The Pm promoter of the TOL plasmid of Pseudomonas putida is expressed at high level along the growth curve. This transcription is dependent on the positive regulator XylS activated by 3-methylbenzoate. The sigma factor σ38 is required for expression in early stationary phase and thereafter. To test whether σ70 was involved in Pm transcription in exponential phase, we have followed mRNA synthesis in a rpoD thermosensitive strain. No difference in Pm transcription was found between the wild type and the thermosensitive strain at the restrictive temperature of 42°C, indicating that transcription was independent of the sigma factor σ70. However, basal levels of mRNA expression from Pm in this strain in exponential phase were more than twofold higher at 42°C, suggesting involvement of σ32 in Pm transcription. In a rpoH background, no expression of Pm took place in the exponential phase, whereas it increased during stationary phase, and in a rpoH rpoS double mutant no activity from the Pm promoter was detected along the growth curve. We have shown that the increase in the amount of σ32 factor necessary for transcription in exponential phase is provided through induction of the heat shock response by the presence of the effector 3-methylbenzoate, which is also required for activation of the positive regulator XylS. We conclude that activation of Pm transcription is achieved through a switch between two stress-responsive factors, σ32 in exponential phase and σ38 in stationary phase. In both cases, transcription is dependent on the activator XylS and presents the same transcription start point.
Pseudomonas putida bearing the TOL plasmid pWW0 is able to grow on 3-methylbenzoate and related compounds as the sole carbon source. The enzymatic pathway for conversion of this aromatic compound to Krebs cycle intermediates, called the meta-cleavage pathway, is coded in an operon of ≈ 11 kb that is under the control of a single promoter, Pm (Inouye et al., 1981; Harayama et al., 1984; Marqués et al., 1993; Ramos et al., 1997). Transcription of the meta-cleavage operon from Pm requires the presence of the positive regulator XylS that belongs to the so-called AraC/XylS family of regulators (for a review, see Gallegos et al., 1997). The XylS protein is constitutively expressed at low levels in an inactive transcriptional form, but becomes active from a transcriptional point of view in the presence of pathway substrates such as 3-methylbenzoate (Ramos et al., 1987; Gallegos et al., 1996a).
A DNA region extending to about 70 bp upstream from its transcription initiation point is sufficient for the XylS-dependent expression from Pm (Ramos et al., 1987; Spooner et al., 1987; Kessler et al., 1993; Marqués et al., 1994; Gallegos et al., 1996b). The XylS protein recognizes in Pm two direct repeats (TGCAN6GGNTA) located between −70 and −35 with respect to the main transcription initiation point (Kessler et al., 1993; González-Pérez et al., 1999) (Fig. 1). The Pm promoter does not display a DNA sequence similar to that reported in other promoters in the −10 and −35 regions (Fig. 1). Part of this divergence from consensus sequences in the −35 region could be explained by the fact that the Pm promoter requires binding of the XylS activator to promote transcription. Divergence from consensus sequences in the −10 region could be due to the fact that expression from Pm can be mediated by RNA polymerase with different sigma factors. Indeed, expression of Pm in the stationary phase is dependent on the alternative sigma factor σ38. High levels of Pm transcription in an E. coliσ38 minus background were only obtained in the early logarithmic phase, in contrast with the situation in a σ38 proficient strain in which high levels of transcription were obtained in both exponential and stationary phases (Marqués et al., 1995). In both cases, this activity remained dependent on the positive regulator XylS activated by a benzoate effector, and exhibited the same transcription initiation point. In this regard, Pm is one of the few σ38-dependent promoters known to strictly require a specific activator protein, XylS, to mediate transcription from its target promoter. This led us to suggest that the high levels of expression of Pm were maintained throughout the growth curve through a switch in sigma factors (Marqués et al., 1995).
A number of promoters have been shown to work in vitro with σ70 and/or σ38 (Tanaka et al., 1993), although this has only been observed in vivo for a few promoters, i.e. those directing the transcription of the dps, csiE and osmY genes among others (Altuvia et al., 1994; Hengge-Aronis, 1996; Bouvier et al., 1998). Sigma-70 is the major sigma factor present in the cell both in exponential growth phase and stationary growth phase, whereas σ38 seems to be expressed at low levels during exponential growth and increases progressively thereafter (Jishage et al., 1996). It was not surprising, therefore, that the requirement of σ38 for Pm transcription previously described was restricted to the late logarithmic and stationary growth phases (Marqués et al., 1995). However, because Pm was transcribed in the early exponential phase by an RNA polymerase different from E-σ38, it remained unclear which sigma factor was used by RNA polymerase to transcribe Pm in this phase. In this study, we intended to elucidate the participation of different RNA polymerases in transcription of Pm along the growth curve, and the signals that determine their sequential action in Pm.
Transcription from Pm in the exponential phase is independent of σ70 RNA polymerase
We previously suggested that high levels of expression of Pm were maintained through a switch between σ70 and σ38 RNA polymerases, σ70 functioning during the exponential phase and σ38 when the stationary phase enters and thereafter (Marqués et al., 1995). To directly test whether σ70 was involved in transcription from Pm in exponential phase, we analysed the expression of Pm in genetic backgrounds proficient and deficient in σ70. Because a null mutant of σ70 is lethal, we used the thermosensitive (ts) E. coli strain UQ285 (Harris et al., 1978; Nakamura et al., 1983). This strain bears a short deletion in the rpoD gene that makes the mutant sigma factor unstable at 42°C (Table 1), so that mRNA synthesis from σ70-dependent promoters ceases 10–20 min after an upshift to 42°C (Harris et al., 1978). Strain UQ285 was transformed with pNM185, a pKT231 derivative bearing the gene for the positive regulator XylS, and the 0.4 kb PstI fragment of TOL plasmid containing the Pm promoter (Mermod et al., 1986). A freshly diluted culture of this strain was supplemented or not with 1 mM 3-methylbenzoate and grown at 30°C. After 1.5 h incubation, when cells were in the exponential phase, a fraction of the main culture was taken and split into two aliquots, one of which was transferred to 42°C whereas the other was maintained at 30°C as a control. Samples from all subcultures were taken for mRNA analysis 30 and 60 min after the temperature upshift. Figures 2 and 3 show the expression of the Pm promoter under each condition. Surprisingly, in the induced culture, after the temperature upshift in the exponential phase, expression from Pm continued to be high (Fig. 2), and similar to the levels obtained in the isogenic wild-type rpoD strain P90 A5c (not shown). The slight decrease in Pm activity observed in both strains after 60 min is probably due to instability of the XylS protein at 42°C (Ramos et al., 1988).
Table 1. . Strains and plasmids used in this study. Apr, Kmr, Cmr refer to ampicillin, kanamycin and chloramphenicol resistance respectively (Harayama et al. (1989)).
Expression from the Pm promoter under non-induced conditions, i.e. in the absence of an effector, is only detectable with exposure times much longer than that required for induced cultures. Figure 3 shows the basal levels of mRNA obtained when cells were transferred to 42°C. It is interesting to note that these levels were at least twice as high as those found in cells kept at 30°C, thus suggesting a possible induction of basal Pm levels as a consequence of temperature upshift.
As a positive control, we determined expression of the xylR gene from the constitutive σ70-dependent tandem Pr1 and Pr2 promoters in pAH120 (Holtel et al., 1992). As expected, in the UQ285 host background bearing pAH120, expression from the Pr1 and Pr2 promoter took place at 30°C but did not occur at 42°C (data not shown).
σ32 RNA polymerase is required for transcription from Pm in the exponential phase
The observation that basal expression of Pm in the exponential phase was increased at 42°C (Fig. 3) prompted us to investigate the pattern of Pm expression in a genetic background lacking σ32. KY1429 is an E. coli MC4100 derivative that bears a mutation in rpoH gene that makes it unable to grow at 42°C (Tobe et al., 1984; Zhou et al., 1988). Strains MC4100 (wild type) and its two isogenic derivatives KY1429 (rpoH ) and the rpoS mutant RH90 (Lange and Hengge-Aronis, 1991) were transformed with plasmids pJLR107, which bears a Pm::′lacZ fusion, and pMT500 bearing the xylS regulator gene. β-Galactosidase activity in response to the addition of 3-methylbenzoate was followed along the growth curve (Fig. 4). Activity increased rapidly in the wild-type strain and remained high throughout the curve. In the rpoH strain, no expression was observed in the early exponential phase, although activity did occur in later growth stages. As found by Marqués et al. (1995), in the rpoS mutant RH90, activity was high in the exponential phase and decreased thereafter. In the absence of 3-methylbenzoate, no activity was detected in any of the strains (not shown).
Because both the level and activity of minor sigma factors are influenced by various factors, including growth phase and the genetic host background, we decided to run a parallel series of controls with well-characterized promoters fused to ′lacZ that were under the control of one of the three sigma subunits assayed in this study: the constitutive σ70-dependent tandem Pr1 and Pr2 promoters that drive the expression of the xylR gene (Holtel et al., 1992); PrpoDhs, the well-established σ32-dependent promoter of rpoD (Yano et al., 1987); and the Pfic promoter, the well-established σ38-dependent promoter of the fic gene (Miksch and Dobrowolski, 1995). High levels of expression of Pr1 and Pr2 were obtained in the wild-type MC4100 and its mutant derivatives RH90 and KY1429, regardless of the growth phase (Table 2); in contrast, expression from the Pfic promoter was high only in the stationary phase in the wild-type and the KY1429 background, and negligible in the σ38-deficient background provided by RH90 (Table 2). The lower level of activity in KY1429 (σ32 deficient) with respect to wild-type strain is probably due to the lack of DnkA in the RpoH-deficient background (Muffler et al., 1997). Expression from PrpoDhs increased when the wild type or the RH90 strains were incubated at 42°C, but not when the assay was run with the KY1429 strain. Given that the Pm promoter behaves like the σ32-dependent PrpoDhs regarding incubation temperature, and like the σ38-dependent Pfic regarding growth phase, we suggest that expression from Pm can be achieved through a switch between σ32 and σ38 at some point on the growth curve.
Table 2. . β-Galactosidase activity expressed from well-characterized promoters in different host genetic backgrounds used to study expression of the promoter for the meta operon of TOL plasmid pWW0. The wild-type (MC4100), rpoS-deficient (RH90) and rpoH-deficient (KY1429) strains were transformed with pAH120 (Pr1, Pr2::′lacZ), pGM1115 (Pfic::′lacZ) and lysogenized with λ phage bearing a PrpoDhs::′lacZ fusion. Strains bearing pAH120 or pGM115 were grown on LB with appropriate antibiotics, and when the mid-log phase (A660: 0.5–0.8) or the early stationary phase (A660: 2.5–3.0) was reached, β-galactosidase activity was determined as described previously (Gallegos et al., 1996a). The strain bearing PrpoDhs was grown on LB medium until mid-log phase, then the culture was divided into two halves. One half was kept at 30°C and the other was transferred to 42°C. β-Galactosidase activity was determined 105 min later. β-Galactosidase activity is given in Miller units. Data are the average of three independent determinations with a standard deviation below 10% of the given values.
To confirm this hypothesis, a series of assays were carried out with a double mutant rpoH rpoS, constructed in our lab. This mutant was expected to exhibit no activity from Pm. The mutant was generated through P1 phage transduction of KY1429 with a lysate obtained from the rpoS mutant strain MC1061-14 (Arnqvist et al., 1994). This double-mutant strain was transformed with plasmids pJLR107 and pMT500, and β-galactosidase activity was recorded as above. Figure 4 shows that in the double mutant, no significant activity was produced at any time during growth. This supported the hypothesis that expression of Pm was dependent on σ32 in the exponential phase and on σ38 in the stationary phase.
The presence of the inducer 3-methylbenzoate triggers heat shock response
E. coli responds to heat shock by increasing the level of σ32 in the cell by 15- to 20-fold. This is a consequence of both stabilization of the σ32 protein and increased translation of its mRNA. Subsequently, this increase in σ32 concentration directs RNA polymerase to the so-called heat shock (hs) promoters, which are then transcribed at a high rate. The heat shock response can be mediated by other stresses, such as viral infection, ethanol, methylating agents, hydrogen peroxide or toluene (Neidhardt and VanBogelen 1987; Van Dyk et al., 1995).
The above results suggested that Pm expression in the exponential phase is mediated by RNA polymerase with σ32 protein. Like other aromatic compounds, 3-methylbenzoate may be able to trigger the heat shock response. To test this hypothesis, we studied whether the addition of 3-methylbenzoate to a cell culture was able to induce activity of a heat shock promoter. We used a λpF13–(PrpoDhs–lacZ) lysogen of MC4100 that bears on the chromosome a fusion of lacZ to the σ32-dependent promoter of rpoD (Yano et al., 1987).
A freshly diluted culture of the λpF13–(PrpoDhs–lacZ) lysogen was supplemented or not with increasing concentrations of 3-methylbenzoate, and β-galactosidase activity was determined 150 min after addition. As a positive control, a culture without 3-methylbenzoate was incubated at 42°C. The results are shown in Fig. 5. The low levels of activity from PrpoDhs (10 ± 3 U) at 30°C were similar to the levels originally reported by Yano et al. (1987). The strain behaved as expected at 42°C, showing the reported twofold increase in β-galactosidase activity. Figure 5 clearly shows that the presence of a wide range of concentrations of 3-methylbenzoate in the culture medium induced the reporter promoter to a level similar to the temperature upshift. We concluded that as for other toxic compounds, the addition of the TOL inducer 3-methylbenzoate triggers the heat shock response in E. coli, and may therefore increase the concentration of σ32 in the cell to the level required for Pm expression.
In the absence of heat shock response, Pm is not expressed in exponential phase
From the results presented above, we suggest that: (1) transcription from Pm in the exponential phase requires σ32; and (2) under normal conditions, this factor is provided through the induction of the heat shock response by the addition of 3-methylbenzoate. We can, therefore, predict that in a system in which induction of Pm does not require the addition of 3-methylbenzoate, no activation (or only a low level of activity) should take place in the exponential phase.
XylSP37L is a single mutant of XylS that promotes expression from Pm in the absence of an effector (Michán et al., 1992). According to the above prediction, in E. coli cells bearing a Pm::′lacZ fusion and this allele of xylS, low levels of expression should take place in the exponential phase without 3-methylbenzoate. Figure 6 shows expression from Pm in MC4100 bearing pJLR107 and pERD138, a pKT231 derivative bearing the xylS mutant allele. As expected, no significant expression from Pm took place during the first 2 h in the absence of effector, whereas, in the presence of 3-methylbenzoate, the pattern of expression resembled that obtained with the wild-type XylS. In the stationary phase, expression in the absence and in the presence of effector was similar. This supports our model in which Pm expression throughout the growth curve does require both σ32 and σ38.
The pattern of expression from Pm in Pseudomonas putida and E. coli has been found to be almost identical (for a review, see Ramos et al., 1997). It was, therefore, expected that in the wild-type P. putida host background, expression from Pm with XylSP37L would show a pattern of expression similar to that seen in E. coli. To test this, P. putida was transformed with pERD100 (Pm::′lacZ) and pLZ4 (bearing xylS37 which encodes XylSP37L), and cells were grown in the absence and in the presence of 3-methylbenzoate. The results were similar to those obtained with E. coli (data not shown).
Our previous studies showed that the expression of Pm involved RNA polymerase with different sigma factors depending on the growth phase, with σ38 involved in transcription in the stationary phase of growth (Marqués et al., 1995). In this work, we determined which sigma factor was used during the exponential phase. The XylS-dependent high expression of Pm seems to be maintained throughout the growth curve by a switch between σ32- and σ38-dependent RNA polymerases. Experiments that analysed the mRNA produced from Pm in the thermosensitive rpoD mutant strain UQ285 at 42°C excluded the involvement of σ70 in transcription from Pm in the exponential phase. In this series of assays, we found that the basal level of activity of the promoter increased significantly at 42°C; we therefore analysed the role σ32 might play in XylS-dependent Pm transcription by studying transcription in the rpoH mutant strain KY1429. No significant transcription from Pm was found in the early exponential phase, although transcription increased thereafter. This clearly established the requirement of σ32 for XylS-dependent Pm transcription in early phases of growth, confirmed that neither σ38 nor σ70 was able to transcribe Pm in the early exponential phase, and confirmed that in the late exponential and stationary phases σ38 was the main sigma subunit involved in expression from Pm. Because the transcription initiation point of Pm was the same throughout the growth curve in the wild-type strain MC4100, in the early exponential phase in the rpoS mutant RH90 (Marqués et al., 1995) and in the stationary phase in the KY1429 mutant, we suggest that the different RNA polymerases use exactly the same promoter. From a mechanistic point of view, this is not unexpected because in all cases the different polymerases must interact with the regulatory protein XylS bound upstream between −41 and −70 with respect to the transcription initiation point (Kaldalu et al., 1996; Kessler et al., 1993; González-Pérez et al., 1999).
Sigma-32 is the factor ultimately responsible for the cellular response to exposure to several stressing agents, such as high temperature, chelating agents, viral infection. Expression of the σ32 regulon is dependent on the amount of active σ32 protein present in the cell. A generally accepted hypothesis postulates that the heat shock response is triggered by an increase in the cellular amount of unfolded protein (Gross, 1996). In this model, any factor that causes protein denaturation would trigger the so-called heat shock response. We have demonstrated (Fig. 5) that the XylS effector 3-methylbenzoate (Ramos et al., 1986) was able to trigger this response. We now note a second function for the effector: it is not only required to activate the regulator, but also to trigger the so-called heat shock response that stabilizes σ32 and provides the appropriate RNA polymerase for transcription from Pm during the exponential growth phase.
The results with KY1429 show that the sigma factor used in the exponential phase is not exchangeable, i.e. σ38 cannot be used in the early exponential phase (no activity in KY1429). Levels of σ38 are lower in the exponential phase than in the stationary phase. In contrast, it has been suggested that transcription by RNA polymerase with σ38in vitro is enhanced when the process uses templates with low superhelical density, an observation consistent with the decrease in DNA superhelical density in stationary phase of growth (Kusano et al., 1996). This may be the reason why σ38 cannot replace σ32 in the exponential phase, as shown by our results with strain KY1429 (Fig. 4).
An interesting feature of the heat shock response is that it is transient: the response starts with a fast burst of transcription of heat shock genes, and an adaptation phase ensues during which the level of transcription decreases to a lower steady state. In the rpoS mutant RH90, levels of Pm activity are maximal in the exponential phase, decreasing thereafter because of the absence of σ38. However, a certain level of activity is maintained. Our suggestion is that this level of activity may be maintained through the action of σ32, although we cannot rule out that a different sigma factor may be used.
Our results indicated that σ32 and σ38 are markedly preferred in the early exponential and late exponential/stationary phase respectively. However, in the corresponding isogenic deficient backgrounds, there was always residual activity of Pm, even in the absence of the effector 3-methylbenzoate (2, 3Figs 2 and 3; Marqués et al., 1995). We therefore cannot rule out a certain degree of ‘cross-readthrough’ of the Pm promoter, i.e. low levels of Pm transcription with other sigma factors, such as σ70.
Figure 1 shows the sequence of Pm, aligned with the proposed consensus sequences for the binding of several sigma factors. Almost no sequence homology was found with the σ70 consensus, which supports our finding that σ70 does not participate in Pm expression. In the −10 region, we found a reasonable homology to the consensus sequences proposed for both σ32 and σ38. However, no homology was found at the appropriate distance in the −35 region. In activatable σ70-dependent promoters, it has been proposed that interaction of the RNA polymerase with the −35 region can be replaced by protein–protein contact between RNA polymerase and the corresponding activator. In such cases, the −35 sequence diverges considerably from the consensus expected for a σ70-dependent promoter (Collado-Vides et al., 1991). The XylS regulator belongs to the AraC/XylS family of transcriptional regulators. Most members of this family are transcriptional activators that bind adjacent to or overlapping the −35 region of the promoter they recognize (Gallegos et al., 1997). Therefore, the lack of any homology at this site in Pm can be explained by the requirement for XylS binding near this region for activation. In this connection, Hiratsu et al. (1995) found that specificity of σ38 binding was conferred by the −10 region, which was essential for the σ38-dependent transcription of the fic promoter. These authors suggested that recognition of a −35 region was not essential for σ38 promoter recognition when a functional −10 region was present.
In recent years, it has become clear that bacteria have evolved adaptive mechanisms to survive various environmental stresses. P. putida, the natural host of the Pm promoter, is a soil bacterium subjected to frequent changes in growth conditions, probably including stress periods and starvation cycles. It is conceivable that the use of 3-methylbenzoate as a carbon source is one of these adaptive mechanisms. It is becoming more and more evident that different pathways may be necessary to specifically protect the extracytoplasmic compartments of the cell from putative injuring agents. Activation of the Pm promoter through two different stress-responsive sigma factors is a good example of adaptation of a plasmid-borne dispensable metabolic function that becomes necessary under certain conditions of stress.
Bacterial strains, phages, plasmids and growth conditions
The bacterial strains and phages used in this work are listed in Table 1. Unless otherwise stated, bacteria were grown at 30°C in Luria-Bertani (LB) broth supplemented when required with 100 μg ml−1 ampicillin, 30 μg ml−1 chloramphenicol, 25 μg ml−1 kanamycin and 10 μg ml−1 tetracycline. Growth was determined turbidometrically at 660 nm.
DNA transformation was carried out according to standard procedures (Ausubel et al., 1991). The 5′ mRNA start of the transcript originated from Pm promoter was determined according to Marqués et al. (1993). The oligonucleotide 5′-GATGTGCTGCAAGGCGATTAAGTTG-3′ was 5′ end-labelled with [γ-32P]-ATP and polynucleotide kinase, and annealed to 20 μg of total RNA prepared from the different E. coli strains grown under different conditions. cDNA was synthesized by using AMV reverse transcriptase as described previously (Marqués et al., 1993). The products of reverse transcription were analysed in urea–polyacrilamide sequencing gels. Gels were exposed, during the time required, to Amersham RPN-8 films for autoradiography.
E. coli strains were grown overnight on LB medium. Cultures were diluted 100-fold in the same medium supplemented or not with 3-methylbenzoate, and β-galactosidase activity was determined at the indicated times in permeabilized whole cells, as described previously (Ramos et al., 1986). β-Galactosidase is expressed in Miller units.
*Present address: Department of Genetics, Faculty of Biology, University of Seville, Seville, Spain
**Present address: Imperial College, Biomedical Science Building, Imperial College Road, London SW7 2AZ, UK
We thank T. Yura and R. Calendar for strains and advice. This project was supported by grants from the Comisión Interministerial de Ciencia y Tecnología BIO-97–0641.