The RNA polymerase-binding protein DksA is a cofactor required for guanosine tetraphosphate (ppGpp)-responsive control of transcription from σ70 promoters. Here we present evidence: (i) that both DksA and ppGpp are required for in vivoσ54 transcription even though they do not have any major direct effects on σ54 transcription in reconstituted in vitro transcription and σ-factor competition assays, (ii) that previously defined mutations rendering the housekeeping σ70 less effective at competing with σ54 for limiting amounts of core RNA polymerase similarly suppress the requirement for DksA and ppGpp in vivo and (iii) that the extent to which ppGpp and DksA affect transcription from σ54 promoters in vivo reflects the innate affinity of the promoters for σ54-RNA polymerase holoenzyme in vitro. Based on these findings, we propose a passive model for ppGpp/DksA regulation of σ54-dependent transcription that depends on the potent negative effects of these regulatory molecules on transcription from powerful stringently regulated σ70 promoters.
The σ54-factor is involved in controlling many physiological processes that are responsive to environmental cues ranging from assembly of motility organs and chemotaxis transducers, through nitrogen assimilation and the utilization of different carbon sources, to alginate biosynthesis (reviewed in Valls et al., 2004). Hence, global regulatory factors that alter the activity or availability of σ54-RNA polymerase (σ54-RNAP) have the potential to mediate far-reaching effects on integrated bacterial responses. One such factor is the nucleotide alarmone guanosine tetraphosphate (ppGpp), which is synthesized in response to a variety of nutritional limitations and physicochemical stresses through the action of RelA (synthetase I) and the dual-function SpoT protein (synthetase II) (reviewed in Cashel et al., 1996). This nutritional/stress alarmone was first identified through its role in negative regulation of powerful stringent σ70-dependent promoters (e.g. rRNA and tRNA promoters) to adjust translational capacity to growth demands. ppGpp is assisted in this process by the RNAP-binding protein DksA which acts synergistically with ppGpp to amplify its effects on σ70 transcription (Paul et al., 2004a,b; 2005). Structural modelling suggests that the coiled coil of DksA protrudes through the secondary channel of RNAP to stabilize ppGpp bound adjacent to the active site, thus providing a structural basis for the observed synergy of ppGpp and DksA on transcription (Artsimovitch et al., 2004; Perederina et al., 2004).
The molecular mechanism(s) by which ppGpp and DksA affect transcriptional initiation from σ70 promoters is not fully resolved. Both of these regulatory molecules reduce the lifetime of competitor-resistant open complexes at all σ70 promoters analysed so far. The lifetimes of competitor-resistant complexes at rRNA promoters are intrinsically very short, and further shortening by ppGpp/DksA has been proposed to underlie the direct negative effects of ppGpp and DksA on transcription from these promoters (Paul et al., 2004a, and references therein). Support for this mechanism also comes from suppressor mutants within the β- and β′-subunits of core RNAP, which likewise destabilize competitor-resistant complexes at σ70 promoters (Zhou and Jin, 1998). During rapid growth, transcription from the powerful stringent σ70-rRNA promoters occupies approximately 60–70% of the transcriptional machinery (Bremer and Dennis, 1996). Thus, a likely consequence of ppGpp/DksA-mediated downregulation is an increase in the available pool of core RNAP for holoenzyme formation. In addition to having direct negative effects on transcription from stringent σ70 promoters, ppGpp and DksA also exert direct positive effects on transcription from σ70 promoters controlling amino acid biosynthetic operons (Paul et al., 2005). The lifetimes of competitor-resistant complexes at ppGpp/DksA-stimulated amino acid biosynthetic promoters are unusually long, and ppGpp/DksA-reduced half-life is not rate-limiting for transcription in these cases. Rather, ppGpp and DksA have been proposed to stimulate transcription by reducing the energy of a transition state intermediate(s) to accelerate rate-limiting formation of open complexes (Barker et al., 2001; Paul et al., 2005).
Previous work has also demonstrated that efficient in vivo transcription from many promoters that are dependent on alternative σ-factors also requires ppGpp (Jishage et al., 2002, and references therein). In this capacity, ppGpp has been proposed to modulate the outcome of σ-factor competition for limiting amounts of core RNAP resulting in increased association of alternative σ-factors such as stationary-phase σS, the heat-shock σH and the structurally distinct σ54, over that of the housekeeping σ70-factor (reviewed in Nystrom, 2004; Magnusson et al., 2005). Although yet to be tested, stabilization of ppGpp binding by DksA also implicates DksA in these global regulatory mechanisms. In the case of σ54-dependent transcription, this proposed role of ppGpp is based on analysis of the DmpR-regulated σ54-Po promoter that controls transcription of a (methyl)phenol catabolic operon of the pVI150 plasmid of Pseudomonas sp. strain CF600 (reviewed in Shingler, 2003). The σ54-factor is structurally unrelated to the σ70 family of proteins, and programmes the RNAP to bind the unusual −24, −12 class of promoters (consensus TGGCAC N5 TTGCa/t; Barrios et al., 1999). Expression levels of σ54 are about 16–20% of the levels of σ70, and the levels of both of these σ-factors are constant throughout the growth curves and under different growth conditions (Ishihama, 2000). In contrast to σ70 and σ70-like alternative σ-factors, σ54 imposes kinetic constraints on open complex formation by the holoenzyme polymerase (reviewed in Zhang et al., 2002). Consequently, transcription from this class of promoters requires activators that utilize nucleotide hydrolysis to remodel the closed complex to allow initiation of transcription. For DmpR, binding of aromatic phenolic compounds to its N-terminal regulatory domain is required to alleviate interdomain repression to give the transcription-activating form of the protein (O’Neill et al., 1998; 2001; Wikström et al., 2001).
Superimposed on the aromatic-responsive regulation of σ54-Po transcription mediated through DmpR, global regulation mediated by ppGpp links the output from Po to the physiological status of the cell. Transcription from Po is low during exponential growth on rich media, but swiftly increases at the transition from exponential to stationary phase where ppGpp rapidly accumulates, and in response to growth conditions or artificial manipulations that elicit high ppGpp levels (Sze et al., 1996; Sze and Shingler, 1999). Consistently, transcription from this σ54-dependent promoter is severely reduced in Escherichia coli and Pseudomonas putida strains devoid of ppGpp (Sze and Shingler, 1999; Sze et al., 2002). Two main lines of evidence indicate that this dominant level of physiological regulation involves ppGpp-mediated modulation of the ability of σ54 to gain access to limiting core RNAP in intact cells: (i) that the requirement for ppGpp can be suppressed by underproduction and/or sequestering of σ70, leading to transcription from the σ54-Po promoter during exponential growth and to a dramatic increase (> 10-fold) in transcriptional output in the post-exponential phase, and (ii) that the in vivo requirement for ppGpp can also be suppressed by four mutant σ70 proteins that all exhibit defects in competing against σ54 for core RNAP in vitro (Laurie et al., 2003). Hence, ppGpp-mediated enhancement of the otherwise poor ability of σ54 to access limited amounts of available core RNAP has been proposed to lead to elevated levels of the σ54-RNAP holoenzyme, thus allowing occupancy and transcription from the Po promoter (Laurie et al., 2003).
The identification of DksA as a critical protein in ppGpp-mediated positive and negative regulation of σ70 promoters prompted us to evaluate the role of DksA in ppGpp-mediated regulation of σ54-dependent transcription. Based on the results from both in vitro and in vivo assays, we propose a passive model for ppGpp/DksA regulation of σ54-dependent transcription that depends on their potent negative effects on transcription from powerful stringently regulated σ70 promoters.
Both ppGpp and DksA are required for efficient σ54 transcription of Po in vivo
Transcription from the Po promoter of pVI150 is growth phase-regulated, and growth phase-dependent transcription from this promoter is maintained with a Po–luxAB transcriptional reporter when carried as a single copy on the host chromosome or in multiple copies on an RSF1010-based plasmid in P. putida (Sze et al., 1996; 2002; Sze and Shingler, 1999). For in vivo transcription analysis of promoter activity in wild-type and mutant E. coli strains, we used the RSF1010-based σ54-Po promoter luciferase transcriptional reporter plasmid pVI466 (dmpR-Po–luxAB), which carries the dmpR gene in its native configuration with respect to Po, with the luxAB genes fused at +291 relative to the transcriptional start. Because of the low expression from the native promoter of dmpR in E. coli, this genetic system maintains regulator levels close to those of the native pVI150 plasmid of Pseudomonas CF600, and reproduces the transcriptional profile observed from the Po promoter in its native context in P. putida (Fig. 1A, squares; Sze et al., 1996; Sze and Shingler, 1999). The dependence of the Po promoter on ppGpp leads to a 7- to 10-fold decrease in transcription in the ppGpp0 strain (Fig. 1A, circles; Laurie et al., 2003, and references therein). Using this genetic system, we also found a large decrease in transcription from the σ54-Po promoter in two independent DksA null mutant strains (Fig. 1A, triangles). When combined, the lack of both ppGpp and DksA essentially abolishes detectable transcription from Po in E. coli (Fig. 1A, diamonds). These results demonstrate the potent effects of both of these regulatory molecules in DmpR-controlled σ54 transcription. The additive negative effect upon loss of both of these regulatory molecules is consistent with the proposed role of DksA in stabilizing the binding of ppGpp to RNAP. However, as DksA can also modulate transcriptional properties in the absence of ppGpp (Paul et al., 2004a; 2005), the ppGpp-independent effects of DksA may also contribute to the net in vivo effect on σ54 transcription.
Western blot analysis revealed that the temporal expression profiles of DmpR and σ54 in wild-type and mutant strains are similar. As shown in Fig. 1B for the wild type, DmpR levels increase approximately twofold over the growth curve while σ54 levels remain constant. We found a previously undetected small decrease in DmpR levels (approximately twofold; Sze and Shingler, 1999) in the strain lacking ppGpp, while the levels σ54 are similar (Fig. 1B, lower). However, coexpression of additional dmpR from a plasmid (pVI899) to provide DmpR levels in the ppGpp0 strain slightly exceeding those in the wild type did not influence the level of dependence on ppGpp (data not shown). Thus, we conclude that ppGpp and DksA exert their action in vivo mainly through a mechanism that is independent of associated alterations in the levels of DmpR or σ54.
Mutants of σ70 suppress the need for ppGpp and DksA for efficient in vivo σ54 transcription
The individual and combined in vivo effects of lack of ppGpp and DksA are consistent with the known synergistic and ppGpp-independent effects of DksA in regulation of σ70 promoters (Paul et al., 2004a; 2005). The genetic system described in the preceding section has been used previously to demonstrate that mutants of σ70 that are defective in competing against σ54 for core RNAP in vitro by-pass the need for ppGpp for efficient σ54-Po transcription in vivo (Laurie et al., 2003). Thus, it was of interest to determine whether σ70 mutants could also suppress the need for DksA for efficient σ54-dependent transcription. To this end, we monitored transcription in strains harbouring the σ70-40Y allele, which has a three-amino-acid insert [∇DSA(536–538)] and exhibits the most extensive defects in in vitro competition assays, and the σ70-35D allele, which harbours a single-amino-acid substitution (Y571H) and shows the least extensive defects in competition assays (Laurie et al., 2003). As shown in Fig. 1C and D, both σ70 mutant proteins result in increased σ54-Po transcription in both ppGpp0 and DksA null strains. The σ70-40Y allele is an efficient suppressor in both cases, restoring transcription in the absence of ppGpp or DksA to approximately 150% and 175% of the levels observed in the wild-type parent respectively. The σ70-35D allele is a comparatively poor suppressor, restoring transcription in the ppGpp0 and DksA null strains to ∼50% and 85% of the level observed in the wild-type parent respectively. These data clearly demonstrate that altered σ70 properties can efficiently compensate for the need for ppGpp or DksA in vivo to restore transcription from the σ54-dependent Po promoter, and are consistent with both of these regulatory molecules mediating their effect through σ70-dependent transcription.
DksA and ppGpp do not have any direct stimulatory effect on reconstituted in vitro σ54 transcription
To assess the potential direct effects of ppGpp and DksA on σ54 transcription in vitro, we used a previously developed multiple-round transcription assay that fully reconstitutes aromatic effector- and ATP-dependent transcription from Po with purified E. coli components (O’Neill et al., 2001). To compare the effects of these two regulatory molecules, we used a template carrying σ54-Po (pVI695) in parallel with assays using templates that carry σ70 promoters previously shown to be either positively regulated by ppGpp/DksA (PthrABC, pRLG5073), or negatively regulated by ppGpp/DksA (rrnB P1, pRLG6214) (Paul et al., 2004a; 2005). As a first step in the analysis, we added increasing concentrations of either ppGpp (0–300 µM) or DksA (0–5 µM) alone into reaction mixtures. Neither ppGpp nor DksA had any major stimulatory effect on σ54-Po transcription (Fig. 2A). With the exception of a previously observed stimulatory effect of DksA alone on transcription from the σ70-PthrABC (Paul et al., 2005), neither ppGpp nor DksA alone had any notable influence on transcription from either σ70-PthrABC or σ70-rrnB P1 when tested in isolation. In contrast, increasing concentrations of ppGpp in the presence of a constant level of DksA (2 µM), or increasing concentrations of DksA in the presence of a constant level of ppGpp (200 µM), both resulted in the anticipated synergistic effects on transcription from σ70 promoters (positive for PthrABC and negative for σ70-rrnB P1; Fig. 2B). However, the simultaneous presence of these two regulatory molecules had no influence on the levels of transcription from the σ54-Po promoter (Fig. 2B). Thus, neither ppGpp and/or DksA directly stimulates transcription from the σ54-Po promoter under in vitro conditions that recapitulate known positive and negative effects on σ70 promoters.
Transcription from σ54 promoters is strictly dependent on binding and hydrolysis of nucleotides by the obligatory transcriptional activator. DmpR is a dedicated (d)ATPase that can use either ATP or dATP efficiently, but not other nucleotides, to activate transcription from the +1G start of the Po promoter (Wikström et al., 2001). The levels of the initiating nucleotide influences the observed effects of ppGpp and DksA at σ70-rRNA promoters, and has been proposed to compete with ppGpp at the active site of RNAP (Jores and Wagner, 2003; Paul et al., 2004a,b). Thus, we considered that the 4 mM ATP used in the in vitro transcription assays might mask potential effects of ppGpp and DksA on transcription from the σ54-Po promoter. To test this possibility, we used multiple-round transcription assays as described under Fig. 2, but with dATP replacing ATP as the regulator nucleotide. The assays used pVI695, which has the native +1G of the σ54-Po promoter, or pVI900, in which an A replaces the +1G. ATP titrations into reaction mixtures remained unchanged in the presence of ppGpp and DksA with both the +1G or +1A template (Fig. 2C). Similar results were obtained when GTP was titrated into reaction mixtures (data not shown). Thus, we conclude that the level of initiating nucleotide has no influence on the absence of effect of these two regulatory molecules on σ54-dependent transcription from the Po promoter in vitro.
Lifetime of competitor-resistant complex at σ54-Po is shortened by ppGpp
The molecular mechanism(s) by which ppGpp and DksA exert their direct effects on transcription from σ70-dependent promoters are still ill-defined. Both these regulatory molecules reduce the lifetime of competitor-resistant open complexes at ppGpp-inhibited and ppGpp-stimulated σ70 promoters alike. However, reduced open complex stability will only have regulatory consequences on transcription from negatively regulated promoters such as rRNA operon promoters that have open complex stability as the rate-limiting step. The potential effects of ppGpp and DksA on competitor-resistant open complex stability at σ54 promoters are unknown. We therefore tested the lifetimes of competitor-resistant complexes at the σ54-Po promoter in the presence and absence of ppGpp/DksA using an assay based on in vitro transcription. In these experiments, open complexes were first accumulated under the same conditions as described above, and then new open complex formation was blocked by addition of competitors. The level of transcript formation at subsequent time points measures the stabilities of the preformed competitor-resistant open complexes. As shown in Fig. 3A, the lifetime of competitor-resistant complexes at σ54-Po lay between those of σ70-rrnB P1 promoter (Fig. 3B) and σ70-PthrABC (Fig. 3C). Addition of ppGpp (200 µM) reduced the lifetime (Fig. 3A). In contrast, addition of DksA (2 µM) had little effect on the lifetimes when added either alone or in combination with ppGpp (Fig. 3A). Thus, ppGpp appears to be more potent than DksA at reducing the lifetime of competitor-resistant complexes at the σ54-Po promoter. However, this reduction in lifetime does not result in reduced transcription from the σ54-Po promoter under the same conditions (Figs 2 and 3), suggesting that open complex stability is not the rate-limiting step for σ54-dependent transcription from this promoter. Taken together with the data described in the preceding sections, these results suggest that potential ppGpp and DksA effects on transcription through open complex formation or stability do not account for the major stimulatory effect of these molecules on σ54-Po transcription observed in vivo (Fig. 1).
DksA and ppGpp do not directly alter σ-factor competition in vitro
Holoenzyme RNAPs utilizing σ70 or σ54 do not show cross-recognition of the distinct promoter classes that they control. Thus, it is highly unlikely that the σ70 mutants compensate for lack of ppGpp or DksA (Fig. 1) by acting directly at the σ54-Po promoter. A more plausible interpretation of these in vivo results would be that the σ70 mutants, which are defective in competing against σ54 for limiting amounts of core RNAP, operate at the level of σ-factor association with core RNAP, by either an active or a passive mechanism (Laurie et al., 2003). Both σ54 and σ70 had similarly high affinity for core RNAP when assessed in isolation; however, σ54 was significantly poorer at out-competing σ70 than the converse in a previously developed multiple-round in vitro transcription competition assay that simultaneously monitors the transcriptional output from the σ54-Po promoter and the σ70-RNA1 promoter of pVI695 (Laurie et al., 2003). Addition of ppGpp alone does not influence competition between σ54 and σ70 in this assay system (Laurie et al., 2003). Here we used this competition assay to test for a possible active role of ppGpp in the presence of DksA in facilitating association of σ54 over that of σ70. The experiment shown in Fig. 4 compares the ability of increasing concentrations of σ54 to compete with 20 nM σ70 for 10 nM core RNAP in the presence or absence of both ppGpp (200 µM) and DksA (2 µM). Although σ54 effectively competed with σ70 to reduce σ70-RNA1 transcription to ∼20% at the highest concentration tested, the addition of ppGpp and DksA did not alter the level of competition observed. Furthermore, addition of ppGpp and DksA did not enhance σ54 transcription over that of σ70 transcription in similar experiments in which components were allowed to associate for 5 min rather than for 2 h as in Fig. 4, nor when fixed concentrations of core RNAP (10 nM) and σ54 (40 nM) were challenged with increasing concentration of σ70 (data not shown). The lack of any direct effects of addition of ppGpp and DksA in σ54 versus σ70in vitro competition assays (Fig. 4) contrasts with the direct effects of ppGpp on competition between σH and σ70 that are apparent in similar assays (Jishage et al., 2002). However, this finding is fully consistent with previous in vivo data on manipulation of the levels or availability of σ54, σS and σ70 in wild-type and ppGpp0 strains, namely that in vivoσ-factor competition, although integral to the efficiency of σ54 transcription, is not directly influenced by the presence or absence of ppGpp in vivo (Laurie et al., 2003). The lack of any detectable direct effect of ppGpp and DksA on in vitro competition between σ70 and σ54 suggests that these two regulatory molecules do not collaborate to alter the binding properties of σ70 or σ54 for core RNAP to favour association of one σ over that of the other.
Hybrid promoters of Po have different affinities for σ54-RNAP
The Po promoter has relatively low affinity for σ54-RNAP (Sze et al., 2001). The effects of ppGpp and DksA on transcription from other σ54-dependent systems that depend on promoters that have different affinities for the holoenzyme have not been tested previously. All the results described above are consistent with a model in which ppGpp and DksA mediate their effects on σ54-dependent transcription in vivo indirectly through the activity or availability of σ54-RNAP holoenzyme. If this were indeed the case, then poorly occupied low-affinity promoters would be more susceptible to loss of these regulatory molecules than high-affinity σ54 promoters which are easy to saturate. These considerations prompted us to generate and test the activities of a series of hybrid σ54 promoters that differed in their innate affinity for σ54-RNAP.
To generate σ54 promoters with different affinities for σ54-RNAP as the test variable, we constructed a template that allows simple replacement of the Po −24, −12 promoter region by DNA linkers specifying the desired sequence as depicted in Fig. 5A. This strategy maintains the test promoters in the same context as the native Po promoter with respect to its integration host factor (IHF) binding site and the upstream activating sequence (UAS) binding sites for the divergently transcribed dmpR gene product, and places the promoters in control of the luciferase luxAB genes. In addition to reconstructing the Po promoter, designated Po/Po, hybrid promoters containing the −33 to +2 regions of σ54 promoters originating from different bacteria were generated. These included: (i) the P. putida-derived Pu promoter which, like Po, appears to have low affinity for σ54-RNAP (Bertoni et al., 1998; Sze et al., 2001), (ii) the Klebsiella pneumoniae-derived nifH promoter and its mutant derivative nifH049, in which substitution of the −17 to −15 CCC by TTT increases both affinity and transcriptional output in vitro and in vivo (Morett and Buck, 1989; Buck and Cannon, 1992) and (iii) the strong E. coli-derived pspA promoter (Weiner et al., 1995) and glnA promoter that form stable closed complexes with σ54-RNAP (Popham et al., 1989).
To directly compare the relative affinities of the hybrid promoters, EcoRI fragments spanning from −171 to +126 of the hybrid promoters were used in a mobility shift assay to assess closed complex formation in the presence of increasing concentrations of σ54-RNAP (Fig. 5B). The relative affinities of the hybrids were found to lie in the order Po/nifH < Po/Pu < Po/Po < Po/nifH049 < Po/glnA < Po/pspA, which is consistent with limited comparative data available on these promoters in their native context (Morett and Buck, 1989; Buck and Cannon, 1992; Sze et al., 2001). In vitro occupancy of the hybrid promoters was also measured by following the formation of productive open transcriptional complexes over time. The data in Fig. 6 show that promoter occupancy kinetics differs such that the higher-affinity promoters rapidly approach plateau values, while the lower-affinity promoters show slower kinetics. Similar results were obtained when DmpR and its aromatic effector (which are required for open complex formation) were omitted until 8 min before each time point (data not shown). The relative order of the slower kinetics for the lower-affinity promoters (Po/nifH slower than Po/Pu slower than Po/Po) was the same as the comparative affinities for σ54-RNAP (Fig. 5B). These results define the relative affinities and hierarchy of promoter occupancies of the hybrid promoters under in vitro conditions (namely, Po/nifH < Po/Pu < Po/Po < Po/nifH049 < Po/glnA and Po/pspA), and suggest that the promoters would be capable of recruiting σ54-RNAP from the available in vivo pool with the same hierarchy.
Hybrid promoters show different dependency on ppGpp and DksA
For in vivo transcription analysis of hybrid promoters, transcriptional reporter plasmids carrying the dmpR-Po/Px–luxAB cassettes were constructed. These plasmids are analogous to pVI466 (dmpR-Po–luxAB) used in Fig. 1, except that the luxAB genes are fused downstream of +2 relative to the transcriptional start. In the reconstructed Po/Po promoter, the only nucleotide changes from the wild-type sequence are limited to −39 to −36 relative to the transcriptional start, which are outside the critical binding sites for DmpR, IHF and σ54-RNAP (Fig. 5A). As shown in Fig. 7A, the Po/Po promoter reproduced the induction profile and dependency on ppGpp and DksA for efficient promoter output, but with a previously noted approximately fivefold higher luciferase activity relative to the +291 fusion transcriptional reporter pVI466 used in Fig. 1 (Sze and Shingler, 1999; Sze et al., 2002).
The temporal expression profiles of all the hybrid promoters showed growth phase-dependent profiles similar to those shown for Po/Po and Po/glnA (Fig. 7A and B), with absolute output levels that differed by 18-fold (Fig. 7C). The powerful high-affinity promoters such as the Po/pspA and Po/glnA hybrids also allowed some level of transcription even in the exponential phase of growth (Fig. 7B, and data not shown). Most importantly, all the Po/Px hybrids were found to require ppGpp and DksA for full transcriptional output to an extent that reflects their affinity for σ54-RNAP in vitro (Fig. 7C). The ratio between peak transcriptional output in the presence or absence of ppGpp or DksA from the low-affinity Po/nifH, Po/Po and Po/Pu promoters was reduced 6- to 10-fold while that from the higher-affinity Po/nifH049, Po/pspA and Po/glnA promoters was only reduced two- to threefold. Thus, while all the promoters require both ppGpp and DksA for maximal output, low-affinity σ54 promoters are markedly more sensitive to the loss of these two regulatory molecules in vivo. As is the case for the native Po promoter, addition of ppGpp and DksA had no significant stimulatory effect on in vitro transcription from these hybrid promoters (Fig. 7D). With the exception of the Po/Pu hybrid, the hierarchy of promoter strengths observed in vivo was the same as that observed in vitro (compare Fig. 7C and D).
Dependence on ppGpp and DksA is observed in the absence of IHF binding capacity
A simple interpretation of the data described above is that the greater sensitivity of the low-affinity promoters to loss of ppGpp and DksA is due to poor occupancy by the σ54-RNAP available. However, optimal localization of the activator via IHF-mediated DNA bending is particularly important for transcriptional initiation from low-affinity σ54 promoters that are rarely occupied by σ54-RNAP (Hoover et al., 1990; Claverie-Martin and Magasanik, 1992; Santero et al., 1992; Carmona et al., 1997). As IHF levels are partially under the control of ppGpp and show an abrupt increase at the exponential-to-stationary phase transition (Aviv et al., 1994; Ditto et al., 1994; Valls et al., 2002), we considered that the influence of ppGpp and DksA on IHF levels might contribute to the differences in dependence of the hybrid promoters on ppGpp/DksA in vivo. To test this possibility, we generated a series of promoter derivatives analogous to Po/Pu, Po/Po, Po/pspA and Po/glnA hybrids that lacked the IHF consensus binding site of Po. This approach, as opposed to the use of an IHF-deficient strain, avoids pleiotropic effects of the loss of IHF, which has global regulatory consequences in vivo (Arfin et al., 2000). Construction of these plasmids involved introduction of a non-native XhoI site at −122 to −117 relative to the transcriptional start, and alterations that disrupt the IHF binding site consensus but maintain the base composition of Po (Fig. 8A). These hybrids were designated xh-Po/Px and xh-Po(-IHF)/Px to indicate these changes.
In vitro transcription assays in the presence and absence of IHF showed the high IHF dependence of the low-affinity xh-Po/Pu hybrid (32- to 35-fold), the intermediate IHF dependence of xh-Po/Po (10- to 12-fold), and the relatively IHF-independent behaviour of high-affinity xh-Po/pspA and xh-Po/glnA hybrids (only approximately threefold stimulation) in vitro (Fig. 8B). As could be anticipated, transcription from cognate derivatives with a disrupted IHF consensus binding site was unaffected by addition of IHF (Fig. 8B). For in vivo transcriptional analysis, we used the same genetic systems as described under Fig. 7 to compare transcriptional output from promoters with or without the IHF binding capacity (Fig. 8C). IHF dependence was assessed as the fold decrease in transcriptional output of xh-Po(-IHF)/Px compared with cognate IHF binding-proficient xh-Po/Px derivatives (Fig. 8C, hatched bars). The results show an approximately fivefold decrease in transcription from the Po/Po promoter and an approximately 2.5- to 3.5-fold decrease from the Po/pspA and Po/glnA promoters in the absence of IHF binding in vivo. It was initially surprising that the activity of the xh-Po/Pu hybrid, which is extremely IHF-dependent in vitro, was unaffected by loss of IHF binding capacity in E. coli (Fig. 8C). However, this phenomenon has also been observed with the native Pu promoter at which the E. coli HU protein, which binds DNA with little or no site specificity, can functionally replace IHF – both in vitro and in vivo (Pérez-Martín and De Lorenzo, 1997; Valls et al., 2002). The action of HU on the Po/Pu promoter probably also underlies the somewhat different transcriptional profile observed in vivo (see below).
To determine the effect of loss of ppGpp and DksA in the absence of any potential influence through modulation of IHF levels, we measured the relative transcriptional output from the hybrids carrying the inactive shuffled IHF binding site [xh-Po(-IHF)/Px derivatives] in ppGpp0 and DksA null strains. Figure 8D shows transcriptional profiles from the test σ54 promoters in the absence of ppGpp and/or DksA relative to that in wild-type E. coli. In the absence of any potential effect through IHF, ppGpp and DksA were still both required for maximal promoter output in all cases, with simultaneous loss of both molecules essentially abolishing detectable transcription (Fig. 8D), as was the case with the IHF binding-proficient Po promoter (Fig. 1A). Lack of ppGpp had somewhat less effect on σ54-dependent transcription using these IHF-independent derivatives as compared with IHF binding-proficient derivatives, while loss of DksA had a similar effect (compare Figs 7A and B and 8D). Importantly, the low-affinity hybrid promoters, Po/Po and Po/Pu, still showed greater dependency on these two regulatory molecules in vivo than high-affinity Po/pspA and Po/glnA counterparts (Fig. 8D). Thus, we conclude that although IHF is required for optimal σ54 transcription in most cases, the major role of ppGpp and DksA in in vivoσ54 transcription is not mediated through their effects on IHF levels.
In this study we have investigated the effects of ppGpp and DksA on σ54-dependent transcription and the consequences of the loss of these two regulatory molecules with respect to regulation of σ54-dependent promoters in E. coli. Both DksA and ppGpp are required for efficient in vivo transcription from the σ54-dependent Po promoter and variants thereof that differ in their innate affinities for σ54-RNAP (Figs 1A and 7). The effects of these regulatory molecules on σ54-dependent transcription are major, with the simultaneous absence of both ppGpp and DksA essentially abolishing detectable transcription from the σ54-Po promoter in vivo (Figs 1 and 8D). However, neither of these regulatory molecules, either alone or in combination, directly stimulated reconstituted in vitroσ54 transcription from six σ54 promoters tested (Figs 2 and 7D). These data clearly suggest that DksA and ppGpp are not required for σ54-dependent transcription per se, but rather that they mainly act in collaboration to mediate their effects in vivo indirectly. Our finding that the in vivo requirement for either ppGpp or DksA for efficient transcription from the σ54-Po promoter can be by-passed in strains expressing mutant σ70 proteins (Fig. 1) provides independent support for this conclusion, and strongly suggests that ppGpp and DksA both affect σ54 transcription via their effects on σ70-dependent transcription. We were unable to detect any alteration in the ability of σ54 and σ70 to compete for core RNAP upon addition of ppGpp and DksA to in vitro competition assays (Fig. 4). The lack of any direct effect of addition of ppGpp and DksA in σ54 versus σ70in vitro competition assays is fully consistent with previous in vivo data that the outcome of manipulation of the levels or availability of σ54, σS or σ70 is not directly influenced by the presence or absence of ppGpp (Laurie et al., 2003). The lack of any evidence that ppGpp and/or DksA play an active role in competition between σ54 and σ70 leads us to propose a model in which these regulatory molecules affect σ54-dependent transcription by a purely passive (indirect) mechanism, as depicted in Fig. 9.
This proposed passive mechanism operates through predicted global regulatory consequences of the negative action of ppGpp and DksA at the seven powerful stringent σ70-rRNA operon promoters of E. coli. While the levels of ppGpp change dramatically, DksA is produced at constant levels during all growth phases and under different growth conditions, and has been proposed to sensitize the transcriptional apparatus to physiological levels of ppGpp (Brown et al., 2002; Paul et al., 2004a). Thus, this passive model primarily operates through changes in ppGpp levels. The seven σ70-rRNA operon promoters sequester approximately 60–70% of the transcriptional machinery during rapid growth of E. coli on rich media (Bremer and Dennis, 1996). Thus, under these conditions in which ppGpp levels are low (Fig. 9A), sequestering of the majority of the transcriptional machinery by the rRNA operons would leave little free core RNAP available for association with the constant levels of σ54 and σ70, leading to low levels of σ54 holenzyme and cognate σ54 promoter occupancy and output. However, under conditions that elicit high levels of ppGpp (Fig. 9B), the dramatic downregulation of transcription from the σ70-rRNA operon promoters in response to the synergistic action of ppGpp and DksA would lead to increased levels of available free core for association with σ-factors. Consequently, σ54 holenzyme levels would increase, leading to enhanced σ54 promoter occupancy. This passive model for the action of ppGpp and DksA would predict that low-affinity promoters that have σ54 holenzyme binding as a rate-limiting step would be more susceptible to loss of these regulatory molecules than their high-affinity counterparts. Here, we have shown that this indeed appears to be the case, as among six σ54 promoters tested, those with lower affinity for σ54 holenzyme were found to exhibit greater sensitivity to loss of ppGpp and DksA in vivo than high-affinity counterparts (Figs 7 and 8). Within this model, the σ70 suppressor mutants that have defects in their ability to compete against σ54 for core RNAP would by-pass the need for a ppGpp/DksA-mediated increase in the pool of core RNAP by simply allowing σ54 to associate with a greater fraction of that available. Consistent with this interpretation, the σ70-40Y mutant that is most severely defective in competing with σ54 is a better suppressor of the lack of both ppGpp and DksA than the σ70-35D mutant that is less severely defective in competing with σ54 (Fig. 1C and D; Laurie et al., 2003).
In vitroσ54 transcription can be reconstituted using just a few basic components, namely σ54 promoter DNA, an active form of the regulator, holoenzyme σ54-RNAP and nucleotides. Lack of ppGpp and/or DksA does not alter the in vivo levels of σ54, and has only a minor effect on the level of DmpR that does not account for their action in vivo (Fig. 1). Although not necessarily essential in vitro, IHF assists promoter output through architectural changes to allow productive contact between the regulator and the σ54 holoenzyme. IHF levels are partially under the control of ppGpp and increase at the transition from exponential to stationary phase of growth (Aviv et al., 1994; Ditto et al., 1994; Valls et al., 2002). Nevertheless, the in vivo requirement for ppGpp and DksA, although slightly different in magnitude, is still clearly manifested at σ54 promoters that lack the capacity to bind IHF (Fig. 8D). Thus, it seems reasonable to propose that passive regulation accounts for a large proportion of the affects of ppGpp and DksA on σ54-dependent transcription. We cannot, however, exclude the possibility that a currently unknown regulator(s) could also contribute to the requirement for ppGpp and DksA for efficient σ54-dependent transcription in vivo. Akin to the discovery of the role of DksA in ppGpp-mediated regulation of σ70 transcription, it is possible that some yet unknown factor could also act in conjunction with ppGpp and DksA to directly control σ54-dependent transcription, or alternatively, could directly affect σ-factor competition for limiting core RNAP. However, only in the latter case could the existence of such an unknown player account for the σ54 promoter affinity-dependent differences in the requirement for ppGpp and DksA.
As depicted in Fig. 9B, the σ70-binding Rsd protein, the levels of which are themselves partially under ppGpp control, may also aid access of σ54 to the available core RNAP by sequestering σ70 from association with core and by actively removing σ70 from the σ70 holoenzyme (Jishage and Ishihama, 1998; Jishage et al., 2001; Ilag et al., 2004; Westblade et al., 2004). While further work is required to directly investigate the participation of Rsd in passive regulation of σ54-dependent transcription, support for its involvement comes from the observation that like underproduction of σ70, overexpression of Rsd both restores and enhances σ54-Po promoter output in ppGpp0 cells to above that found in wild-type E. coli (Laurie et al., 2003). Future analysis of the in vivo occupancy of σ54 promoters under different growth conditions in strains lacking Rsd should help to directly test the comparative importance of the effects of both ppGpp/DksA effects on core RNAP pools and the potential effects of these regulatory molecules through their control of Rsd levels. Irrespective of the relative contributions of different factors that potentially contribute to passive regulation, transcription from even the strongest high-affinity σ54 promoters is still diminished in ppGpp- and DksA-deficient strains (Figs 7C and 8D). This suggests that passive regulation will affect the performance of many σ54 systems in response to nutritional stress and other cues that stimulate ppGpp production. However, as transcription from σ54 promoters requires co-occupancy of the regulator and the holoenzyme, the degree to which ppGpp/DksA-mediated passive regulation of σ54 transcription would affect different σ54 systems will depend on both the affinity of the promoter, and the levels and binding properties of the cognate regulator in each case.
Escherichia coli strains (Table 1) were grown in Luria–Bertani medium (LB; Sambrook et al., 1989) supplemented with the following antibiotics as appropriate for the strain and resident plasmid selection: carbenicillin (Cb, 100 µg ml−1), kanamycin (Km, 50 µg ml−1), spectinomycin (Sp, 50 µg ml−1), tetracycline (Tc, 5 µg ml−1). E. coli DH5 was used for construction and maintenance of plasmids. E. coli strains carrying rpoD mutations linked to aer-3075::Tn10 or the dksA::Tc mutation of TE8114 were generated by P1 transduction using the associated Tc resistance marker in each case (Table 1). DNA sequencing confirmed co-transduction of the rpoD alleles, as previously described (Laurie et al., 2003).
Table 1. Bacterial strains.
Escherichia coli strain
Cm, chloramphenicol; Km, kanamycin; Tc, tetracycline.
The dmpR-Po–luxAB luciferase transcriptional reporter plasmid pVI466 carries the dmpR gene and σ54-Po promoter in their native configuration fused at +291 to the luxAB genes on an RSF1010-based vector (IncQ; 16–20 copies per cell) (Sze et al., 1996). To provide a compatible plasmid expressing additional controllable levels of DmpR, the KpnI site of the polylinker of the SpR pEXT21 vector (IncW; three copies per cell; Dykxhoorn et al., 1996) was converted to an NdeI site via a linker to give pVI898. Subsequent cloning of the dmpR gene as an NdeI to HindIII fragment from pVI399 (Shingler and Moore, 1994) between these sites of pVI898 to give pVI899 places the dmpR gene cloned under the control of the lacI/Ptac promoter of the vector.
Construction of plasmids carrying different promoters was by a common step-wise procedure using standard DNA techniques, and the fidelity of all polymerase chain reaction (PCR)-generated fragments and linker sequences was confirmed by DNA sequence analysis. First, as depicted in Fig. 5A, templates that allowed introduction of the desired promoter sequence by insertion of linkers between a unique NdeI and BamHI sites were assembled on pBluescript-SK (Stratagene). Plasmid pVI769 harbours the native Po upstream region with an NdeI site at −39 to −34 relative to the transcriptional start of Po. Plasmid pVI733 carries the same DNA but with an XhoI site at −122 to −117. Plasmid pVI734, which harbours the Po upstream region with a disrupted IHF binding sequence, was generated by replacement of the XhoI to NdeI fragment of pVI733 with custom-designed linkers. In the next step, linkers with the desired promoter sequences, and designed to regenerate the BamHI site but not the NdeI site (Fig. 5A, right) were introduced into these three templates to generate the pBluescript-based plasmids listed in Table 2. To generate the dmpR–Px/Px–luxAB transcriptional reporter plasmids listed in Table 2, the RSF1010-based broad-host-range vector pVI432 was constructed by introducing HindIII–XhoI–HpaI sites via a linker into the SalI site of pVI397 (Pavel et al., 1994) to provide a PstI–HindIII–XhoI–HpaI–BamHI–SmaI–EcoRI polycloning cassette. The pVI432 vector was used to clone the dmpR–Px/Px–luxAB fusions as HindIII fragments.
Table 2. Reporter and in vitro transcription plasmids carrying hybrid promoter regions.
dmpR–Px/Px–luxAB on pBluescript-SK
dmpR–Px/Px–luxAB on pVI432
Px/Px–luxAB on pVI432
Px/Px on pTE103
Promoters (Px/Px) of the different plasmids contain the indicated combinations of promoter upstream regions (−122 to −40) and promoter regions (−39 to +2) in the context of the dmpR-dmpK intervening region with co-ordinates given relative to the transcriptional start of Po. The prefix xh- denotes the presence of a non-native XhoI site, while (-IHF) indicates disruption of the IHF DNA binding sequence as shown in Fig. 8.
Templates for in vitro transcription studies were generated using pTE103, which contains a strong T7 termination signal downstream of a polycloning site (Elliott and Geiduschek, 1984). Construction of pVI695, which carries DNA spanning the binding sites for DmpR and the Po promoter (−480 to +26) as an EcoRI–BamHI fragment, has been described previously (Laurie et al., 2003). An analogous plasmid, pVI900, with the +1G exchanged for an A, was constructed by first mutating the fragment on pBluescript-SK using the QuickChange™ from Stratagene, and subsequent cloning into pTE103. To generate pTE103-based plasmids carrying hybrid promoters, the Px/Px–luxAB fusions were first cloned as BglII to HindIII fragments between the BamHI and HindIII sites of pVI432 (Table 2). The EcoRI to BamHI fragments spanning the −578 to +2 hybrid promoter regions of the resulting plasmids were then cloned between these sites of pTE103 (Table 2).
Luciferase reporter gene assays
Luciferase assays of the luxAB gene product were performed on cultures grown and assayed at 30°C. To ensure balanced growth, overnight cultures were diluted and grown into exponential phase before a second dilution to an OD600 of 0.05–0.08, and initiation of the experiment by addition of the DmpR effector 2-methylphenol to a final concentration of 0.5 mM. Light emission from 100 µl of whole cells using a 1:2000 dilution of decanal was measured using a PhL Luminometer (Aureon Biosystems). Data points are the average of duplicate determinations from each of two or more independent experiments in which output values varied less than 20%.
Western blot analysis
Crude extracts of cytosolic proteins, sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), electrotransfer and Western blot analysis were essentially as described previously (Shingler and Pavel, 1995). Monoclonal antibodies against E. coliσ54 were obtained from Neoclone (W0005). Polyclonal rabbit antibodies were raised against the N-terminal 232 residues of DmpR and further affinity purified against the same peptide with an SHHHHHH C-terminal fusion by Agrisera, Sweden. Antibody-decorated bands were revealed using Amersham Biosciences PVDF membrane and ECL-Plus reagents as directed by the supplier.
Gel retardation assay
Reactions were carried out at 20°C in 20 µl of glutamate buffer (G-buffer) comprising 20 mM Tris-glutamate, 10 mM Mg-glutamate, 400 mM K-glutamate, 0.1 mg ml−1 BSA and 5% glycerol. EcoRI fragments generated by digestion of DNA amplified using primers 5′-GCGCCGAATTCATTTGCT CAAGCGGCC-3′ and 5′- CCACAGAATTCAGACGCTTTGC CCAG-3′ were end-labelled with [α-32P]-dATP and the Klenow fragment of DNA polymerase as previously described (Sze et al., 2001). Labelled DNA (2 nM) was incubated for 30 min with the indicated concentration of σ54-RNAP reconstituted at a 1:1 molar ratio of core to σ54 for 10 min before addition to the reaction mix. The entire reaction volume was then loaded onto a 4% non-denaturing polyacrylamide gel, electrophoresed, dried. The relative band intensities were quantified using a Molecular Dynamics Phosphorimager.
In vitro transcription assays
Reactions (20 µl) were performed at 20°C in G-buffer (see above), or at 30°C in T-buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 10 mM MgCl2, 1 mM dithiothreitol, 0.1 mM EDTA and 0.275 mg of BSA per ml), as indicated. E. coli core RNAP and σ70-RNAP holoenzyme were obtained from Epicentre. Synthesis and purification of E. coli IHF, σ70, σ54, DmpR-His and ppGpp were as previously described (O’Neill et al., 1998; 2001; Laurie et al., 2003). N-terminally His-tagged E. coli DksA, which is functionally indistinguishable from native DksA in in vivo and in vitro assays (Paul et al., 2004a), was a generous gift from A. Åberg (Umeå University, Sweden).
For σ54-RNAP holoenzyme formation, core RNAP was pre-incubated with an eightfold molar excess of σ54 for 5 min, unless otherwise stated. Holoenzyme RNAPs were incubated with ppGpp and/or DksA (or appropriate storage buffer) for 10 min, or as indicated, before initiation of experiments. Open complex formation (20 min) was initiated by the addition of supercoiled plasmid DNA. For σ54-dependent transcription, reactions were also supplemented at the same time with IHF (10 nM unless otherwise indicated), DmpR-His (50 nM), 4 mM ATP (or dATP required for DmpR activity) and the DmpR aromatic effector 2-methylphenol (0.5 mM). Transcription was initiated by adding 2.5 µl of a mixture of ATP (final concentration, 0.4 mM), GTP and CTP (final concentration 0.2 mM each), UTP (final concentration 0.08 mM) and [α-32P]-UTP (5 µCi at > 3000 Ci mmol−1; Amersham Biosciences). In multiple-round assays, re-initiation was prevented after 7 min by the addition of heparin (0.125 mg ml−1 final concentration) for σ54-dependent transcription assays, or by addition of a 200-fold molar excess of a 60 bp double-stranded DNA fragment containing the full con promoter (Gaal et al., 2001) for σ70-dependent transcription assays. Reactions were further incubated for 3–5 min to allow completion of initiated transcripts. For single-round assays, heparin or competitor DNA was added simultaneously with the nucleotide mix, reactions were incubated for 10 min then terminated by addition of formamide loading buffer, and transcripts were analysed on a 7 M urea-4% acrylamide gel as previously described (Laurie et al., 2003).
In competitor-resistant open complex stability assays, reaction components were incubated in T-buffer at 30°C in the presence or absence of ppGpp (200 µM) and DksA (2 µM) as described for transcription assays, to allow open complex formation. At time zero, heparin or a 200-fold molar excess of competitor full con promoter DNA was used to compete the formation of new complexes. At the indicated times, 20 µl aliquots were taken and single-round transcription was performed as described above.
We are indebted to C.C. Sze and A. Laurie for construction of pVI899 and pVI900, respectively, and to M. Cashel, R.L. Gourse and A. Åberg for fruitful discussions and for generously providing reagents used in this study. L.M.D.B. is a recipient of a student fellowship from the Foundation for Science and Technology (FCT) Portugal. This work was supported by grants from the Swedish Foundation for Strategic Research and the Swedish Research Council.