Supercoiled pJCD01 derivatives were transcribed in vitro with Eσ70 (lanes 1–3) or Eσs (lanes 4–6). Autoradiography of a fixed and dried electrophoresis gel of the products is shown. The templates used were: pJCD01- inline image (lanes 1 and 4); pJCD01-osmEp12T (up-promoter; lanes 2 and 5); pJCD01-osmEp28Δ (down-promoter; lanes 3 and 6). Positions of the osmE′ mRNA (≈150 nucleotides) and vector-derived RNA I (107–108 nucleotides) are indicated. Numbers at the bottom of the lanes indicate the ratio osmE′ RNA/RNA I.

Inactivation of σ70 abolished induction of osmEp after an osmotic shock but not during decelerating phase

The preceding observations demonstrated that Eσs can transcribe osmEpin vitro. In order to investigate whether it actually does so in vivo, we took advantage of rpoD800, a mutation giving a thermosensitive σ70 (Grossman et al., 1983). This mutation was introduced by P1 transduction in strain CLG570, which carries a transcriptional fusion controlled by osmEp, leading to strain CLG621. When grown at 30°C in rich medium at low osmotic pressure, CLG621 exhibited the decelerating phase accumulation of β-galactosidase activity typical of osmE promoter regulation (Fig. 2A, compare with Fig. 4B for expression in an rpoD+ isogenic strain at 37°C). This accumulation stopped when cells had entered stationary phase, and the A600 of the culture did not increase any more. When shifted to 40°C at the onset of decelerating phase (A600 of 0.8), CLG621 still showed an accumulation of β-galactosidase, lasting until that the A600 stopped increasing (Fig. 2A). In contrast, the induction observed at 30°C after an osmotic shock during exponential phase (A600 of 0.15) was abolished by a shift to 40°C (Fig. 2B). These data indicated that σ70 is necessary for the expression of osmEp during exponential phase, and confirmed the rapid inactivation of σ70 after a shift to 40°C. Yet, osmEp remained transcribed during decelerating phase at 40°C, strongly suggesting that Eσs is able to transcribe the osmE promoter in vivo.

Figure 2.

Effect of inactivation of σ70 on induction of osmEp during decelerating phase or after osmotic shock. Strain CLG621 was grown in LB0N at 30°C, and A600 of the culture (open symbols) and β-galactosidase specific activity (closed symbols) were measured over time.

A. At the point indicated by an arrow, part of the culture was shifted to 40°C (triangles), while part was maintained at 30°C (circles).

B. At the time indicated by an arrow, 0.4 M NaCl was added, and part of the culture was shifted to 40°C (triangles), while part was maintained at 30°C (circles).

Isolation of mutations within the osmE promoter region

In order to identify the nucleotides necessary for recognition of osmEp by Eσ70 and Eσs, we isolated point mutations altering the promoter efficiency. The oligonucleotides osmE1 and osmE2 were used to amplify from the matrix pCG512 a DNA fragment encompassing the promoter region (Table 2). The amplification was performed with Tfl DNA polymerase, after incubation of pCG512 DNA with nitrous acid to increase the frequency of transitions (Diaz et al., 1991). Amplified DNA fragments were cleaved with BamHI and HindIII and cloned in the promoter probe vector pCB267 (Schneider and Beck, 1986). After transformation of strain DH5α, plasmids with active promoter fragments gave blue colonies on rich medium with Xgal. Darker blue and pale to very pale blue colonies were kept as candidates carrying up- and down-promoter mutations respectively. The candidate plasmids were purified and sequenced: 12 down-promoter (Fig. 3) and two up-promoter (Fig. 4) single mutations were obtained. They were named according to the position of the modified nucleotide with respect to the transcription start, and according to the nature of the nucleotide introduced in the substitution. Two down-promoter mutations were single base deletions, named osmEp28Δ and osmEp13Δ, according to the position of the deleted nucleotide. EcoRI DNA fragments (154 bp) carrying either the wild-type or the mutated osmE promoter were purified and cloned in the unique EcoRI site of the vector pOM41. The resulting plasmids were then used to introduce the various osmE promoter regions in front of a malP–lac operon fusion by homologous recombination (see Experimental procedures). In the resulting strains, the osmEplac transcriptional fusions were present in single copy at the malA locus on the chromosome and produced intact β-galactosidase under the control of the wild-type or mutant osmE promoter. An rpoS359::TeK mutation was introduced in all the fusion strains by P1 transduction, using a P1 stock grown on strain CLG141 (Table 2).

Figure 3.

osmE p down-promoter mutations. Sequence of the osmE promoter region is indicated. The − 10 and − 35 hexanucleotides are boxed. + 1: transcription start. rpoS+ and rpoS derivatives of fusion strains carrying the wild-type or mutated promoter were grown in LB0N (L0) or LB0N +  0.4 M NaCl (L0.4), and β-galactosidase activities were measured in overnight cultures. Average values of at least three independent assays are shown. Numbers between parentheses indicate the inhibition ratio resulting from the mutations.

Effect of down-promoter mutations on transcription of osmEp

The 12 mutations presented at the bottom of Fig. 3 gave a large range of effects on promoter efficiency. Among the 10 single base substitutions, mutation osmEp49C only had a very weak effect (1- to 1.3-fold). The nine others reduced the β-galactosidase production from twofold (osmEp6C) to an almost complete abolition (osmEp30G or osmEp7C). These mutations clustered within or next to the − 35 or − 10 sequences of osmEp. Comparison of the effect of the mutations in wild-type or rpoS backgrounds demonstrated that the effects on expression were within the same range in both conditions. The only exceptions were the mutations giving an expression of only a few β-galactosidase units (osmEp35C, osmEp30G and osmEp7C). This might result from overestimation of the very low activities, giving values in the order of the background [1–2 Miller units, as measured with strain pop2887, a derivative of pop3125 carrying a deletion of the malP promoter in front of the φ(malP–lac) fusion]. Therefore, as the substitutions isolated here alter the expression of osmEp similarly in the presence and absence of σs, we conclude that the mutated nucleotides must participate in the recognition of the promoter by the two forms of RNA polymerase.

Two mutations resulted in single basepair deletions, which shortened the spacer from 17 to 16 bp. osmEp28Δ reduced expression approximately 12-fold in the presence of σs, and nine- and 13-fold at low and high osmotic pressure, respectively, in the absence of σs. Therefore, this deletion exerted similar effects on σs-dependent and -independent transcription, like the substitutions. In contrast, osmEp13Δ affected σs-dependent transcription much more than σs-independent transcription. Indeed, osmEp13Δ reduced transcription in the presence of σs approximately 52-fold (fourfold more than osmEp28Δ) but only six- and 11-fold at low and high osmotic pressure, respectively, in the absence of σs (that is less than osmEp28Δ). To clarify the role of nucleotides C13 and A28 and check it with a spacer of normal length, the two mutations osmEp13T and osmEp28G were constructed by site-directed mutagenesis on the 154 bp osmE promoter fragment. Their effect was tested after construction of the corresponding fusion strains (Fig. 3, top). osmEp28G reduced the promoter expression twofold, in both rpoS+ and rpoS backgrounds. In contrast, osmEp13T reduced the expression twofold in the absence of σs, but sixfold in an rpoS+ background. Therefore, changing the C at position − 13 appeared to be more severe for transcription by Eσs than by Eσ70.

We also constructed and analysed mutation osmEp8G in the same way. Changing C8 into a G reduced expression twofold, similarly in rpoS+ and rpoS backgrounds.

Effect of up-promoter mutations

Strain CLG570, carrying a wild-type osmE promoter, gave kinetics of β-galactosidase production (Fig. 4B) similar to that reported earlier with longer promoter fragments (Conter et al., 1997), demonstrating that 154 bp are sufficient for a normal regulation of osmEp. The mutations osmEp8A, constructed by site-directed mutagenesis, or osmEp31C and osmEp12T, isolated after nitrous acid mutagenesis, made promoters closer to the consensus for the − 35 (TTGACA) or − 10 (TATAAT) regions of the σ70-dependent promoters (Fig. 4A). The three of them increased the promoter activity. osmEp8A mostly resulted in higher expression during exponential phase, observed in both rpoS+ and rpoS backgrounds (compare Fig. 4B and C). An rpoS-dependent decelerating phase induction was still observed with this mutation, and its magnitude was similar to that observed with the wild-type promoter. As for osmEp31C and osmEp12T, both led to an important increase in promoter efficiency and completely modified the kinetics of expression (Fig. 4D and E). After a period of very weak transcription in early exponential phase, resulting in a decrease in the β-galactosidase specific activity, the transcription of the mutant promoters started in mid-log phase, leading to progressive accumulation of β-galactosidase, which reached a plateau in stationary phase. No specific induction during the decelerating phase was observed with the osmEp31C and osmEp12T mutant promoters. Another striking feature was that both promoters had lost σs dependency. They could even be more efficient in the absence of σs than in a wild-type genetic background. When used to programme in vitro transcription, osmEp12T behaved as an up-promoter mutation, with both Eσ70 and Eσs forms of RNA polymerase (Fig. 1, lanes 2 and 5). However, the stimulation of transcription resulting from this mutation was more important for Eσ70 than for Eσs.

Relative efficiency of a number of σs-dependent promoters

In order to compare the relative level of transcription from a number of σs-dependent promoters in identical rpoS+ or rpoS backgrounds, we used a set of strains in which the same φ(malP–lacZ) transcriptional fusion was expressed under the control of different promoters. Strain CLG442 carries an osmCp2lac fusion (Bouvier et al., 1998). We constructed strains CLG294, CLG700 and CLG701, which carry fusions with treAp (Hengge-Aronis et al., 1991; Repoila and Gutierrez, 1991), osmYp (Yim et al., 1994) and ficp (Hiratsu et al., 1995) respectively (see Experimental procedures). As shown in Table 1, with this reporter system, we reproduced the σs dependency of all these promoters. All but ficp were more active at elevated than at low osmotic pressure. In the absence of σs, all but ficp conserved some residual expression, very low but still significantly above the background. Finally, osmEp appeared as the strongest of the five tested promoters.

Discussion

In the present work, we have demonstrated that osmEp, the promoter of E. coli osmE, can be transcribed in vitro by Eσs and Eσ70, two forms of RNA polymerase containing the sigma factors σs or σ70. As shown in Fig. 2, expression of this promoter requires an intact σ70 factor during the exponential phase of growth, but is still possible without σ70 during the decelerating phase. Taken together with previous data (Conter et al., 1997), our results strongly suggest that the two forms of RNA polymerase are sequentially involved in transcription of osmEpin vivo. As osmEp is poorly transcribed during exponential growth and becomes more active in the decelerating phase (Fig. 4B), this implies that Eσs is more efficient than Eσ70 in transcribing osmEp. During the exponential phase, osmEp is weakly transcribed by Eσ70, ensuring a low basal level of OsmE. During the decelerating phase, accumulation of σs shifts the holoenzyme to Eσs, resulting in a higher level of transcription and an important accumulation of OsmE.

Among the different sigma factors identified in E. coli, σs is the most similar to σ70, especially in the DNA-binding regions (Lonetto et al., 1992). Eσs is able to transcribe a number of promoters also recognized by Eσ70 (Tanaka et al., 1993), suggesting that common features could be involved in promoter recognition. This hypothesis is reinforced by DNase I footprinting and FeBABE cleavage experiments, which demonstrated similar positioning of both holoenzymes on common target promoters: lacUV5, galP1 and osmYp (Tanaka et al., 1995; Colland et al., 1999). The down-promoter mutations isolated in this work show that it is also the case with osmEp. Indeed, all the down-promoter mutants isolated here, affecting the − 35 and − 10 regions or the length of the spacer, reduce efficiency of transcription by both Eσs and Eσ70. Furthermore, most of them appear to affect transcription by both forms of RNA polymerase to similar extents, indicating that the nucleotides of osmEp recognized by the two holoenzymes must be essentially the same. The only noticeable exception is the mutation osmEp13T, which affects transcription by Eσs more severely than by Eσ70 (Fig. 3). From 33 σs-dependent promoters, Espinosa-Urgel et al. (1996) derived a consensus − 10 region of seven nucleotides (CTATACT), with Cs at positions one and six, corresponding to the Cs in position − 13 and − 8 in osmEp. osmEp13T confirms the importance of a C nucleotide at position one of the − 10 motif for recognition by Eσs. However, expression of this promoter in an rpoS background shows that a C at this position is also preferred to a T for transcription by Eσ70 (Fig. 3). As for the C nucleotide in position six of the − 10 motif, changing it into a G has the same effect for Eσ70 and Eσs (Fig. 3). Mutation osmEp8A changes this C into an A, consensus for Eσ70-dependent promoters. As shown in Fig. 4C, this mutation stimulates transcription by Eσ70, resulting in a higher expression during exponential phase, observed in both the presence and the absence of σs. However, as it does not significantly affect the σs-dependent decelerating phase induction, osmEp8A appears to be almost neutral for transcription by Eσs. Altogether, our results suggest that, although both forms of RNA polymerase holoenzyme recognize similar sequence determinants, an important difference between Eσs and Eσ70 is that they tolerate variations from the consensus differentially.

The two up-promoter mutations osmEp31C and osmEp12T point out another element that seems important in designating osmEp as a σs-dependent promoter. In vitro experiments demonstrate that the osmEp12T promoter is stronger than the wild type with both forms of RNA polymerase holoenzyme (Fig. 1, compare for instance lanes 1 and 2 for Eσ70, lanes 4 and 5 for Eσs). However, the two up-promoter mutants become functionally σs independent in vivo (Fig. 4D and E). Therefore, in order to be σs dependent in vivo, osmEp has to be a relatively weak σ70-dependent promoter. Owing to its overall base composition, wild-type osmEp would be intrinsically less active when transcribed by Eσ70 than by Eσs. The two up-promoters would be variants transcribed by Eσ70 at equivalent or higher levels than by Eσs. No induction of these up-promoters is observed during the decelerating phase because, in contrast to wild-type osmEp, the appearance of Eσs would no longer result in increased transcription efficiency.

In conclusion, the model suggested by this work is as follows. Eσ70 and Eσs would have a similar consensus for their target promoters. However, to be σs dependent, a given promoter should differ significantly from the consensus, and thus be poorly transcribed by Eσ70. The two forms of RNA polymerase could accommodate differently to deviations from the consensus, and each promoter would have a characteristic efficiency of transcription by Eσ70 and Eσsin vivo, according to the nature of the modifications it carries. This model is consistent with all the observations presented here with osmEp. It is also consistent with observations with other σs-dependent promoters. First, it is well established that a number of promoters can be recognized by Eσ70 and Eσsin vitro (Tanaka et al., 1993), and also in vivo (Schellhorn et al., 1998). Second, promoters fitting this model should deviate from the consensus, and therefore be rather weak. As shown in Table 1, the programme described by Mulligan et al. (1984) to predict σ70-dependent promoters gives very low scores to the five examples of σs-dependent promoters studied here. These five promoters are weak when compared with the induced lac promoter, which gives β-galactosidase activities ≈10-fold higher than osmEp (not shown). The weakest is ficp, the only promoter totally dependent on σs. In addition, after a random search for σs-dependent promoters, Schellhorn et al. (1998) found that, among 10 such promoters, osmYp was by far the stronger, confirming that weakness is a rather general property of σs-dependent promoters. We also note that the relative efficiency of the five promoters tested here is dependent upon the presence of σs. Indeed, osmYp is stronger than osmCp2 in the presence of σs, but becomes weaker in its absence. Such behaviour is also expected with the proposed model. Finally, Wise et al. (1996) described two types of promoter mutations that also argue in favour of this model. On the one hand, osmYTT is a mutation making the − 35 region of osmYp closer to the consensus for σ70. It creates a derivative of osmYp with an increased and σs-independent expression during the exponential phase, but remains inducible during the decelerating phase in a σs-dependent manner. This behaviour, similar to that of osmEp8A, demonstrates that the type of mutants identified here with osmEp can also be found with other σs-dependent promoters. On the other hand, although the wild-type proU promoter is not σs dependent and does not exhibit any induction upon entry into stationary phase, mutation proUcc, changing the − 35 hexamer (TTGCCT) into a region matching the σ70 consensus at only two positions (CCGCCT), results in a promoter with a much reduced expression that is now stimulated upon entry into stationary phase in a σs-dependent fashion (Wise et al., 1996).

Obviously, other factors are probably also important in establishing the complete growth phase regulation of σs-dependent promoters. For instance, DNA supercoiling is likely to be such an additional factor in the case of osmEp. Our previous results suggested that a high level of DNA supercoiling stimulates transcription of osmEp by Eσ70 (Conter et al., 1997), and the great increase in negative supercoiling after an osmotic shock is most probably an important component of the high level of transcription of osmEp observed under such conditions in Fig. 2B. In contrast, Kusano et al. (1996) have shown that supercoiling is able to reduce transcription of some promoters by Eσs, and this could explain why expression of osmEp after an osmotic shock is not ensured by Eσs, in spite of the stabilization of σs observed at elevated osmotic pressure (Muffler et al., 1996). Several global regulators have also been shown to modulate expression of promoters, such as osmYp (Lange et al., 1993), csiDp (Marschall et al., 1998) or osmCp2 (Bouvier et al., 1998). A differential effect of repressors or activators on transcription by Eσ70 and Eσs can lead to a specific induction upon accumulation of σs, as described for the effect of H-NS on the csgBA promoter (Arnqvist et al., 1994). However, it is becoming clear that multiple mechanisms contribute to the regulation of many stress-responsive promoters (Hengge-Aronis, 1999), and we believe that an intrinsic differential efficiency of recognition by Eσ70 and Eσs is very likely to contribute to the growth phase regulation of transcription.

Experimental procedures

Bacterial strains, plasmids and phages

The bacterial strains, all derived from Escherichia coli K-12, and the plasmids used in this study are listed in Table 2. Bacteriophage P1vir and recombinant λ624 (Kohara et al., 1987) were from the laboratory collection.

Table 2.  Bacterial strains and plasmids.
Strain or plasmidGenotypeReference or source
E. coli strains
MC4100FaraD139Δ(argF-lac)U169 ptsF25 Casadaban (1976)
rpsL150 relA1 flbB5301 deoC1 rbsR  
pop3125MC4100 φ(malP–lac) Débarbouilléet al. (1978)
CAG1130Fthi leu thr lacY tonA supE galK rpoD-800 zgh::Tn10 Grossman et al. (1983)
CLG141MC4100 rpoS-359::Tek Conter et al. (1997)
CLG294MC4100 φ[treAp+-φ(mal –lac)]This study
CLG442MC4100 φ[osmCp2+-φ(malP–lac)] Bouvier et al. (1998)
CLG570MC4100 φ[osmEp+-φ(malP–lac)]This study
CLG621CLG570 rpoD-800 zgh::Tn10This study
CLG700MC4100 φ[osmYp+-φ(malP–lac)]This study
CLG701MC4100 φ[ficp+-φ(malP–lac)]This study
Plasmids
pCG512pUC18 derivative carrying a 910 bp osmE DNA fragment Gutierrez et al. (1995)
pOM41pBR322 derivative, promoter recombination vector Vidal-Ingigliardi and Raibaud (1985)
pSB118pUC18 with an EcoRI site on both sides of the linker Vidal-Ingigliardi and Raibaud (1985)
pJCD01pUC19 derivative, vector for in vitro transcription Marschall et al. (1998)

Chemicals and media

Media were made as described by Miller (1992). Cells were grown aerobically at 37°C in LB0N medium (Lucht et al., 1994). LB0N is LB medium (Miller, 1992) without NaCl. When indicated, this medium was supplemented with 0.4 M NaCl to increase the osmolarity. Xgal was added to solid media at a concentration of 40 µg ml− 1. Ampicillin, kanamycin and tetracycline were used at concentrations of 100, 40 and 10 µg ml− 1 respectively.

Enzyme assays

β-Galactosidase activities were assayed by measuring the hydrolysis by SDS/chloroform-treated cells of ONPG, as described by Miller (1992).

Methods used with nucleic acids

Isolation of plasmid DNA, digestion with restriction enzymes, ligation with T4 DNA ligase and transformation were carried out as described by Maniatis et al. (1982) and Silhavy et al. (1984). Mutagenesis of the osmEp region was performed according to Diaz et al. (1991). pCG512 DNA (10 µg) was lyophilized and incubated in 100 µl of 250 mM sodium acetate, pH 4.3, 1 M sodium nitrite for 1 min. After precipitation, aliquots of the treated DNA were mixed with oligonucleotides osmE1 (5′-GGGGATCCGAATTCGGGCTGAAT GGTGG-3′) and osmE2 (5′-GGGAAGCTTAATACCGCCGCTGC-3′) and used for amplification of DNA fragments with Tfl thermostable DNA polymerase (Epicentre Technologies) according to the manufacturer’s protocol. Site-directed mutagenesis of the osmE promoter region was performed on plasmid pCG512 with the USE mutagenesis kit (Pharmacia Biotech), as described by Gordia and Gutierrez (1996). The mutagenic primers used were: osmE8A (5′-GTATTCCAGGATTATCTAACAC-3′), osmE8G (5′-GTATTCCAGGGTTATCTAACAC-3′), osmE13T (5′-CCCAATGTATTTCAGGCTT ATC-3′) and osmE28G (5′-GGGCTTGAAAAGGCGCCCAATG-3′).

Amplification of EcoRI DNA fragments carrying the osmY and fic promoters was performed using the polymerase chain reaction (PCR) with the following couples of primers and matrix DNA: osmY1 (5′-CCCCGAATTCCCAAGCTTCGATATCTACGCATTGAAC-3′), jcd1 (5′-GCGTTCAGCAGTTCTGCC-3′) and plasmid pJCD02 (Colland et al., 1999) for osmYp; fic1 (5′-GGGGAATTCCGCTAAAGTCCTGCCG-3′), fic3 (5′-GGGGAATTCCCAGAGACGGGACTTTTGC-3′) and DNA of phage 624 from Kohara’s collection (Kohara et al., 1987) for ficp.

Run-off transcription assays

Supercoiled plasmid templates ( inline image, pJCD01osmEp12T or pJCD01osmEp28Δ prepared from an overnight culture of a recA1 strain) were used for in vitro transcription assays as described previously (Marschall et al., 1998). The transcription buffer was adjusted to 0.3 M K+ glutamate. Intensities of the bands were quantified on the scanned autoradiography using tina (version 2.07c) software.

Genetic procedures

Standard procedures were used for growth of bacteria and bacteriophages and generalized transduction with phage P1vir (Silhavy et al., 1984; Miller, 1992). EcoRI DNA fragments carrying the wild type or mutated osmEp promoter or the osmY and fic promoters were cloned in the recombination vector pOM41. A 638 bp BamHI–HaeIII DNA fragment carrying the treA promoter (Repoila and Gutierrez, 1991) was cloned in the vector pSB118 cleaved by BamHI and SmaI, and a 689 bp EcoRI DNA fragment carrying treAp was then subcloned in pOM41. Strain pop3125 carries a φ(malP–lacZ) transcriptional fusion in which an intact lac operon is fused to the first gene of the malPQ operon (Débarbouilléet al., 1978). The series of pOM41 derivatives was used to insert the corresponding promoters in front of the φ(malP–lacZ) fusion of pop3125 by homologous recombination, as described previously (Gutierrez and Devedjian, 1991).

Acknowledgements

We thank J.-P. Bouché for help with the computer work. Part of this work was supported by grants from the French Ministère de l’Education Nationale, de la Recherche et de la Technologie (Programme de Recherche Fondamentale en Microbiologie, Maladies Infectieuses et Parasitaires) and the Région Midi-Pyrénées (no. 9609793). We thank the Fondation pour la Recherche Médicale for a fellowship to F.R., and the NATO Science Program for a Collaborative Research Grant (CRG 972150)

Ancillary