The alarmone (p)ppGpp mediates physiological-responsive control at the σ54-dependent Po promoter


Victoria Shingler. E-mail; Tel. (+46) 90 7852534; Fax (+46) 90 771420.


Transcription from the Pseudomonas-derived σ54-dependent Po promoter of the dmp operon is mediated by the aromatic-responsive regulator DmpR. However, physiological control is superimposed on this regulatory system causing silencing of the DmpR-mediated transcriptional response in rich media until the transition between exponential and stationary phase is reached. Here, the positive role of the nutritional alarmone (p)ppGpp in DmpR regulation of the Po promoter has been identified and investigated in vivo. Overproduction of (p)ppGpp in a Pseudomonas reporter system was found to allow an immediate transcriptional response under normally non-permissive conditions. Conversely (p)ppGpp-deficient Escherichia coli strains were found to be severely defective in DmpR-mediated transcription, demonstrating the requirement for this metabolic signal. A subset of mutations in the β, β′ and σ70 subunits of RNA polymerase, which confer prototrophy on ppGpp0E. coli, was also found to restore specific DmpR-mediated transcription from Po, suggesting that the metabolic signal is mediated directly through the σ54-RNA polymerase. These data provide a direct mechanistic link between the physiological status of the cell and expression from σ54 promoters.


Eubacterial RNA polymerase (RNAP) holoenzyme is composed of a core enzyme with the subunit structure α2ββ′, associated with one of a number of different sigma (σ) factors required for promoter binding and initiation of transcription. The associated sigma factors programme the holoenzyme to engage different sets of promoters depending on promoter sequence. In Escherichia coli, seven different sigma factors have been identified (reviewed in Gross et al., 1992). With the exception of σ54 (also called σN), all the sigma factors share a notable degree of structural and functional similarity. The σ54 factor, on the other hand, remains in a class of its own, recognizing well-conserved −24, −12 (TGGCAC N5 TTGCa/t) promoter sequences (reviewed by Merrick, 1993).

Promoters of this class direct the transcription of genes involved in a variety of physiological processes in response to environmental signals that lead to modulation of the activity of their cognate regulator (reviewed by Shingler, 1996). Having achieved their active form, the regulators are thought to activate transcription through a common mechanism involving looping out of the intervening DNA between the regulator-bound upstream activator sites (UAS) and the cognate σ54-RNAP-bound promoter (reviewed by Kustu et al., 1991). In some σ54-dependent systems, including the dmp system used here (Sze et al., 1996), this process has been suggested as being assisted by binding of the DNA bending protein integration host factor (IHF), whereas in others, binding of the HU protein or intrinsic bends have been implicated (see Pérez-Martin et al., 1994; Pérez-Martin and de Lorenzo, 1995 and references therein).

Holoenzyme RNAP using the σ54 factor (Eσ54), unlike Eσ70, is isomerization incompetent and therefore incapable of forming open transcriptional complexes in the absence of a positive regulator. The formation of open transcriptional complexes requires an ATPase activity mediated by the conserved central C-domain of the regulators (Weiss et al., 1991; Austin and Dixon, 1992; for a review, see Morett and Segovia, 1993). In some cases, the C-domain has also been shown to mediate oligomerization required before ATPase activity and thus transcriptional activation (Porter et al., 1993; 1995), a process that is facilitated by binding to the UAS sequences via their carboxy-terminal DNA binding domains (Mettke et al., 1995). The unusual property of the formation of stable closed promoter complexes appears to be imposed by a leucine-rich region within the N-terminal of σ54, which keeps the activity of RNAP in check (Wang et al., 1995; 1997a; Syed and Gralla, 1997). Cross-linking studies with DctD have suggested that the activating interaction takes place with both the σ54- and β-subunits of Eσ54 and the conserved C3 region of the regulator (Lee and Hoover, 1995; Wang et al., 1997b). Thus, the regulator C-domain couples ATP hydrolysis with Eσ54 transcriptional initiation. The mechanism underlying this coupling is not fully understood but may involve a process by which ATP binding and hydrolysis mediate conformational changes that allow the regulator to interact successfully with Eσ54, and thereby relieve the repressive effect of σ54 on the ability of core RNAP to form open transcriptional complexes (Pérez-Martin and de Lorenzo, 1996).

The σ54-dependent regulator DmpR is the specific regulator controlling expression of the (methyl)phenol catabolic enzymes encoded by the plasmid-located dmp operon of Pseudomonas sp. strain CF600 (reviewed in Powlowski and Shingler, 1994). The activity of this regulator is directly modulated by binding of the aromatic pathway substrates to its signal-sensing A-domain. Effector binding alleviates interdomain repression, leading to the expression of its otherwise repressed ATPase activity, and thus transcriptional activation (Ng et al., 1996; O′Neill et al., 1998). In addition to the specific DmpR-mediated control mechanism described above, DmpR regulation of Po is subject to modulation by the physiological status of the cell. DmpR expression is constant over the growth curve. However, in rich media, expression of the dmp operon in the presence of effector aromatics is suppressed until the exponential to stationary phase transition (Sze et al., 1996). This level of regulation can be mimicked in both E. coli and Pseudomonas putida by a minimal system comprising the dmpR regulatory gene, the Po promoter and the DmpR binding sites. These findings are consistent with the idea that a global regulatory signal is transmitted via the action of DmpR at the Po promoter and is a host-encoded property common to E. coli and Pseudomonas alike.

One potent mechanism by which promoter-specific transcription is modulated in response to metabolic cues and the growth status of the cell involves the unusual nucleotides guanosine tetraphosphate (ppGpp) and guanosine pentaphosphate (pppGpp), collectively referred to as (p)ppGpp. These nucleotides are thought to function as metabolic alarmones to integrate cellular responses to the various forms of nutritional stress. The levels of (p)ppGpp in E. coli are determined by the action of the RelA (p)ppGpp synthase I and the dual-function SpoT protein mediating both (p)ppGpp 3′-pyrophosphohydrolase and synthase II activities (Xiao et al., 1991). These molecules were originally identified because of their effects on Eσ70 transcription of rRNA operon promoters during the classical stringent response to accumulating uncharged tRNA during amino acid starvation. Since then, however, their involvement in both positive and negative modulation of a continually increasing number of σ70-dependent systems has been reported (see Cashel et al., 1996). Here, we present evidence that (p)ppGpp is the metabolic signal required to allow transcriptional activation from the Po promoter, thus bringing this σ54 promoter, and potentially other σ54 promoters, into the (p)ppGpp regulon. On the basis of in vivo phenotypes of Po transcription with mutant RNA polymerase that compensate for the lack of (p)ppGpp, we propose mechanisms by which these metabolic signal molecules could modulate the action directly of Eσ54 at σ54 promoters.


Media composition determines when DmpR-mediated regulation is observed

The involvement of a nutritional component in the control of DmpR-mediated regulation of the Po promoter prompted us to investigate the role of media composition more extensively. Previous data had shown that, in LB or minimal media containing casamino acids (CAA) as the sole carbon and energy source, transcription from the Po promoter was delayed until the exponential to stationary phase transition, whereas in minimal media containing glucose or organic acids, the transcriptional response from Po was immediate (Sze et al., 1996). These results suggest that the presence of amino acids in the media may be responsible for the delayed response, causing silencing of DmpR-mediated transcription until they had been depleted. In order to test this directly, we used the previously constructed minimal system of P. putida KT2440::dmpR, with a single copy of dmpR locked down on the host chromosome with the aid of a mini-transposon and the Po-luxAB reporter plasmid pVI360. Cultures of KT2440::dmpR (pVI360) were then grown to steady state in test media, and transcription from the Po promoter in the presence of the best DmpR effector, 2-methylphenol, was monitored as luciferase activity. As shown in Fig. 1, growth at the expense of minimal medium containing succinate resulted in a slow growth rate and an immediate response to the presence of the DmpR effector (Fig. 1A), whereas the same medium containing CAA elicited fast growth and a delayed response (Fig. 1B). Similar results were obtained with minimal glucose media in the presence and absence of CAA (Table 1). To determine if this property could be tracked down to a single amino acid, and therefore a particular amino acid-responsive protein such as the leucine-responsive protein Lrp, supplements of different groups of amino acids were tested for their effect. No group of amino acids was found to mediate a delay in the response (Table 1). From this, we infer that it is the general availability of amino acids that elicits the delay.

Figure 1.

. Transcriptional response from Po in KT2440::dmpR (pVI360) grown in the indicated media and assayed for luciferase activity as described in Experimental procedures. Arrows indicate the point at which the DmpR effector was added.

Table 1. . Transcriptional response from Po in KT2440::dmpR (pVI360). a. Concentration of amino acids as in 0.2% casamino acids (CAA) (Dumas et al., 1972).b. Concentration of amino acids as standard amino acid supplements (Experimental procedures).Doubling times were calculated from at least two independent experiments.Thumbnail image of

Artificial manipulation of (p)ppGpp levels overcomes silencing

As the cellular response to a deficiency in amino acids is the accumulation of (p)ppGpp, the simplest interpretation of the growth experiments described above is that low levels of cellular (p)ppGpp are incompatible with efficient transcription from the Po promoter. Cellular levels of these nucleotides can be manipulated in the presence of amino acids by overexpression of the E. coli relA gene product from the Ptac promoter (Schreiber et al., 1991). To test the effect of elevated levels of cellular (p)ppGpp, we used the dual reporter strain KT2440::dmpR-Po-dmpB/luxAB, which carries the entire minimal system in monocopy on the host chromosome. A plasmid carrying relA transcribed from a heterologous promoter completes the system. Using a Ptac-relA plasmid (pVI534), even the uninduced RelA expression caused by the leakiness of the Ptac promoter was found to have severe effects on P. putida growth (data not shown). This suggests that P. putida is very sensitive to (p)ppGpp levels. To overcome this problem and to provide fine control of low-level expression, an equivalent plasmid (pVI536) with relA under the control of the PBAD promoter was constructed. As shown in Fig. 2 (top), PBAD-relA expression in KT2440::dmpR-Po-dmpB/luxAB, had a minimal effect on growth, while allowing an immediate transcriptional response from the Po promoter. Production of (p)ppGpp by RelA is still ribosome context dependent, explaining the hyperexpression from Po as the culture enters the stationary phase, where the frequency with which ribosomes encounter hungry codons increases. Measurement of the total levels of (p)ppGpp (Fig. 2, bottom) indicates that an approximate twofold increase in basal (p)ppGpp levels is sufficient to allow an immediate transcriptional response from Po.

Figure 2.

. Overexpression of RelA and transcription from Po. KT2440::dmpR-Po-dmpB/luxAB harbouring pVI533 (vector control, A) or the PBAD-relA plasmid pVI536 (B) were grown in modified low-phosphate MOPS minimal glucose medium containing total amino acid supplements. Cells were diluted periodically in prewarmed fresh media to extend the exponential phase. Luciferase activity (top) and (p)ppGpp levels (bottom) were determined as described in Experimental procedures. Arrows indicate the point at which arabinose and the DmpR effector was added.

DmpR-mediated regulation of Po is dependent on (p)ppGpp

In view of the finding that (p)ppGpp increases transcription from Po, the very low levels of transcription from Po during exponential growth on rich media suggest that (p)ppGpp is positively required for Po activity. Hence, the DmpR-Po regulatory circuit would be predicted to be defective in a (p)ppGpp0 background. Pseudomonas strains deficient in (p)ppGpp synthesis are not available. However, the physiologically regulated response from Po can be mimicked in E. coli using plasmid pVI466, which carries dmpR in its native configuration relative to Po-luxAB (see Fig. 3; Sze et al., 1996). Therefore, we turned to E. coli (p)ppGpp-deficient strains to address this possibility. Figure 3A shows the normal delayed response from Po exhibited by the growth of wild-type E. coli MG1655 (pVI466) on rich media. In the isogenic ΔrelA derivative CF1652, which still accumulates (p)ppGpp through the action of SpoT (Xiao et al., 1991), the transcriptional profile from Po is the same as in the wild type (data not shown). This is consistent with the high (p)ppGpp sensitivity of the DmpR-Po regulatory circuit shown in Fig. 2. However, in the isogenic ΔrelA, ΔspoT derivative CF1693, which is devoid of all detectable (p)ppGpp, transcription of Po is severely impaired, being reduced to less than 10% of the peak activity in the wild-type counterpart. Qualitatively similar results were obtained with wild-type W3110 and its (p)ppGpp0 counterpart CF1946 (data not shown). In each case, transcription could be restored by minimal expression of RelA from the low-copy-number plasmid pVI535, which carries a promoterless relA gene under the control of a kanamycin resistance promoter (Fig. 3B).

Figure 3.

. Requirement for (p)ppGpp for transcription from Po. E. coli MG1655 (wild type; squares) and its (p)ppGpp0 derivative CF1693 (circles) harbouring pVI466 alone (A) or pVI466 and pVI535 (Pkm-relA) (B). Cells were grown in LB and assayed for luciferase activity as described in Experimental procedures. Arrows indicate the point at which the DmpR effector was added. C. Western analysis of DmpR and σ54 within total soluble proteins (12 μg in lanes 1 and 6 μg in lanes 2) from cells shown in (A) and harvested at OD600 = 2.0.

Modulation of Po activity by (p)ppGpp does not operate at the level of DmpR activation

In view of the findings above that (p)ppGpp is necessary for efficient transcription from Po, we considered a possible indirect mechanism by which this metabolic signal could mediate its effect. One possibility considered was that (p)ppGpp levels may control entry of the aromatic effectors of DmpR into the cell, for example by modulating the expression of a putative aromatic compound uptake system, and thereby control when DmpR-mediated regulation is observed. To test directly if physiological control is mediated at the level of activation of DmpR by its effector, we made use of a deleted derivative of DmpR, ΔA2-DmpR, which lacks the amino-terminal regulatory domain. This derivative is constitutively active both in vivo and in vitro and thus bypasses the need for the aromatic effector activation step (Shingler and Pavel, 1995). Plasmid pVI453 carries ΔA2-dmpR downstream from the IPTG-inducible promoter of the vector. This plasmid was introduced into P. putida KT2440::Po-luxAB carrying the Po–luxAB fusion on the chromosome. The resulting strain was cultured in LB containing 0.15 mM IPTG to produce ΔA2-DmpR at approximately 80% of the level of wild-type DmpR produced from its native promoter at monocopy (Shingler and Pavel, 1995; data not shown). The results in Fig. 4 demonstrate that Po transcription using ΔA2-DmpR is still under physiological control, and thus that the (p)ppGpp signal is not mediated at the level of DmpR activation by aromatic effectors.

Figure 4.

. Transcriptional response in KT2440::Po-luxAB harbouring pVI453 (ΔA2-dmpR ) in LB medium containing 0.15 mM IPTG. The culture was assayed at different times for luciferase activity as described in Experimental procedures.

Modulation of Po activity by (p)ppGpp is not caused by indirect effects on the levels of key players

The occurrence of transcription from Po at the exponential to stationary phase transition could be interpreted as suggesting a direct or indirect role for the stationary phase sigma factor RpoS, whose expression is upregulated in response to (p)ppGpp (Gentry et al., 1993). However, we have previously eliminated this possibility using a null rpoS mutant E. coli strain (Sze et al., 1996). Another obvious indirect mechanism by which the (p)ppGpp signal could act would be at the level of expression of known positive players in the DmpR-Po regulatory circuit. These include DmpR, σ54 itself and IHF. The levels of DmpR and σ54 have been shown to be constant along the growth curve (Cases et al., 1996, Jishage et al., 1996; Sze et al., 1996). In addition, quantitative Western blot analysis of DmpR and σ54 in the wild-type E. coli and its (p)ppGpp0 derivative used here confirm that (p)ppGpp does not mediate variations in the protein levels of these two players (Fig. 3C).

Previously, (p)ppGpp has been shown to be positively involved in regulating the levels of the two IHF subunits, encoded by himD and himA (Aviv et al., 1994). IHF is required for efficient transcription from Po. However, the residual DmpR-mediated transcription (of about 25% of wild-type levels) observed in a himD ::mini-tet derivative is still delayed until the exponential to stationary phase transition (Fig. 5A, Sze et al., 1996). While this observation rules out a role for IHF per se, it raises the possibility that the almost complete obliteration of Po transcription in (p)ppGpp0 backgrounds is in part the result of downregulation of IHF. To test this possibility, we constructed a plasmid expressing himD-himA from the Ptac promoter. Expression of IHF from this plasmid was found to restore Po transcription fully from the minimal system carried on pVI466 in DPB101 IHF strain (Fig. 5A), while having no effect on the levels observed in the (p)ppGpp0CF1693 background (Fig. 5B). Similar data were obtained with CF1946 (data not shown). These results demonstrate that the severe defect seen in (p)ppGpp0 backgrounds is not compounded by IHF deficiency. Hence, we conclude that (p)ppGpp either modulates the expression of some as yet unidentified factor involved in Po activity or exerts its action directly through its effects on the transcriptional machinery.

Figure 5.

. A. Effect of IHF on transcription from Po. E. coli S90C (IHF+) and its isogenic IHF counterpart DPB101, each harbouring pVI466 and the vector pEXT21 or pVI537 (carrying IHF genes himA-himD ), were grown in LB and assayed for luciferase activity as described in Experimental procedures. B. As (A) but using MG1655 and its isogenic (p)ppGpp0 counterpart. Arrows indicate the point at which the DmpR effector was added.

Po activity is restored by mutations within the β, β′ and σ70 subunits of RNA polymerase

Recent data has demonstrated that ppGpp binds directly to RNAP in the C-terminal domain of the β-subunit (Chatterji et al., 1998). Genetic analysis, while predominantly implicating this subunit, has also implicated the β′- and σ70-subunits, suggesting a role for these subunits in transducing the ppGpp binding signal. Over 100 independent mutations that suppress survival defects and/or the inability of (p)ppGpp0-E. coli strains to grow on minimal glucose media have been mapped to rpoB, rpoC and rpoD (Hernandez and Cashel, 1995; Cashel et al., 1996). Seventeen of these mutant derivatives, which all confer prototrophy, have been tested here for their ability to restore DmpR-mediated regulation of the Po promoter in the absence of (p)ppGpp (Table 2). The seven rpoB (β) alleles were chosen as the strongest suppressors of the growth defect, the eight rpoC (β′) alleles represent the spread of mutant clusters found within this gene and the two rpoD70) alleles are the only two found out of six independent mutations isolated (M. Cashel, personal communication; Hernandez and Cashel, 1995). The minimal DmpR-Po reporter plasmid pVI466 was introduced into (p)ppGpp0 derivatives harbouring each of the mutant alleles, and the resulting strains were monitored for luciferase activity during growth on LB. Nine alleles were found to restore detectable Po activity (shown in Fig. 6); the remaining derivatives did not restore activity above basal levels (data not shown). Five out of seven mutants in rpoB were found to restore Po activity even during exponential growth (Fig. 6A and B; note the change of scale for the hypersuppressors in Fig. 6A). Two of the eight rpoC mutants restored expression (Fig. 6C): rpoC-(215–220)Δ in a growth phase-independent manner similar to that of rpoB mutants, while the rpoC-R1148C allele, like the two rpoD alleles (Fig. 6D), showed a marked growth phase dependency in their ability to restore expression from Po. The finding that only a subset of the mutants tested could restore Po activity suggests that prototrophy can be restored by at least two different mechanisms, only one of which is shared in the restoration of transcription at Po. As expanded on in the Discussion, we interpret the different behaviours of restorative mutants of transcription from Po as reflecting different facets of (p)ppGpp modulation of RNAP.

Figure 6.

. Restoration of Po activity by mutations within RNAP. E. coli (p)ppGpp0 strains with the indicated alleles of RNAP and carrying pVI466 were grown in LB and assayed for luciferase activity as described in Experimental procedures. Dotted lines indicate the level observed in an isogenic strain with wild-type RNAP. Arrows indicate the point at which the DmpR effector was added.


Bacteria have to respond to specific signals in their environment with a rapid adaptive transcriptional response. However, the appropriate response has to incorporate considerations relating to the physiological status of the cell. In the case of the Pseudomonas-derived DmpR-Po σ54-dependent regulatory system, the specific regulatory circuit is silenced in rich media until amino acids become limiting (Fig. 1; Sze et al., 1996). In addition to Po, a number of other σ54 promoters have been reported to be growth phase regulated in rich media in an analogous manner, including two promoters from the toluene/xylene catabolic TOL plasmid, Pu and Ps (Hugouvieux-Cotte-Pattat et al., 1990; Marqués et al., 1994; Cases et al., 1996), the PnifH promoter involved in nitrogen assimilation (Cases et al., 1996), the DctD-controlled promoter involved in dicarboxylic acid transport (Gu et al., 1994) and a promoter controlling a carbon starvation survival gene (Kim et al., 1995). Here, we have sought to understand how global regulatory events dominate over the specific regulatory mechanism to achieve such an effect.

Using overexpression of RelA (Fig. 2) and specific E. coli mutants (Fig. 3), we identified (p)ppGpp as the regulatory signal required to allow transcription from Po. One consequence of lack of transcription from Po in the absence of (p)ppGpp, i.e. in rich media eliciting fast growth, would be effectively to block expression of the specialized (methyl) phenol catabolic enzymes when other more favourable sources of energy are available. The inverse linear relationship between growth rates and the intracellular concentration of (p)ppGpp provides a simple interpretation of the previous finding of inverse correlation between Po transcription levels and growth rates (Sze et al., 1996). A positive role for (p)ppGpp may also, at least in part, explain catabolite repression on aromatic degradation observed with the σ54-dependent TOL toluene/xylene pathway (Duetz et al., 1994; Holtel et al., 1994) and the σ54-dependent phenol catabolic pathway of P. putida H (Müller et al., 1996). Although the generality of involvement of this metabolic signal has to be verified, it is our expectation that other σ54 promoters reported to control specialized catabolic functions of soil and water microorganisms will fall into the (p)ppGpp regulon, fulfilling the same function as glucose repression in enterics, namely causing silencing of energetically less favourable specialized catabolic functions until needed.

A subset of mutations that restore prototrophy to (p)ppGpp-deficient strains is able to restore transcription from Po in the absence of (p)ppGpp (Fig. 6). While mutations in RNAP are, by definition, likely to have pleiotropic effects, six of the mutants that restore Po activity have been analysed in terms of in vivo and in vitro properties at σ70 promoters (Hernandez and Cashel, 1995; Gaal et al., 1997; Bartlett et al., 1998; Zhou and Jin, 1998). In each case, the mutations appear to partially or completely mimic the effect of ppGpp binding to the β-subunit in vivo and to alter the kinetics of transcriptional inactivation in vitro, providing strong evidence that ppGpp exerts its effect directly through modulation of Eσ70 function. A role for (p)ppGpp in modulating the levels of known factors participating in the formation of the DmpR/Eσ54 complex at Po has been eliminated (Fig. 3C and 5; Cases et al., 1996; Sze et al., 1996). This leads us to conclude that the (p)ppGpp signal either modulates the level of an unknown component or exerts its effect directly through the modulation of Eσ54 function at Po. While we cannot eliminate the former possibility as yet, the mutant analysis shown in Fig. 6 is consistent with the latter. Therefore, we interpret the results in terms of the modulation of Eσ54 activity below.

The kinetics of restoration of Po activity over the growth curve differs depending on the specific RNAP mutation, suggesting two related facets of ppGpp modulation through its action on the transcriptional machinery. The first facet involves competition between sigma factors for limiting core RNAP. Competition between sigma factors that affected Po transcription as the cell enters stationary phase has been suggested previously to explain the hyperinduction of transcription from Po observed in an RpoS-deficient strain (Sze et al., 1996). A similar phenomenon observed at the σ70uspA promoter has recently been investigated directly, and the data suggest that passive regulation at the level of sigma factor competition for core can have marked effects on gene expression levels (Farewell et al., 1998). Sigma factor competition is also suggested by the phenotype of the two σ70 alleles (rpoD-P504L and S506F), which restore Po activity to higher than wild-type levels but still show a marked growth phase dependency (Fig. 6D). It may at first seem surprising that a mutation in σ70 can restore transcription from a strictly σ54-dependent promoter. However, previous work on these mutants provides a plausible mechanistic explanation. These two mutations map in the conserved region 3.1 of sigma70-like proteins, which has been implicated by work on rpoH encoding σ32 as being important in sigma–core RNAP interaction (Zhou et al., 1992). Similarly, both RpoD-P504L and S506F decrease the cellular fraction of σ70-associated holoenzyme, a situation that is exacerbated by the presence of ppGpp (Hernandez and Cashel, 1995). Hence, ppGpp appears to influence the affinity of core RNAP for σ70 negatively. The high levels of ppGpp present during the exponential to stationary phase transition may thus influence competition between sigma factors in favour of σ54. The growth phase dependency of restoration of transcription from Po in the absence of (p)ppGpp by the RpoD mutant proteins may reflect their mimicry of the action of ppGpp by favouring the association of core with σ54 over that with σ70. This effect may be further enhanced by the accumulative action of factors that modulate the activity/availability of sigma factors. For example, the anti-sigma factor protein Rsd, which is expressed at the exponential to stationary phase transition and specifically binds σ70 (Jishage and Ishihama, 1998), and regulation of a putative anti-σ54 factor by the FtsH protease, which has recently been shown to be essential for σ54-mediated transcription (Carmona and de Lorenzo, 1998).

A second facet of the mechanism by which ppGpp may exert its physiological control on σ54 promoters is suggested by the ability of RpoB and RpoC alleles to restore Po activity (Fig. 6A–C). In vivo and in vitro analysis of three of the RpoB (A532Δ, L533P and T563P) and one of the RpoC derivatives (215–220Δ) has been performed using both ppGpp-regulated (stringent) and ppGpp-independent σ70 promoters (Gaal et al., 1997; Bartlett et al., 1998; Zhou and Jin, 1998). In each case, the mutations have been shown to destabilize the Eσ70–promoter open complex, and complementary models have been proposed to explain how the unifying theme of destabilization can co-ordinately decrease and increase transcription depending on the kinetic properties of the specific σ70 promoter. All four of these derivatives are shown here to restore activity of the σ54 promoter Po even in the exponential growth phase. As outlined in the Introduction, the events taking place at the −24,–12 promoters with Eσ54 are different from those taking place at the −35,–10 promoters. Nevertheless, these results suggest that conformational changes mediated by ppGpp binding may also modulate the stability of σ54–promoter complexes, leading to enhanced transcription. The mechanism involving ppGpp-mediated modulation of Eσ54–DNA interactions or the conformational enhancement of regulator–Eσ54 interaction can be envisaged.

Overexpression of DmpR in Pseudomonas and in E. coli alleviates the delay imposed on DmpR-mediated transcription from Po seen in amino acid-rich media and promotes immediate and high-level transcription from the Po promoter (Sze et al., 1996; C. C. Sze and V. Shingler unpublished). The identification of (p)ppGpp as the metabolic signal required to allow activated DmpR to mediate its regulatory role at the Po promoter suggests that overexpression of DmpR overcomes the need for the ppGpp-mediated allosteric change in RNAP. DmpR is naturally made at very low levels in the cell and, under these conditions, the presence of its DNA binding sites are necessary for the activation of transcription from Po (Sze et al., 1996). It is possible that increased levels of DmpR lead to efficient multimeric occupancy of its binding sites and/or sequestering of σ54 or limiting Eσ54 to the promoter, thus leading to an increased chance of productive interaction with RNAP and transcriptional activation in the absence of (p)ppGpp. However, a common feature of σ54-dependent regulators is that, while they normally act in cis (i.e. bound to the DNA), at high concentration they can also function in trans (from solution), suggesting that high regulator concentrations eliminate two of the functions of DNA binding, namely facilitating oligomerization and concentration of the regulator near the RNAP. At high concentrations of DmpR, its ATPase activity required for transcriptional activation is independent of DNA binding (C. C. Sze and E. O'Neill, unpublished). Hence, upon overexpression, DmpR may also activate Eσ54 from solution. This raises the interesting possibility that physiological-responsive control via ppGpp-Eσ54 is restricted to the DNA-bound regulator. In such a scenario, ppGpp may be required to overcome constraints on the DmpR–Eσ54 interaction imposed by DNA binding, while the interaction may occur freely with the regulator from solution.

The present study has identified (p)ppGpp as an essential positive modulator of transcription from the Po σ54-dependent promoter and elucidated potential and testable mechanisms by which (p)ppGpp may modulate Eσ54 to link transcription from σ54-dependent promoters to the physiological status of the cell. Evaluation of the mechanistically different, and possibly synergistic, facets by which ppGpp may modulate the level and function of Eσ54 is a future goal of our laboratory. Previous work has implicated (p)ppGpp as having a direct or indirect role in transcription from the σ54-nif operon of Klebsiella pneumoniae (Riesenberg et al., 1982). It is our prediction that many σ54 promoters are potential targets for (p)ppGpp-mediated regulation, but that the extent to which this is manifested will depend on the innate ability of its cognate regulator to interact with Eσ54 and thereby release the repression of σ54 on core RNAP.

Experimental procedures

Bacterial strains and plasmids

The bacterial strains and plasmids used are listed in Table 2. Plasmids were constructed using standard recombinant DNA techniques (Sambrook et al., 1989) and were introduced into E. coli strains by transformation and into P. putida strains by electroporation using a Bio-Rad gene pulser.

Plasmid pVI365 was constructed by insertion of a KpnI–EcoRI–KpnI–SspI–HpaI–BamHI–HindIII linker into the unique HindIII site within the kanamycin resistance gene of pRK2501-E. The RSF1010-based plasmid pVI533 was constructed by replacing the HpaI to EcoRI fragment of pMMB66EH, spanning lacI q and Ptac, with a polymerase chain reaction (PCR)-amplified araC-PBAD fragment of pBAD/Myc-His A. The EcoRI to HindIII promoterless relA fragment from pALS10 was inserted into the multiple cloning sites of pMMB66EH, pVI365 and pVI533 to generate pVI534, pVI535 and pVI536 respectively. Plasmid pVI537 was constructed by cloning the SacI to BamHI (himD–himA) fragments of pHX3-8 into pEXT21.

Growth and culture conditions

All strains were grown at 30°C in Luria broth (LB), unless otherwise stated. Minimal M9 salts (Sambrook et al., 1989) were supplemented with carbon sources in the form of 0.2% casamino acids, 10 mM succinate or 10 mM glucose. Modified low-phosphate MOPS minimal medium (Neidhardt et al., 1974) was supplemented with 20 mM glucose. Amino acid supplements to these minimal media are as indicated in Table 1 or as standard supplements of Ala (0.00712%), Arg (0.00696%), Asn (0.00528%), Asp (0.00532%), Cys (0.0012%), Gln (0.00876%), Glu (0.00882%), Gly (0.006%), His (0.0031%), Ile (0.00524%), Leu (0.0105%), Lys (0.00584%), Met (0.00298%), Phe (0.00648%), Pro (0.0046%), Ser (0.1052%), Thr (0.00476%), Trp (0.00014%), Tyr (0.00362%) and Val (0.00704).

Before all assays or measurements, cells were grown overnight in the indicated media containing appropriate antibiotics for plasmid selection. To ensure balanced growth, cells from overnight cultures were diluted 1:500, grown into exponential phase and then diluted once more before initiation of the experiment. When needed, the effector 2-methylphenol was added to a final concentration of 2 mM for P. putida or 0.5 mM for E. coli, at the point when an experiment was initiated.

Luciferase assays

Luciferase assays of the luxAB gene product were performed on whole cells, using a 1:2000 dilution of decanal substrate, as described previously (Sze et al., 1996).

(p)ppGpp assays

Cells to be assayed for intracellular (p)ppGpp levels were grown in MOPS supplemented with glucose and all 20 amino acids. Aliquots of the parent culture were taken at the indicated intervals and grown further in the presence of 100 μCi ml−132P (Amersham) for two doubling times. Cells were then collected by centrifugation, resuspended in ice-cold 1 M formic acid and treated as described by Bochner and Ames (1982). Samples were spotted on polyethyleneimine cellulose plates (Merck), fixed by methanol and dried. Ascending chromatography in 1.5 M KH2PO4 buffer, pH 3.4, allowed separation of the nucleotide species. Radioactivity was quantified using a Molecular Dynamics PhosphorImager. The sum of ppGpp and pppGpp arbitrary units is expressed as a ratio of GTP or cell density (A600).

Western blot analysis

Crude extracts of cytosolic proteins, SDS–PAGE, transfer to nitrocellulose filters and Western blot analysis with polyclonal rabbit anti-DmpR or anti-σ54 sera were as described previously (Shingler and Pavel, 1995). Antibody-decorated bands were revealed using Amersham's chemiluminescence reagents as directed by the supplier. Differences in expression levels were assessed by comparison with dilution series of the samples. Specificity of antisera was monitored using purified proteins and genetic control strains proficient and deficient in the expression of each gene product (data not shown).


We wish to thank Dr M. Cashel for providing numerous unpublished strains, Dr V. de Lorenzo for providing information before publication, and Drs R. Dixon, S. Austin and M. Buck for σ54 strains, antisera and purified protein respectively. We thank both Dr T. Nyström and Dr M. Gullberg for fruitful discussion and Dr Gullberg for critical reading of the manuscript. This work was supported by grants from the Swedish Research Councils for Natural and Engineering Sciences, the Swedish Foundation for Strategic Research and the J. C. Kempe Foundation.