The mechanism by which XylR, the toluene-responsive activator of the σ54-dependent Pu and Ps promoters of the Pseudomonas TOL plasmid pWW0, downregulates its own σ70 promoter Pr has been examined. An in vitro transcription system was developed in order to reproduce the repression of Pr observed in cells of P. putida (pWW0) both in the presence and in the absence of the XylR inducer, benzyl alcohol. DNA templates bearing the two σ70-RNA polymerase (RNAP) binding sites of Pr, which overlap the upstream activating sequences (UAS) for XylR in the divergent σ54 promoter Ps, were transcribed in the presence of a constitutively active XylR variant deleted of its N-terminal domain (XylRΔA). The addition of ATP, known to trigger multimerization of the regulator at the UAS, enhanced the repression of Pr by XylR. Furthermore, we observed activation of the divergent σ54 promoter Ps during Pr downregulation by XylRΔA. These results support the notion that activation of XylR by aromatic inducers in vivo triggers a transcriptional switch between Pr and PsSuch a switch is apparently caused by the ATP-dependent multimerization and strong DNA binding of the protein required for activation of the σ54 promoter. This device could reset the level of XylR expression during activation of the σ54Pu and Ps promoters of the TOL plasmid.
In addition to controlling the expression of the σ54 promoters Pu and Ps of the TOL (toluene and xylene biodegradation) plasmid of Pseudomonas putida, the XylR protein also downregulates its transcription by repressing its own σ70-dependent promoter Pr (Inouye et al., 1987; Abril et al., 1989; Gomada et al., 1992; Holtel et al., 1992; Delgado and Ramos, 1994; Ramos et al., 1997) To achieve this effect, the upstream activating sequences (UAS) of the Ps promoter, which is activated by XylR for transcription of the xylS gene, happen to overlap with two −10/–35 hexamers of the divergent promoter Pr (Fig. 2). We have shown previously (Bertoni et al., 1997) that the downregulation caused by XylR on Pr can be separated genetically from activation of the σ54 promoter Ps and that XylR binding to the UAS suffices to cause a degree of repression. However, it is also known (Pérez-Martín and de Lorenzo, 1996a) that interactions of XylR with the UAS do change upon induction: release of the intramolecular repression exerted by the A-domain (Fernández et al., 1995) triggers an ATP-induced multimerization of XylR through the UAS accompanied by changes in the local conformation of the DNA region. This multimerization event may further hinder the entry of σ70-RNAP to the divergent promoter Pr, thereby enhancing its downregulation. On this basis, inducer addition is predicted to have a influence not only on Ps activation but also on Pr repression, ATP-driven multimerization of XylR being the basis of both effects.
In this paper, we present evidence supporting the notion that, besides its performance as a σ54 activator, XylR is also an ATP-responsive repressor of the σ70-RNAP in Pr, not unlike other prokaryotic repressors, such as TyrR (Pittard and Davidson, 1991). Our results suggest that Ps and Pr promoters are submitted to a transcriptional switch mediated by the transition from the non-co-operative, weak binding of XylR to the UAS to the ATP-dependent multimerization and strong DNA binding of the protein to the same region.
Results and discussion
XylR activation represses Pr in vivo
As discussed above, the mechanism by which XylR activates its cognate promoters (Pérez-Martín and de Lorenzo, 1996a) predicts that Pr repression should be enhanced when XylR is activated by the presence of aromatic inducers of the TOL system. However, the degree of Pr repression observed in cells exposed to inducers does vary considerably among authors (Abril et al., 1989; Holtel et al., 1992; Delgado and Ramos, 1994). To clarify this point in a well-defined assay system in vivo, we used a primer extension protocol for monitoring the activity of the Pr promoter in P. putida cells bearing the TOL plasmid pWW0 and exposed to benzyl alcohol. As shown in 3Fig. 3A, under the conditions of the assay, it became apparent that inducer addition resulted in a 90% repression of Pr transcripts compared with their non-induced counterparts. This result provided a preliminary clue that activation of XylR by its aromatic inducers could enhance the repression of its promoter Pr.
To investigate whether the basis of this effect was the ability of XylR to multimerize at the UAS (Fig. 2) under induced conditions, we reproduced all the regulatory elements concerned in Escherichia coli GB1, which carries a chromosomal Pr-lacZ reporter system. Owing to the inherent toxicity of TOL inducers in E. coli (Abril et al., 1989; Bertoni et al., 1997), we did not expose cells to benzyl alcohol as above. Instead, we transformed E. coli GB1 cells with equivalent plasmids encoding either wild-type XylR or a fully constitutive XylR variant called XylRΔA. This variant behaves as XylR in the presence of its aromatic inducers because of the deletion of the intramolecular repressor A domain (Fernández et al., 1995; Pérez-Martín and de Lorenzo, 1996b). As a control, we also used a plasmid encoding a XylRΔA derivative called G268N, which bears a mutation within the ATP-binding site of XylR (GETGV[G268xN]KE: Walker A motif; Walker et al., 1982; Inouye et al., 1988; North et al., 1993). This mutation inactivates the protein because it affects the ability to bind ATP, oligomerize and occupy the UAS co-operatively (Pérez-Martín and de Lorenzo, 1996a,b). On this basis, each of the three XylR variants used (wild-type XylR, XylRΔA and G268N) represent one of the three stages of the activation of the protein by aromatic inducers: before activation (wild-type XylR), activated but before ATP binding (G268N) and activated after ATP binding (XylRΔA).
By using this Pr-lacZ reporter system assembled in E. coli (Fig. 3B), it became apparent that the constitutively active XylR variant capable of ATP binding and oligomerization at the UAS (XylRΔA) had a superior repression ability on Pr than either wild-type XylR or the variant unable to bind ATP (G268N). On this basis, we hypothesized that Pr could be subjected to two levels of repression: on the one hand, an ATP-independent level resulting from non-co-operative interaction of non-activated XylR with the distal and proximal sites of Ps UAS, as described previously (Bertoni et al., 1997); and on the other hand, an ATP-dependent repression level, resulting from the co-operative oligomerization of activated XylR at the UAS (Pérez-Martín and de Lorenzo, 1996a).
Autorepression of Pr by XylR is responsive to ATP
In order to ascertain the basis of the enhanced downregulation exerted by the constitutively active XylRΔA protein, we examined the effect in vitro with isolated components. To this end, XylRΔA was purified by means of a His-tag placed at its N-terminus (Pérez-Martín and de Lorenzo, 1996b) and mixed with σ70-RNAP from E. coli and either supercoiled or linear DNA templates bearing Pr. The concentration of ATP used in the standard NTP mix (400 μM) present in the in vitro transcription buffer is insufficient to cause a significant oligomerization of XylRΔA (Pérez-Martín and de Lorenzo, 1996a). Therefore, we refer to such a condition hereafter as ‘without ATP’. On the contrary, 2 mM ATP has been shown to trigger the multimerization process in XylRΔA at the UAS that precedes transcription activation (Pérez-Martín and de Lorenzo, 1996a). On this basis, the addition of 2 mM ATP to the transcription buffer is referred to as ‘with ATP’.
In the in vitro system used, the ability of increasing XylRΔA concentrations to repress transcription arising from the Pr promoter region was monitored through the analysis of the resulting transcripts generated in the presence or absence of 2 mM ATP (Fig. 4). A preliminary piece of information emerging from the data shown in Fig. 4 is that the two potential transcription initiation sites at the Pr region proposed from the analysis of in vivo transcripts with S1 nuclease or primer extension (Inouye et al., 1987; see above) do have their correspondence in the transcripts observed in vitro. This confirms rigorously that transcription from the Pr region originates in vivo from two tandem promoters instead of the processing of a longer transcript.
The results shown in Fig. 4 indicate that purified XylRΔA represses by itself the synthesis of the T1 and T2 transcripts raised by σ70-RNAP from Pr in a range of concentrations from 60 to 300 nM. Repression occurred to the same extent regardless of whether the promoter is born in a supercoiled (Fig. 4A) or linear (Fig. 4B) template. Non-co-operative binding of XylRΔA to the UAS without ATP therefore sufficed to exert a degree of repression on Pr. Under the assay conditions used, the apparent affinity constant (K′s) of the protein for its target in linear DNA, calculated on the basis of its ability to inhibit transcription, was in the region of 140 nM (Fig. 5). This inhibition accounts for the downregulation of the Pr promoter under non-induced conditions in vivo (Fig. 3; Bertoni et al., 1997). However, the addition of ATP to the in vitro system strengthened Pr downregulation: while full Pr repression by XylRΔA without ATP required the addition of > 300 nM protein to the assay, the same occurred in the presence of ATP with < 90 nM protein, thus bringing down K′s values by 3.5-fold (40 nM, Fig. 5).
A second aspect of the experiment shown in 4Fig. 4B is the change in the relative ratio of T1 and T2 transcripts in the absence or in the presence of ATP. In principle, the difference in T1 vs. T2 may simply reflect the relative affinity of the σ70-RNAP for each of the sites (Fig. 2), which, in any case, only vaguely resemble the consensus promoter sequences of E. coli (Collado-Vives et al., 1991) or Pseudomonas (Deretic et al., 1987; Soldati et al., 1987). In the absence of ATP, the ratio T1:T2 is about 2.5, while the addition of the nucleoside consistently varies the ratio to > 4.0. Thus, T2 becomes far more sensitive to ATP than T1. As T2 transcript originates in a −35/–10 sequence that overlaps the Ps-distal UAS for XylR (Fig. 2), which is only weakly bound by the protein in the absence of ATP (Pérez-Martín and de Lorenzo, 1996c), the rapid inhibition of T2 in the presence of ATP suggests a stronger binding to such a distal binding site.
Taken together, the results from the in vitro assays presented in Fig. 4 indicate that the formation of an ATP-dependent multimer at the Pr promoter accounts for the enhanced downregulation of the promoter in vivo in conditions of XylR activation. This notion is supported further by the results on the behaviour of G268N mutant in vitro (Fig. 4C). As mentioned above, this variant binds ATP to a much lesser extent and is unable to multimerize (Pérez-Martín and de Lorenzo, 1996a). If enhanced repression of Pr is linked to the multimerization of XylR at the target sequences, G268N protein should then repress less than the wild type. According to this prediction, 4Fig. 4C shows that purified G268N protein downregulated Pr activity, but the effect of ATP addition was remarkably weaker than in the samples with XylRΔA.
XylR oligomerization at the UAS switches Pr off and turns Ps on
The picture that emerges from the results above is that induction of the XylR protein into a transcriptionally competent form (i.e. release of intramolecular repression, ATP binding, oligomerization at the UAS and σ54-RNAP activation) causes a transcriptional switch from the σ70 promoter Pr to the σ54 promoter Ps. To test such a switch, we ran a series of in vitro transcription assays, in which we monitored simultaneously the activity of both promoters on the basis of the different sizes of the transcripts initiated at each of them at the Ps/Pr region. To this end, we took advantage of the fact that the RNAP used (see Experimental procedures) was only approximately 50% saturated with σ70. Therefore, by adding 50 nM purified σ54 protein (i.e. a 3:1 ratio σ70:σ54), the two holenzyme forms became available to the system. Under these conditions, we added increasing amounts of XylRΔA so that Ps could become activated. The results of such an experiment are shown in Fig. 6, which indicated that Ps activation was concomitant with Pr repression. The addition of increasing amounts of XylRΔA to the σ70/σ54-containing system switched transcription progressively from Pr to Ps. As ATP-dependent oligomerization is a prerequisite for XylR activation (Pérez-Martín and de Lorenzo, 1996a), this result suggested that the same XylR/UAS complex at the Pr/Ps region accounts for both the activation of the σ54 promoter and the repression of the divergent σ70 system.
XylR is a TyrR-like repressor of Pr
A number of transcriptional regulators, CAP and IHF proteins being the best examples (Goosen and van de Putten, 1995; Kolb et al., 1993; Valentin-Hansen et al., 1996), act as either activators or repressors by virtue of the location of their target sequences with respect to the corresponding promoters. These dual effects frequently rely on straight competition between proteins for binding sites or architectural effects (Pérez-Martín et al., 1994; Pérez-Martín and de Lorenzo, 1997). In some cases, however, the ability of a regulator to both activate or repress transcription requires the presence of different activities in the same protein. We have shown previously that XylR can downregulate its own transcription independently of Ps activation, so that the two functions can be separated genetically (Bertoni et al., 1997). In this work, we have addressed whether the series of events that occur at the UAS after XylR activation by xylene but before the engagement of the σ54-RNAP (i.e. release of intramolecular repression caused by the A domain, ATP binding and multimerization; Pérez-Martín and de Lorenzo, 1996a) have an influence on XylR repressor activity on the Pr promoter. Our results indicate that ATP binding and the ensuing oligomerization of XylR at the UAS required for Ps activation enhance repression of Pr activity by extended occupation of the region by a XylR oligomer. The ability of XylR to shift from weakly downregulating to strongly repressing Pr in response to ATP sheds some light on the fine tuning of the TOL system, in particular the resetting of XylR biosynthesis after induction. From the data obtained in this work, it becomes clear that, as soon as the protein is activated and the UAS strongly bound by the regulator, XylR expression is minimized, thus limiting the period of time in which the Ps and Pu promoters of the TOL plasmid last in an activated state.
The ATP-dependent repressor activity of XylR may not be alien to the fact that the central domain of the proteins that act in concert with σ54 is highly homologous to the central module of TyrR, the repressor/activator of various operons for the biosynthesis of aromatic amino acids in E. coli (Pittard and Davidson, 1991). In the presence of tyrosine and ATP, TyrR dimers associate to form a hexamer that represses transcription from cognate σ70 promoters through the co-operative binding of TyrR to multiple TyrR boxes (Wilson et al., 1994). Similarly to XylR, mutations analogous to G268N that alter the Walker A motif of the ATP-binding site (Pittard, 1996; Song and Jensen, 1996) or changes in residues involved in hexamerization (Kwok et al., 1995) abolish the repressing ability of TyrR. Furthermore, the changes in the patterns of interaction of TyrR with its target sequences in the presence of ATP resemble those observed with XylR at the UAS (Pérez-Martín and de Lorenzo, 1996a). On the basis of these similar characteristics of TyrR and XylR, it seems, therefore, that the same protein module that oligomerizes in response to ATP has been recruited during evolution for either repression of some σ70 promoters or activation of other σ54 promoters containing properly positioned target sequences. This notion is further sustained by the observation that the σ54-dependent regulator PhhR protein, which controls expression of the phhABC operon for the metabolism of aromatic amino acids in P. aeruginosa (Song and Jensen, 1996), was able to replace E. coli TyrR for the repression of the aroF–tyrA operon of E. coli, but not for the activation of the mtr promoter, which is positively controlled by the TyrR protein as well.
Strains, plasmids and general procedures
P. putida KT2442 (pWW0) has been described previously (Marqués et al., 1994). Escherichia coli GB1 is a reporter strain that carries a chromosomal Pr–lacZ fusion (Bertoni et al., 1997) driven by a 0.3 kb BamHI fragment spanning positions + 83 of Ps to + 26 of the further downstream promoter (P2) of the two −10/−35 hexamers present in Pr (Fig. 2). Recombinant DNA manipulations were carried out according to published protocols (Sambrook et al., 1989). Plasmids pFX, pFX1 and pFX2 drive the expression of wild-type XylR, XylRΔA and XylRΔA-G268N respectively. The three proteins were engineered bearing a polyanionic tag at their C-terminus (i.e. a modified FLAG epitope DYKDEGGK; Kodak) in order to increase their stability in vivo (Keiler et al., 1996). To this end, each of the expression plasmids was constructed by assembling the sequence of the previously described xylR alleles (de Lorenzo et al., 1991; Pérez-Martín and de Lorenzo, 1996b) in the Ptrc/lacIq expression vector pTrc99A (Amann et al., 1988) as EcoRI–BamHI fragments, along with a synthetic FLAG linker. Promoter activity was monitored in all cases by assaying the accumulation of β-galactosidase raised by lacZ fusions in cells permeabilized with chloroform and SDS as described by Miller (1972) under the conditions specified. His-tagged proteins XylRΔA and XylRΔA-G268N were purified from E. coli DH5α carrying the expression plasmids pQXΔ and pQXΔ-G268N as described previously (Pérez-Martín and de Lorenzo, 1996b). Purified σ54 factor was a kind gift from B. Magasanik and M. Carmona. E. coliσ70-RNAP (50% saturation with σ70) was purchased from Amersham.
Primer extension analysis
Detection of specific transcripts T1 and T2, originated in vivo from the Pr promoter in P. putida (pWW0), was made using the protocol described in Marqués et al. (1994). Briefly, this strain was diluted in minimal medium (Sambrook et al., 1989) with 0.2% glucose as the only carbon source and incubated at 30°C to exponential phase. The culture had 5 mM of the XylR inducer benzyl alcohol (used instead of toluene because of its superior solubility in water) added, and the incubation continued for 4 h, until the cells reached an OD600 = 1.5. mRNA (10 μg) extracted from the cells of each culture (non-induced or induced) were then extended with a primer, 5′-ACGGATCTGGCTGCTAAGGTCTTGC-3′, labelled with 32P at its 5′ end in the presence of reverse transcriptase. This originates two cDNAs of 208 nucleotides (arising from P1) and 180 nucleotides (arising from P2). The labelled products were resolved and quantified in urea–polyacrylamyde gels as described previously (Marqués et al., 1994).
The linear template for run-off transcription assays was obtained as a 427 bp SmaI–PvuII restriction fragment from pEZ2b. This plasmid is a pUC18 derivative inserted with a segment from the TOL plasmid spanning positions + 83 of the Ps promoter to + 26 of the more downstream −10/−35 hexamers (P2) of the Pr region (Fig. 2). The entire ccc DNA form of plasmid pEZ2b was used in the assays in which a supercoiled template was required. For the reactions, the DNA template at 5 nM was premixed on ice in 20 mM Tris-HCl (pH 7.5), 5 mM MgCl2, 50 mM KCl and 50 μg ml−1 BSA in a final volume of 25 μl, with various amounts of purified XylRΔA or its variant G268N and, where indicated, with 2 mM ATP as well. The mixtures were incubated for 5 min at 30°C before adding σ70-RNAP at a final concentration of 150 nM, and the incubation at 30°C was continued for a further 5 min. In the case of run-off assays, transcription reactions were then initiated by the addition of [α-32P]-UTP (30 μCi; 6000 Ci mmol−1), unlabelled UTP at 50 μM and ATP, CTP and GTP each at 400 μM. When required, purified sigma factor (σ54) of E. coli was also added to the transcription mixtures at a final concentration of 50 nM. After incubation for 10 min, the reactions were stopped by the addition of EDTA to 40 mM. The run-off reactions were then loaded directly on 7 M urea/7% polyacrylamide sequencing gels. When the template of the reaction was supercoiled plasmid pEZ2b, transcription was followed through primer extension of the mRNAs produced in vitro. To this end, the σ54-RNAP, XylRΔA and DNA mixture as above was added with 400 μM cold NTPs, and the reactions were allowed to run for 10 min. Transcripts were then extracted, primed with the oligonucleotide 5′-CGCCAGGGTTTTCCCAGTCACGAC-3′ (which primes the vector sequence counterclockwise of the Pr insert) and extended with reverse transcriptase in the presence of [α-32P]-dATP (Rojo et al., 1993). As before, the labelled cDNAs were loaded in sequence gels, dried and exposed to autoradiographic films for 16–48 h. Transcription was quantified using a Molecular Dynamics scanner system.
This work was funded by the ENV4-CT95-0141 (ENVIRONMENT) Contract of the EU and by Grant BIO95-0788 of the Comisión Interministerial de Ciencia y Tecnología. G.B. was the recipient of a Fellowship of the Spanish Ministry of Education and Science for foreign PhD visitors. The authors are grateful to J. Pérez-Martín, J. L. Ramos and I. Cases for inspiring discussions and to M. Carmona and B. Magasanik (MIT, Cambridge, MA, USA) for various materials used in this work.