Divergent transcription of a regulatory gene and a cognate promoter under its control is a common theme in bacterial regulatory circuits. This genetic organization is found for the dmpR gene that encodes the substrate-responsive specific regulator of the σ54-dependent Po promoter, which controls (methyl)phenol catabolism. Here we identify the Pr promoter of dmpR as a σ70-dependent promoter that is regulated by a novel mechanism in which σ54-RNA polymerase occupancy of the non-overlapping σ54-Po promoter stimulates σ70-Pr output. In addition, we show that DmpR stimulates its own production through Po activity both in vivo and in vitro. Hence, the demonstrated regulatory circuit reveals a novel role for σ54-RNA polymerase, namely regulation of a σ70-dependent promoter, and a new mechanism that places a single promoter under dual control of two alternative forms of RNA polymerase. We present a model in which guanosine tetra-phosphate plays a major role in the interplay between σ54- and σ70-dependent transcription to ensure metabolic integration to couple σ70-Pr output to both low-energy conditions and the presence of substrate.
Bacterial adaptation frequently incorporates reprogramming of the global transcriptional capacity through changes in the relative composition of different forms of holoenzyme RNA polymerase (σ-RNAP). The alternative σ-subunits associated with the catalytic core enzyme (α2ββ'ω) direct the holoenzymes to distinct promoter classes (Gruber and Gross, 2003). All bacteria have a primary σ-factor, designated σ70 in Escherichia coli and Pseudomonas putida, and a variable number of alternative σ-factors for adaptive responses to prevailing conditions (Martinez-Bueno et al., 2002). Most alternative σ-factors show sequence and functional similarity to σ70 and form holoenzymes that can spontaneously initiate transcription. The structurally distinct rpoN-encoded σ54-subunit is an exception. It imposes constraints on closed-to-open complex formation so that transcriptional initiation by σ54-RNAP is strictly dependent on activators that use ATP catalysis to remodel DNA–RNAP interactions (reviewed in Wigneshweraraj et al., 2005).
The σ54-subunit programmes RNAP to recognize −24, −12 promoters (core consensus TGGCACG N4 TTGC; Barrios et al., 1999). One such promoter is the σ54-Po promoter of the dmp-operon of pVI150, which encodes the enzymes for (methyl)phenol catabolism (reviewed by Shingler, 2004). Transcription of the dmp-operon, and thereby the ability of P. putida strains harbouring pVI150 to grow at the expense of phenolics is dependent on the divergently transcribed regulatory dmpR gene product. The activity of DmpR, in turn, is controlled by binding of dmp-pathway substrates (or structural analogues), which allow DmpR to adopt its active multimeric form (O'Neill et al., 1998; 2001; Wikström et al., 2001). Interaction between Po-bound σ54-RNAP and DmpR bound to its distally located binding sites (upstream activation sequences, UASs) is then aided by IHF (integration host factor) through its DNA-bending properties (Sze et al., 2001).
Regulatory systems of catabolic pathways for aromatic compounds are generally submissive to global regulatory input resulting in suppression of transcription of the catabolic genes when preferred carbon sources are present (reviewed in Shingler, 2003; Cases and de Lorenzo, 2005). For the DmpR/Po dmp-system, the global alarmone guanosine tetra-phosphate (ppGpp) is primarily responsible for this level of regulation. Lack of ppGpp severely reduces σ54-dependent Po activity in both E. coli and P. putida without altering the constant levels of σ54 in the cell, while artificial synthesis of ppGpp allows σ54-dependent transcription under normally non-permissive growth conditions (Sze and Shingler, 1999; Sze et al., 2002). The alarmone ppGpp heralds nutrient and physicochemical stress and is rapidly synthesized under such conditions through the activities of the RelA synthetase and the bifunctional SpoT synthetase/hydrolase (reviewed in Cashel et al., 1996). Thus, the levels of ppGpp in the cell vary from very low in rapidly growing high-energy-status cells to vastly elevated levels in cells grown in nutrient-poor medium or as cells enter the exponential-to-stationary phase transition in nutrient-depleted medium.
DksA is a small transcription factor that directly binds RNAP to alter its properties (Paul et al., 2004). Relatively constant levels of DksA sensitize RNAP to changing cellular levels of ppGpp and these two regulatory molecules act together to inhibit or stimulate transcription from some susceptible σ70- and σE-dependent promoters (Paul et al., 2004; 2005; Aberg et al., 2008; Costanzo et al., 2008). However, while lack of ppGpp and/or DksA severely restricts transcription from a range of σ54-dependent promoters in vivo, they do not have any effect on in vitro reconstituted σ54-dependent transcription (Bernardo et al., 2006). These findings, together with the effects of manipulating σ-levels and the properties of RNAP mutations that bypass the need for ppGpp, have led to a model in which ppGpp indirectly elevates the levels of available σ54-RNAP through competition between σ54 and other σ-factors for varying but always limiting levels of core RNAP (Jishage et al., 2002; Laurie et al., 2003; Bernardo et al., 2006; Szalewska-Palasz et al., 2007). In addition to σ54-RNAP levels and promoter affinity, regulator levels and ppGpp enhancements of IHF levels have also been implicated in rendering σ54-dependent transcription submissive to global regulatory input (Valls et al., 2002; Carmona et al., 2005; Bernardo et al., 2006). Here we specifically address the role of regulator levels by dissecting the control of DmpR levels. This analysis revealed an unexpected interplay between σ54- and σ70-dependent transcription and a regulatory scenario that is likely to be pertinent for other bacterial regulatory circuits that involve divergent but non-overlapping transcription.
Po output and DmpR levels are both growth phase-regulated
Because the transcriptional promoting activity of DmpR is controlled by binding of phenolic pathway substrates, the activity of the DmpR-activated Po promoter is also strictly responsive to the presence of these effector molecules. However, even in the presence of a phenolic effector, transcription from the Po promoter is growth phase-regulated. This level of regulation is reproduced with P. putida KT2440::dmpR-Tel, which expresses dmpR from its native Pr promoter from the host chromosome, when carrying Po–luxAB luciferase transcriptional reporters in monocopy or at 16–20 copies per cell on plasmids such as pVI945 (Fig. 1A; Sze et al., 1996; 2002; Sze and Shingler, 1999; Bernardo et al., 2006). In addition to growth phase regulation of transcription from Po, which is characterized by inactivity of the Po promoter until exponential-to-stationary growth transition, the data in Fig. 1A also illustrate the previously documented dependence of Po promoter output on an effector, σ54, and the integrity of the −24, −12 σ54 recognition sequence of the Po promoter.
In contrast to an earlier report, we found that DmpR levels also vary according to the growth phase in both P. putida KT2440::dmpR-Tel (Fig. 1B) and to a similar extent in P. putida CF600 where dmpR is in its native context on the 1–2 copy number pVI150 plasmid (Fig. 1C). We cannot fully explain why this growth phase regulation of DmpR levels was not detected previously (Sze et al., 1996), but it may be related to the different antibodies and/or dose–response of detection. The approximately three- to fivefold elevation of DmpR levels occurs at approximately the same time point as transcription from the Po promoter is detected (Fig. 1, compare A and B). These results suggested to us that regulation of DmpR levels might be integral to the regulation of the levels of (methyl)phenol catabolic enzymes through its action at the Po promoter.
Pr is a σ70 extended −10 promoter
Po and Pr drive divergent transcription from within the 406 bp dmp-operon to dmpR intergenic region. However, the exact location and identity of the Pr promoter has not previously been determined. Primer extension on total RNA isolated from P. putida KT2440 harbouring a plasmid that carries dmpR expressed from its native promoter identified the apparent transcriptional start site as an A base located 123 bp upstream of the translational start codon of dmpR (Fig. 2A). The same transcriptional start site was also identified irrespective of the presence and absence of an aromatic effector, and using total RNA isolated from P. putida CF600 carrying pVI150 (data not shown). The data shown in Fig. 2A were obtained using a primer complementary to a region located 55–78 bp upstream of the ATG initiation codon of dmpR. Other primers designed to identify transcripts originating further upstream or downstream within the intergenic region failed to identify any additional transcriptional start site (data not shown).
Analysis of the DNA sequence upstream of the transcriptional start site did not identify any appropriately located promoter motifs. However, a region that possesses five out of eight bases of a consensus σ70 extended −10 promoter sequence and a potential σ70−10 motif that lacked a cognate −35 promoter motif were found further upstream (Fig. 2B). Because artificial elevation of σ70 levels resulted in higher output from Pr–luxAB reporters (data not shown), we generated a series of transcriptional reporter plasmids that carried various portions of the Pr promoter region controlling transcription of luxAB to determine which of these σ70-related motifs constituted the Pr promoter. As shown in Fig. 2C, this analysis identified the region between −38 and +1 relative to the identified transcript start site as constituting the Pr core promoter. Deletion into the region of homology to the extended −10 promoter sequence abolished transcription (Fig. 2C, −33 to +8 bar), suggesting that the extended −10 promoter sequence is Pr. The importance of this region was verified by generating mutations that maintained the base composition but destroyed potential signature motifs; mutation of the potential −10 motif had little effect on transcription, while disruption of the extended −10 promoter motif totally abolished detectable transcription (Fig. 2C, right-hand hatched). Disruption of the extended −10 promoter motif was also found to abolish detectable transcription in an in vitro transcription assay with P. putidaσ70-RNAP and templates that carry the wild-type −38 to +8 Pr region or the mutated variant (Fig. 2C, insert).
Examination of the data in Fig. 2C suggests that the Pr promoter is unusually DNA context-sensitive. Both the extent of the upstream and downstream regions, and alteration of the downstream sequence (as in the potential −10 motif mutant) result in up to 1.5-fold differences in maximal promoter output. Notably, the critical σ70-Pr extended −10 promoter sequence is located 29 bp from the start site identified by primer extension in vivo. The same transcriptional start site was also identified using transcript generated in vitro from templates that carry the wild-type Pr promoter (Fig. S1). This datum excludes the possibility that the identified start site is a result of mRNA processing by extrinsic factors, and suggests that it might be a result of either premature termination of the primer extension reaction at a stable secondary structure within the mRNA or due to intrinsic cleavage by RNAP. Whichever is the case, for the purposes of orientation, hereafter we still refer to this start site as the +1 start of the transcript from Pr.
Transcription from the Pr promoter is auto-stimulated by effector-activated DmpR
The results outlined above demonstrate that the Po and Pr promoters drive non-overlapping divergent transcription from within the dmpR to dmp-operon intergenic region. The promoter region of Pr (−38 to +1) is separated from Po by 200 bp and is located 56 bp from UAS1 and UAS2 to which effector-activated DmpR binds to activate transcription from Po (Sze et al., 2002). Using a transcriptional Pr–luxAB fusion that carries the intergenic region (pVI938; Fig. 3A, upper), we found that the increase seen in DmpR levels at the exponential-to-stationary phase growth transition (Fig. 1B) is accompanied by an approximate twofold increase in transcription from Pr upon entering the stationary phase of growth (Fig. 3A, open squares). In addition, we unexpectedly found that stationary phase transcription from Pr is further increased by the presence of the DmpR effector 2-methylphenol (two- to threefold; Fig. 3A, open circles) to result in a net approximately fivefold increase in stationary phase transcription over that during exponential growth. This effector-dependent stimulation was only found with pVI938 in P. putida KT2440::dmpR-Tel as in Fig. 3A, and not when pVI938 was resident in P. putida KT2440 derivatives lacking a chromosomal copy of dmpR (data not shown). Thus, because the stimulation requires the simultaneous presence of an effector and DmpR, the observed auto-stimulation requires the effector-activated multimeric form of DmpR.
To determine if the auto-stimulatory effect of DmpR on Pr output at the transition between exponential and stationary growth is also found in the native system, we generated analogous Pr–luxAB transcriptional reporters on the native pVI150 catabolic plasmid of P. putida CF600, and on pVI150::ΔdmpR::Km, an internal dmpR gene replacement derivative that is incapable of producing DmpR. The single site recombination strategy used results in the luxAB reporter genes fused at +215 relative to the transcriptional start of Pr followed by plasmid DNA and a second complete copy of the intergenic region controlling native dmpR or the ΔdmpR::Km allele. As shown by the data in Fig. 3B, the above findings were reiterated in the native system with the presence of the non-metabolizable effector 2,3-dimethylphenol eliciting approximately threefold higher transcription from Pr in the strain with a functional dmpR gene as compared with that of the DmpR null mutant.
Auto-stimulation by DmpR determines sensitivity of the dmp-operon promoter Po to phenolics
Auto-stimulation of Pr activity by effector-activated DmpR prompted us to address if the absolute levels of DmpR are limiting for Po output and if they affect the sensitivity of the Po promoter to the presence of phenolics. To this end we generated a chromosomal transcriptional dmpR-Po–luxAB fusion, with Po in its native configuration relative to dmpR, using P. putida KT2440::dmpR-Tel. We determined the levels of DmpR and the activity of the DmpR-regulated Po promoter of this strain (PP2315) when harbouring either a vector control plasmid or an expression plasmid with dmpR under control of the lacIQ/Ptac expression system (pVI465). The transcriptional output from the Po promoter in response to the varying levels of DmpR was first monitored across the growth curve with saturating concentrations of 2-methylphenol (2 mM). The data illustrate that the low levels of DmpR serve as one regulatory check point during the exponential phase of growth (Fig. 4B) as elevated levels of DmpR provided by pVI465 elicit elevated levels of transcription from Po during this phase of growth. A second regulatory check point that results in elevated transcription at the exponential-to-stationary phase transition is also evident. As expanded on in the following sections, we primarily attribute this second check point to ppGpp-mediated effects on this regulatory circuit.
The artificial increase in DmpR levels that results from leakiness of the lacIQ/Ptac promoter of pVI465 also enhances Po output in the stationary phase, suggesting that even the naturally increased levels of DmpR observed in this phase of growth are limiting for transcription from Po (Fig. 4B). To examine if the auto-stimulatory effect of DmpR on stationary phase transcription from Pr (Fig. 3A and B) resulted in a concomitant accumulation of DmpR, we compared DmpR protein levels in stationary phase PP2315 cells grown in the presence and absence of effector. As anticipated, immuno-detection showed that the presence of 2 mM 2-methylphenol in the culture medium resulted in increased DmpR levels (Fig. 4C, upper panel), to approximately five- to sevenfold higher than those found with cultures grown in the absence of an effector. This increase is somewhat greater than the approximately threefold increase in transcription from Pr elicited by addition of the effector in the culture media (Fig. 3A). Because binding of an effector allows subsequent nucleotide-triggered multimerization of DmpR to a protease-resistant form, differences in DmpR dimer and multimer stabilities may, at least in part, account for these differences (Wikström et al., 2001).
We reasoned that such effector-dependent enhancement in DmpR levels might also influence the sensitivity of Po output to the concentration of the effector. To test this idea, we monitored stationary phase transcription from Po in response to increasing concentrations of 2-methylphenol. Under these conditions, sensitivity of the Po promoter to the effector was severely curtailed at low effector concentrations as compared with when DmpR levels were artificially elevated via pVI465 (Fig. 4C). This is exemplified by the two- to threefold difference in output from the Po promoter at high effector concentration as compared with the ∼20-fold higher output at low effector concentrations. Based on these data, we conclude that DmpR levels are normally limiting for Po activity even in the stationary phase where DmpR levels are at their highest, and that elevated levels of DmpR that occur through auto-stimulation of Pr activity increase the sensitivity of the Po promoter to effectors. These results suggest that the auto-stimulatory activity mediated by DmpR would be integral to the regulation of the levels of (methyl)phenol catabolic enzymes to attune Po-output to both the growth status of the cell and the levels of available substrate for catabolism.
Stationary phase stimulation of Pr activity requires Po promoter elements
Auto-stimulation of Pr activity that is evident in both the native system and with a transcriptional reporter that carries the entire intergenic region was not found with a reporter that carried the core Pr −60 to +8 promoter region (pVI929, data not shown). To further delineate the DNA region required to detect DmpR-mediated auto-stimulation of Pr activity, we used two transcriptional reporter plasmids analogous to the −266 to +215 Pr–luxAB pVI938 reporter that differed only in possession of the Po promoter region or the Po promoter and DNA encompassing the UAS binding sites for DmpR (pVI940 and pVI941), and as a negative control, plasmid pVI942 that lacks part of the core Pr extended −10 promoter sequence (Fig. 5A). Transcriptional analysis using these plasmids in P. putida KT2440::dmpR-Tel demonstrated that removal of DNA features that are associated with divergent transcription from the Po promoter results in defective transcription from the Pr promoter upon entry into the stationary phase (Fig. 5B, up- and down-triangles). As would be anticipated, only background levels of luciferase activity were evident with the negative control plasmid pVI942 (Fig. 5B, diamonds). The results shown in Fig. 5B were generated in the absence of 2-methylphenol; however, the response profiles shown are indistinguishable from those found with these three plasmids in the presence of effector or in P. putida KT2440 that lacks DmpR (data not shown). Hence, efficient transcription from the Pr promoter in the stationary phase of growth is strictly dependent on DNA elements situated ∼200 bp upstream of Pr and, in their absence, auto-stimulation by DmpR is abolished even when DNA encompassing its binding sites is present. This suggested to us that both these features of the regulation of Pr activity are related to the Po promoter rather than DmpR binding per se.
Auto-stimulation of Pr activity requires transcription from Po
To dissect the role of Po promoter occupancy and activity on transcriptional activity from Pr, we generated the same Po mutation that abolishes Po activity (Fig. 1A) in the context of the Pr–luxAB reporter plasmid pVI938. This mutation of the −24, −12 motif of Po maintains the base composition but destroys the signature motif and would be anticipated to prevent σ54-RNAP binding upon which the activity of Po depends. That this mutation indeed abolishes specific binding of σ54-RNAP to the Po promoter was confirmed by Biacore plasmon resonance analysis of P. putidaσ54-RNAP binding to chip-coupled double-stranded DNA encompassing the native Po promoter or its cognate mutant (Fig. 6A).
The activities of two Pr–luxAB transcriptional reporters that only differ with respect to the Po mutation and thus the ability to bind σ54-RNAP were then compared in P. putida KT2440::dmpR-Tel and a derivative thereof that lacks σ54. The results from this analysis are summarized in Fig. 6B, which shows stationary phase levels from the Pr–luxAB transcriptional reporters. As with analysis of reporters that lack the Po promoter region (Fig. 5B), lack of σ54 or the capacity of the Po promoter to bind σ54-RNAP both resulted in failure to induce transcription from the Pr promoter upon entering the stationary phase of growth, resulting in lower levels of transcription in this phase of growth (Fig. 6B, black bars). Similarly, lack of σ54 or the capacity of the Po promoter to bind σ54-RNAP essentially abolished stimulation of Pr activity by effector-activated DmpR (Fig. 6B, hatched bars). These results show that: (i) efficient transcription from the Pr promoter upon entering the stationary phase is dependent on σ54-RNAP occupancy of the Po promoter and (ii) auto-stimulation of Pr activity by effector-activated DmpR is mediated through σ54-RNAP activity at the Po promoter.
DmpR-activated open-complex formation and concomitant transcription from Po-bound σ54-RNAP is aided by IHF (Sze et al., 2001). Lack of IHF binding to its DNA sites situated between Po and the UASs for DmpR reduces transcription from the σ54-Po promoter by approximately fivefold in E. coli and ∼10-fold in P. putida KT2440::dmpR-Tel (Bernardo et al., 2006 and data not shown). We reasoned that if divergent transcription from Po underlies stimulation of Pr activity, then lack of IHF binding – through reduced Po activity – should also reduce the stimulatory effect of DmpR on Pr activity. To test this idea, we employed a previously characterized mutation of the IHF binding site that maintains the base composition but destroys the signature motif and the ability to bind IHF (Bernardo et al., 2006). The IHF site mutation was introduced in the context of the Pr–luxAB reporter plasmids pVI938 (bearing native Po) and pVI939 (bearing mutated Po). Pr activity from these two reporter plasmids (pVI960 and pVI961) was then compared with that of the parental plasmids in P. putida KT2440::dmpR-Tel or in P. putida KT2440 and an IHF null counterpart (Fig. 6B).
Lack of the capacity to bind IHF (Fig. 6B, left-hand black bars), like lack of IHF in the KT2440 IHF null mutant (Fig. 6B, right-hand black bars), resulted in lower levels of transcription from Pr during stationary phase growth. Lack of both σ54-RNAP and IHF binding resulted in an additive effect, with an approximately three to fourfold reduction in Pr output as compared with the reporter bearing the native intergenic region. Importantly, lack of IHF binding (which reduces σ54-Po activity ∼10-fold) markedly reduced the ability of effector-activated DmpR to stimulate transcription from σ70-Pr (Fig. 6B, hatched bars). We conclude from this datum that: (i) efficient transcription from the Pr promoter upon entering the stationary phase is dependent on binding of both σ54-RNAP and IHF to the intergenic region and (ii) that auto-stimulation of Pr activity by effector-activated DmpR is mediated through open-complex formation and/or concomitant divergent transcription from Po.
To assess if active divergent transcription per se can influence Pr activity, we also replaced the σ54-Po promoter by the σ70-λPL promoter or a mutant derivative thereof that is incapable of promoting transcription, and compared the activities of these two reporters in P. putida KT2440::dmpR-Tel as described above. Transcription from Pr with the reporter plasmid bearing the mutated σ70-λPL promoter (pVI963) was similar to that of the reporter bearing the mutated σ54-Po; however, transcription from Pr with the reporter plasmid bearing the active σ70-λPL promoter (pVI962) was twofold higher (Fig. 6B, compare black bars). These results demonstrate that the stimulation seen at Pr by divergent transcription is not unique to σ54-Po and can occur with other types of promoter arrangements. Further, because effector-activated DmpR would not directly affect the activity of the σ70-λPL promoter, the finding that the presence of a DmpR effector does not significantly alter the output from the Pr promoter of these two reporters (Fig. 6B, hatched bars) suggests that in contrast to σ54-RNAP and IHF-binding, binding of DmpR per se does not influence Pr output in vivo.
ppGpp and DksA stimulate transcription from the Pr promoter in vivo and in vitro
The inability to synthesize ppGpp results in approximately twofold decreased stationary phase levels of DmpR in E. coli and P. putida KT2440::dmpR-Tel (Bernardo et al., 2006 and data not shown). The alarmone ppGpp in collaboration with its cofactor DksA has been proposed to elevate the levels of σ54-RNAP upon entry into the stationary phase (Laurie et al., 2003; Bernardo et al., 2006; Szalewska-Palasz et al., 2007). This, combined with our finding that σ54-RNAP via the Po promoter is necessary for efficient stationary phase transcription from Pr (Fig. 6B), prompted us to investigate the impact of ppGpp and DksA on transcription from the Pr promoter. To this end, we monitored transcription from the two Pr–luxAB reporters that differ only by the Po mutation that renders it incapable of binding σ54-RNAP in P. putida KT2440::dmpR-Tel and derivatives thereof that lack the capacity to synthesize ppGpp (ppGpp0) or DksA.
As summarized in Fig. 7A, we found that both ppGpp and DksA were necessary for high-level stationary phase transcription from the Pr promoter. Based on the data in the preceding section, we attribute the approximately twofold reduction in the transcription from Pr caused by the mutation of Po in the wild-type background to the lack of capacity of the mutated Po promoter to bind the available σ54-RNAP and thereby stimulate transcription from Pr. Consistent with the proposed role of ppGpp and DksA in elevating the levels of available σ54-RNAP, transcription from Pr from the reporter bearing native Po was substantially decreased in ppGpp0 and DksA null strains (Fig. 7A). Interestingly, however, Pr activity was also further decreased in ppGpp0 and DksA null strains with the reporter that carries the Po mutation and is thus independent of effects of these regulatory molecules through σ54-RNAP (Fig. 7A, right-hand bars). This result suggested that ppGpp and DksA might additionally affect Pr activity through a mechanism that is independent of their effects through Po and σ54-RNAP.
To test potential direct effects of ppGpp and DksA on Pr activity, we reconstituted transcription from the Pr promoter in vitro using P. putidaσ70-RNAP. For comparison, we also analysed the effects of ppGpp and DksA on the σ70 PthrABC and rrnB P1 promoters, which have previously been shown to be positively (PthrABC) and negatively (rrnB P1) co-regulated by ppGpp and DksA in vitro (Paul et al., 2005). Consistent with previous work, we found that ppGpp has little effect on its own on transcription from any of the promoters tested (Fig. 7B, grey bars), but caused a co-stimulatory (Pr and PthrABC) or co-inhibitory (rrnB P1) effect in the presence of DksA (Fig. 7B, black bars). With the Pr promoter, DksA also markedly stimulated transcription from the Pr promoter even in the absence of ppGpp (Fig. 7B, hatched bar).
To determine the potential additive effects of σ54-RNAP activity at Po and the direct effects of ppGpp/DksA on Pr activity, we performed single-round transcription assays using the same Pr template that carries native Po and a derivative that carries the Po promoter mutation. In these experiments, the reaction mixes also contained the active form of DmpR to elicit open-complex formation and transcription from Po when σ54-RNAP was added. As shown in Fig. 7C and D, activity at Po mediated by the addition of σ54-RNAP results in an approximately twofold stimulation of transcription from Pr both in the presence and absence of ppGpp/DksA. As would be anticipated from the in vivo data (Fig. 6B), no enhanced transcription from Pr was detectable with the template that has the Po mutation. Interestingly, while direct stimulation of Pr output by ppGpp and DksA was found in similar experiments using linear templates, stimulation of Pr activity by the addition of σ54-RNAP was not (Fig. S2). Taken together, these results strongly support the idea that ppGpp and DksA have a dual effect on transcription from Pr in vivo, both directly at the level of transcription initiation from the Pr promoter, and indirectly through their effects on σ54-RNAP levels and activity at the Po promoter.
DmpR stimulates σ70-Pr output through open-complex formation and transcription from σ54-Po
The twofold stimulation of Pr output by the addition σ54-RNAP shown in Fig. 7C was not detected when any of the molecules that are needed for DmpR activity were individually omitted from the reaction mixes, namely DmpR, dATP or the effector 2-methylphenol (data not shown). Thus, consistent with in vivo findings, the presence of the effector-activated multimeric form and binding of DmpR does not influence the output from Pr unless the Po promoter is functional. IHF facilitates open-complex formation and transcription from σ54-Po in vitro when σ54-RNAP levels are limiting (as in the cell). The twofold stimulation in Pr activity in vitro (Fig. 7C) was gained in the presence of saturating levels of σ54-RNAP in the absence of IHF. Addition of IHF to reaction mixes did not result in further stimulation of transcription from either Po or Pr under these in vitro conditions (Fig. S3).
This datum above suggests that DmpR has its stimulatory effects on transcription from Pr through mediating open-complex formation and/or transcription from Po rather than through binding at its UAS sites per se. To examine this issue further, we used variants of DmpR that lack the regulatory N-terminal domain. The constitutively active variant (ΔA2-DmpR), like intact effector-activated DmpR, could stimulate transcription from Pr in the presence of σ54-RNAP. However, a previously defined mutant derivative (ΔA2-DmpR-A315T), which can bind and hydrolyse ATP but cannot activate transcription from Po (Wikström et al., 2001), could not (Fig. 7E). We also examined the effects of a variant consisting of the catalytically active central domain of DmpR (C-DmpR). This derivative lacks both its regulatory N-terminal domain and its DNA binding domain and thus must act from solution. C-DmpR activates transcription from Po although poorly, and in vitro comparison of its transcriptional promoting abilities showed that saturating concentrations of C-DmpR elicited the same level of transcription from σ54-Po as 10 nM effector-activated DmpR under our experimental conditions (data not shown). The data in Fig. 7F show that these concentrations likewise elicit the same levels of stimulated transcription from Pr. Taken together, the in vitro data strongly support the suggestions from the in vivo data (Fig. 6B) that DmpR mediates stimulation of transcription from Pr solely by promoting open-complex formation and concomitant divergent transcription from Po.
The aromatic sensor-regulator DmpR controls transcription from the σ54-Po promoter, and thus formation of the (methyl)phenol catabolic enzymes via the Po-driven dmp-operon. Here we have shown that divergent but non-overlapping transcription from the σ54-Po promoter stimulates transcription from the σ70-Pr promoter of dmpR. We found that efficient stationary phase transcription from Pr requires occupancy of the Po promoter by σ54-RNAP and binding of IHF (Fig. 6B), and that substrate-activated DmpR further stimulates transcription for its own synthesis by mediating open-complex formation and concomitant transcription from Po both in vivo and in vitro (Figs 3, 6 and 7). Based on these findings, we propose a regulatory scheme for this novel role of σ54-RNAP in controlling output from a σ70-dependent promoter as depicted in Fig. 8. This scheme also incorporates new additional roles for ppGpp in this regulatory circuit to reinforce both silencing of Po when preferred substrates are present and robust production of the catabolic enzymes when preferred substrates are lacking. Under high-energy conditions of rapid growth on rich media (Fig. 8A), the levels of σ54-RNAP and IHF are low, resulting in low binding site occupancy and consequent low Pr activity and DmpR levels. Upon nutritional or other stress conditions that elicit ppGpp synthesis (Fig. 8B), however, consequential ppGpp/DksA-mediated elevation of σ54-RNAP and IHF levels increase binding site occupancy and thus Pr activity, resulting in higher levels of DmpR. Direct ppGpp/DksA stimulation of Pr activity (Fig. 7) further enhances DmpR levels. The higher levels of DmpR, σ54-holoenzyme and IHF would thus poise the system for both sensitive detection and rapid high-level production of pathway enzymes should a substrate be available. Under these conditions (Fig. 8C), DmpR takes up its active multimeric form and through promoting transcription from Po imparts an auto-stimulatory feed-forward loop to both elevate its own expression and synthesis of the catabolic enzymes. The positive feed-forward loop would be held in check by the output from Po, which cannot exceed a certain maximum determined by co-occupancy of σ54-RNAP bound to Po and DmpR bound to its UAS sites, and their productive interaction that is facilitated by IHF.
Under the in vitro conditions used here, the effects of the ability of Po to drive transcription (approximately twofold), combined with the direct effects of ppGpp and DksA on the σ70-Pr promoter (approximately threefold), results in a net approximately sixfold enhanced transcription from Pr (Fig. 7C and D). Given that binding of σ54-RNAP and IHF are required for stimulation of Pr output in the stationary phase (Fig. 6B), it appears likely that elevated levels of σ54-RNAP and IHF would contribute to the enhanced transcription from Pr that we observe at the exponential-to-stationary phase transition (Fig. 3). However, because ppGpp/DksA also act directly through σ70-RNAP to stimulate transcription from the Pr promoter, direct action of these two regulatory molecules at Pr would also be anticipated to contribute to elevated transcription from Pr as cells enter the stationary phase. The net effect of ppGpp-mediated activities and auto-stimulation of DmpR levels through Po activity is important for graded transcription and sensitivity of the Po promoter to aromatic effectors (Fig. 4). The resulting enhanced levels of DmpR when expressed at monocopy allow robust transcription from the Po promoter when concentrations of the most potent natural effector of DmpR (2-methylphenol) are ∼0.1 mM or higher (Fig. 4C). These findings suggest that auto-regulation of DmpR levels provides a regulatory mechanism to keep the activity of the Po promoter locked under high-energy conditions, and also to help keep the activity of Po in check unless sufficient substrate is present to provide an appropriate metabolic return for the investment made in producing the enzymes for their catabolism.
The finding that the σ70-Pr promoter is located far upstream from the transcriptional start site identified by primer extension analysis in vivo and in vitro (Fig. 2 and Fig. S1) suggests that the identified start site is either the result of a stable mRNA secondary structure that blocks progression of the primer extension reaction or possibly due to intrinsic cleavage by σ70-RNAP. Whatever is the case, as this transcript start site is 123 bp from the ATG initiation codon of dmpR, the mRNA for DmpR has an extensive 5′-untranslated region. The presence or absence of this 5′-untranslated leader region does not affect growth phase regulation or auto-stimulation of Pr activity across the growth curve; however, its presence does result in an approximately fivefold reduction in absolute output levels of Pr–luxAB transcriptional reporters in vivo (L.U.M. Johansson, unpublished). Non-coding regions of mRNA can affect gene expression by a wide variety of different mechanisms (Fuglsang, 2005; Komarova et al., 2005; Kaberdin and Blasi, 2006), and analysis of the role of this region in controlling DmpR levels is the subject of ongoing research.
Divergent transcription of a regulatory gene and at least one of the cognate promoters it controls is a common theme in bacterial transcriptional regulation and is frequently observed in the regulatory circuits that control catabolism of aromatic compounds. Where known, interplay of the divergent transcriptional units involves some overlap of cognate regulatory regions. For example, in pWW0-encoded toluene catabolism, the binding sites for the XylR regulatory protein used to activate transcription from the divergently transcribed xylS gene overlap the xylR promoter (Ruíz et al., 2004), while a common intergenic binding site that overlaps the promoter for the AtzR regulator controls both its own synthesis and expression of the divergent atzDEF operon involved in degradation of the herbicide atrazine (Porrua et al., 2007). However, to our knowledge, the auto-stimulation mechanism we describe for Pr is the first example of interplay through non-overlapping divergent transcription. Many examples exist, from both chromosomal and plasmid-encoded phenol catabolic systems, which have analogous genetic organizations as found in the dmp-system (Shingler, 2004). Thus, stationary phase transcription of these regulatory circuits is likely to be controlled in a similar manner.
The σ70-Pr promoter is unusually DNA context-dependent (Fig. 2). Interestingly, while direct stimulation of Pr output by ppGpp and DksA could be recapitulated using linear templates, stimulation of Pr output by active transcription from Po was not (Fig. 7C and D; Fig. S2). These properties suggest susceptibility of the stacking properties of the Pr promoter DNA for binding and/or initiation of transcription by σ70-RNAP. The intergenic region between σ54-Po and σ70-Pr is very crowded and has to accommodate two RNAP holoenzymes, IHF and the multimeric form of DmpR. In vivo, binding of σ54-RNAP and IHF (but not DmpR) per se stimulate transcription by σ70-RNAP from the Pr promoter. It is plausible that DNA-bending caused by binding of these proteins may allow upstream DNA to interact with σ70-RNAP to stimulate its activity at Pr, and that this process is further facilitated by open-complex formation and/or transcription from Po. However, further work is required to test these possibilities.
In vivo, lack of σ54-RNAP binding and active transcription from the Po promoter results in five- to sixfold reduced transcription from σ70-Pr (Fig. 6B). This effectively places the σ70-Pr promoter under control of σ54-RNAP without it having a cognate binding site for the σ54-holoenzyme. Thus, output from the Pr promoter is determined by the active pools of two forms of RNAP. Conceptually, this kind of dual sensitivity of a promoter to the levels of two different holoenzymes need not be limited to the case of σ70/σ54-holoenzymes. The activities of many σ-factors important for developmental programmes and for counteracting stress conditions are modulated in response to specific signals to alter the relative composition of alternative RNAP-holoenzymes of the cell. Hence, it is plausible that other regulatory circuits that involve non-overlapping divergent promoters dependent on different σ-factors could likewise be made subservient to conditions that elicit activity of an alternative σ-factor through a mechanism analogous to that shown here for the σ70-Pr promoter.
Pseudomonas putida strains (Table 1) were cultured at 30°C in Luria–Bertani/Lennox (LB) medium (AppliChem GmbH) supplemented with antibiotics as appropriate for the strain and resident plasmid selection. Strains and plasmids were constructed by standard DNA techniques as detailed in the Supporting information. Plasmids used for in vivo and in vitro transcription assays are listed in Table 2, while primers used in DNA manipulations are listed in Table S1. Luciferase luxAB reporter assays were performed as described in Sze and Shingler (1999) and Sze et al. (2002) and data points are the average of duplicate determinations from each of two or more independent experiments ± standard errors. Immuno-detection was performed as described in Bernardo et al. (2006), except that affinity-purified polyclonal antibodies against P. putida RpoN-His were used.
Table 1. P. putida strains .
Resistance abbreviations are Gm, gentamicin; Km, kanamycin; Tc, tetracycline; Tel, tellurite.
Parental strain of the (methyl)phenol catabolic plasmid pVI150
CbR, −266 to +8 Pr on pTE103, in vitro transcription template
CbR, as pVI948 but a mutation of Po
CbR, −38 to +8 Pr on pTE103, in vitro transcription template
CbR, as pVI950 but a mutated extended −10 promoter sequence
CbR, −72 to +16 PthrABC, in vitro transcription template (from R.L. Gourse)
CbR, −66 to +50 rrnB P1, in vitro transcription template (from R.L. Gourse)
Analysis of total RNA isolated using the RNeasy Midi Kit (Qiagen) with [γ-32P]-ATP kinase-labelled oligonucleotide 894 was essentially as described in Balsalobre et al. (2003). For analysis of in vitro generated transcripts from pVI948 and pVI950, a 5′ 6-carboxyfluorescein (FAM)-labelled oligonucleotide 2425 was used. Denatured transcripts were annealed with 1 pmol of primer 2425 at 58°C for 20 min. After cooling to room temperature for 15 min, 40 μl reactions containing dNTPs (final concentration 1.5 mM each), 2 U RNasin (Ambion) and 2 U AMV-RT (Roche) in AMV-RT buffer supplied by the manufacturer, were incubated at 42°C for 60 min. Reactions were then supplemented by an additional 2 U AMV-RT, and incubation continued for a further 60 min. Precipitated cDNA was co-electrophoresed with 0.5 μl of GeneScan 350-Tamra standards (Applied Biosystems) using an ABI DNA sequencer, and the primer extension product sized using the GeneScan Analysis Software.
ppGpp and purified proteins for in vitro assays
The nucleotide ppGpp was synthesized and purified as previously described (Cashel, 1974). The N-terminally histidine-tagged catalytic central domain of DmpR (C-DmpR-His) was expressed from the T7 promoter of pVI647 and purified by nickel affinity chromatography as previously described for DmpR-His, ΔA2-His-DmpR-wt and ΔA2-His-DmpR-A315T (O'Neill et al., 2001; Wikström et al., 2001). Native IHF was as described (Sze et al., 2001). Native P. putida KT2440 core RNAP and σ70-RNAP holoenzymes were purified as described in Hager et al. (1990). N-terminal His-tagged E. coli DksA was a gift from A. Åberg (Umeå University, Sweden). Native P. putidaσ70 and σ54 were purified after overexpression in P. putida KT2440 or an E. coliσ54 null mutant respectively, as detailed in the Supporting information.
Plasmon resonance assays
Plasmon resonance experiments were carried out on a Biacore 3000 system (Uppsala, Sweden). The 116 bp biotin-labelled double-stranded DNA fragments incorporating the native σ54-Po promoter or a mutant Po promoter were coupled to Streptavidin immobilized on a CM4 chip by amine coupling. The DNA fragments were obtained by PCR amplification using primer Po-f2 (homologous to DNA upstream of the Po promoter) and the biotin-coupled primer lux-r2, which is complementary to the luxAB genes of pVI700 (native Po) and pVI947 (Po mutation) that were used as template DNA. Binding experiments were run using B buffer [50 mM Tris-HCl pH 7.5, 100 mM NaCl, 10 mM MgCl2, 1 mM DTT, 0.1 mM EDTA, 0.054% P-Surfactant (Biacore)] at 25°C. P. putidaσ54-RNAP holoenzyme was formed by pre-incubation of 20 nM core RNAP and 160 nM σ54 for 5 min in buffer B, then injected at a flow rate of 20 μl min−1.
in vitro transcription assays
Multiple- and single-round transcription assays (final volume 20 μl) were performed with 10 nM template DNA at 30°C in T-buffer containing 35 mM Tris-Ac pH 7.9, 70 mM KAc, 5 mM MgAc2, 20 mM NH4Ac, 1 mM DTT and 0.275 mg ml−1 bovine serum albumin essentially as described in Bernardo et al. (2006). In brief, P. putidaσ70-RNAP (25 nM) or P. putida core RNAP (25 nM) and P. putidaσ70 or σ54 (200 nM) were pre-incubated for 5 min. ppGpp and DksA or DksA storage buffer were added as appropriate and incubation continued for a further 5 min prior to addition of template DNA and open complex formation (10 min). In multiple-round assays, transcription was initiated by addition of NTPs (ATP, 500 μM; GTP and CTP, 200 μM each; UTP, 80 μM and [α-32P]-UTP (5 μCi at > 3000 Ci mmol−1), and re-initiation prevented after 10 min by the addition of heparin (0.125 mg ml−1). Single-round assays also included DmpR or a derivative thereof, 0.5 mM 2-methylphenol, 4 mM dATP and 200 μM GTP and CTP during open-complex formation as indicated. Transcription was initiated by addition of 500 μM ATP, 80 μM UTP and [α-32P]-UTP in the presence of heparin and allowed to proceed for 10 min. Reactions were terminated by addition of formamide loading buffer, and transcripts were analysed on a 5% acrylamide gel containing 7 M urea, followed by quantification using phospho-imaging.
We thank R.L. Gourse and K. Ygberg for providing plasmids, A. Åberg for His-DksA, A. Szalewska-Palasz for help in purifying proteins, and S. Saaf with primer extension analysis. This work was supported by the Swedish Research Council.