One of the major signalling pathways responsible for intercompartmental communication between the cell envelope and cytoplasm in Escherichia coli is mediated by the alternative sigma factor, σE. σE has been studied primarily for its role in response to the misfolding of outer membrane porins. This response is essentially reactionary; cells are stressed, porin folding is disrupted, and the response is activated. σE can also be activated following starvation for a variety of nutrients by the alarmone ppGpp. This response is proactive, as σE is activated in the absence of any obvious damage to the cell envelope sensed by the stress signalling pathway. Here we examine the mechanism of regulation of σE by ppGpp. ppGpp has been proposed to activate at least two alternative sigma factors, σN and σS, indirectly by altering the competition for core RNA polymerase between the alternative sigma factors and the housekeeping sigma factor, σ70. In vivo experiments with σE are consistent with this model. However, ppGpp and its cofactor DksA can also activate transcription by EσEin vitro, suggesting that the effects of ppGpp on σE activity are both direct and indirect.
One of the hallmarks of bacteria is their remarkable ability to adapt to different environmental conditions and survive a wide range of cellular stresses. This ability relies on a sophisticated array of stress responses, many of which work by altering transcription so that genes required to combat a particular stress are induced, and those that may be deleterious under the stress conditions are repressed. These stress responses allow the cell to remodel its physiology to meet whatever conditions are encountered.
The bacterial RNA polymerase (RNAP) has five subunits, α, β, β′, ω and σ. The catalytic core of the enzyme (E) consists of the ββ'α2ω subunits and is capable of transcription elongation and termination. The sigma subunit is required for promoter binding and specific transcription initiation. All bacteria have a primary sigma subunit, known as σ70 in Escherichia coli, which directs the bulk of cellular transcription. Most bacteria also have one or more alternative sigma factors that direct transcription of specific subsets of genes. In most cases, transcription in bacteria is regulated in response to environmental and cellular cues by repressors or activators, proteins that bind to DNA near specific promoters and modulate the activity of RNAP at that promoter, or by alternative sigma factors, proteins that reprogram gene expression by replacing σ70 and redirecting RNAP to promoters specific for that particular sigma factor. E. coli has six alternative sigma factors, σS, σH, σN, σE, σF and σFecI, which each respond to different cellular stresses (Gruber and Gross, 2003). These two types of transcriptional controls are not mutually exclusive. Repressors or activators can regulate transcription by holoenzymes containing alternative sigma factors, thereby integrating multiple environmental signals at a particular promoter to modulate gene expression appropriately.
Alternative sigma factors are regulated by dedicated signal transduction pathways that are activated by particular stresses, conditions, or developmental programs. For example, in E. coli the extracytoplasmic stress factor σE is activated by degradation of the σE-specific anti-sigma factor RseA in response to conditions that interfere with the folding of outer membrane porin proteins (Alba et al., 2002; Kanehara et al., 2002). The activity of the nitrogen-responsive sigma factor, σN, is regulated by the signal-dependent activation of enhancer-binding proteins, which are required for transcription initiation by EσN (Shingler, 1996). Some alternative sigma factors can respond to several different stresses and are regulated by multiple regulators, utilizing a variety of mechanisms. A classic example is σS, whose activity is modulated in response to an assortment of stresses including osmotic shock, acid stress, cold shock, and entry into stationary phase, by regulators that affect transcription of the rpoS gene, translation of the rpoS mRNA, stability of the σS protein, and activity of the σS protein (Hengge-Aronis, 2002). Regulation of stress responses by these signalling pathways is essentially reactionary; damage or stress occurs, is sensed, and then the cell responds.
Reactionary stress responses, such as those described above, are not always the most effective way to respond to stress. When nutrients are scarce, the cell may not have the resources to rapidly mount a response that requires the energy-consuming processes of transcription and translation. In this case, a more general alarm that activates the individual responses in tandem could effectively preload the cell with stress factors, allowing it to combat stresses should they arise. In E. coli the alarmone ppGpp fulfils this role, as it can activate several alternative sigma factors independently of their dedicated signalling pathways (Magnusson et al., 2005; Costanzo and Ades, 2006).
ppGpp is well known as a general signal of starvation stress (Cashel et al., 1996). The level of ppGpp in the cell is determined by the relative activities of the ppGpp synthase RelA and the bifunctional synthase/degradase SpoT (Xiao et al., 1991). The best-studied cellular role of ppGpp is its involvement in balancing the protein synthetic capacity of the cell with nutrient availability. ppGpp levels rise during nutrient downshifts caused by either abrupt changes in media conditions, such as starvation for amino acids, carbon, phosphate, or nitrogen, or upon more complex growth-limiting conditions, such as entry into stationary phase (Cashel et al., 1996). Under steady-state growth conditions ppGpp levels are inversely correlated with the growth rate of the culture, lower in rich media and higher in nutrient-poor media (Ryals et al., 1982).
ppGpp, in conjunction with a cofactor DksA, can regulate transcription both negatively and positively (Paul et al., 2004a; 2005; Mallik et al., 2006). Unlike more conventional activators and repressors that regulate promoter activity by binding to sites on the DNA and contacting RNAP, ppGpp and DksA bind only to RNAP and modulate its activity directly (Paul et al., 2004a; Perederina et al., 2004). ppGpp and DksA reduce transcription of ribosomal RNA genes by Eσ70 and therefore the number of ribosomes in the cell (Paul et al., 2004a,b). In addition to inhibiting rRNA synthesis, ppGpp and DksA activate transcription of several genes by Eσ70, including a subset of the genes encoding enzymes required for amino acid biosynthesis (Paul et al., 2005). The ω subunit of RNAP, encoded by the rpoZ gene, also contributes to the regulation of transcription by ppGpp. RNAP lacking the ω subunit cannot respond to ppGpp in vitro (Vrentas et al., 2005). ω can also affect transcriptional regulation by ppGpp under some conditions in vivo (Rutherford et al., 2007).
The extracytoplasmic stress factor, σE, is activated during entry into stationary phase and this activation is dependent upon ppGpp, not the dedicated cell envelope stress-sensing pathway (Costanzo and Ades, 2006). In this paper, we further investigate the correlation between σE activity and ppGpp levels, and the mechanism of regulation of σE-dependent transcription by ppGpp. In theory, the model that ppGpp alters the competition among sigma factors for RNAP could be sufficient to explain activation of σE by ppGpp. However, in vitro experiments demonstrate that ppGpp and DksA can also directly activate EσE-dependent transcription. Therefore, we propose that the positive regulation of σE-dependent promoters by ppGpp and DksA has two components: direct regulation of σE-dependent transcription and indirect regulation by increasing the amount of EσE in the cell through negative regulation of transcription of rRNA promoters. This is the first reported example of direct activation of alternative sigma factor-dependent transcription by ppGpp and DksA.
ppGpp regulates σE activity
Previous work demonstrated that the activity of the alternative sigma factor, σE, was regulated with respect to growth phase (Costanzo and Ades, 2006). In wild-type cells growing in rich media, σE activity can be divided into three parts. σE activity is low in early exponential phase (OD600 < 0.3), then increases fourfold during mid-exponential phase (OD600 from 0.4 to 1.8) and 19-fold during entry into stationary phase (OD600 > 1.8) compared with its activity in early exponential phase. The low activity of σE during early exponential phase is not due to recovery from stationary phase and is likely to be a function of growth in fresh media (Costanzo and Ades, 2006). The increase in activity during entry into stationary phase required the alarmone ppGpp, suggesting that ppGpp is a positive regulator of σE-dependent promoters. Because these experiments were performed with cultures grown in a rich medium, Luria–Bertani (LB) broth, and entry into stationary phase is complex in rich medium owing to both depletion of nutrients and accumulation of secondary metabolites, all of which can affect gene regulation; it is possible that signals in addition to ppGpp are required for regulation of σE activity. To further explore the connection between ppGpp and σE activity, we examined σE activity under several conditions in which ppGpp levels are known to increase.
To determine if σE activity increased during entry into stationary phase under more defined conditions caused by the depletion of a specific nutrient, σE activity was monitored in cultures grown in media containing limiting concentrations of glucose (0.02%) or phosphate (0.13 mM). ppGpp levels are known to increase when cells enter stationary phase as either nutrient is depleted (Spira et al., 1995; Cashel et al., 1996). Transcription by EσE was assayed by measuring β-galactosidase activity produced from a chromosomally encoded reporter in which the σE-dependent rpoH P3 promoter directs transcription of the lacZ gene (Mecsas et al., 1993). This reporter contains the σE-dependent promoter from the rpoH gene and has been used extensively to monitor σE activity under a wide range of conditions in a variety of strain backgrounds. σE activity increased when cell growth slowed owing to glucose or phosphate depletion, indicating that additional signals accompanying entry into stationary phase in rich media are not required for the regulation of σE activity (Fig. 1A and B). The increase in σE activity following depletion of a specific nutrient is not a property unique to the rpoH P3 promoter. The σE-dependent fkpA promoter, previously shown to be activated by ppGpp during entry into stationary phase in rich media (Costanzo and Ades, 2006), was activated during entry into stationary phase following phosphate depletion as well (data not shown).
The cellular level of ppGpp in exponential phase varies inversely with the growth rate of a culture, such that ppGpp levels are high in nutrient-poor media supporting slow growth rates, and low in nutrient-rich media supporting high growth rates. Many genes that are subject to regulation by ppGpp also show growth rate-dependent regulation (Cashel et al., 1996; Sze and Shingler, 1999; Paul et al., 2004b). We monitored transcription by EσE during early exponential phase in cultures grown in media that support different growth rates. σE activity was regulated with respect to growth rate, increasing as the growth rate decreased (Fig. 1C).
The above experiments demonstrate that under conditions in which ppGpp levels are known to increase, σE activity also increases. However, each of these conditions is accompanied by physiological adaptations to changes in nutrient availability, in addition to the production of ppGpp. This raises the possibility that EσE could be sensing another signal that is coincident with nutrient depletion or culture conditions. If ppGpp were sufficient to regulate transcription by EσE by itself, then σE activity should increase when ppGpp is made during exponential phase in rich medium. To test this hypothesis a truncated variant of the relA gene that constitutively produces ppGpp without associating with ribosomes was overexpressed from the plasmid pALS13 (Svitil et al., 1993). When this variant relA gene was overexpressed, σE activity increased (Fig. 1D). σE activity did not increase upon overexpression of a catalytically inactive variant of the truncated RelA protein that cannot synthesize ppGpp (Fig. 1D). These results demonstrate that additional signals accompanying nutrient depletion are not required for induction of σE activity by ppGpp. Because cell growth slows when ppGpp production increases, we cannot formally eliminate the possibility that σE activity increases owing to a separate event associated with the transition to slower growth.
If ppGpp is required for increasing σE activity, then eliminating the ability of RNAP to respond to ppGpp should eliminate the increase in σE activity observed in cells that can make ppGpp. As both DksA and ω have been shown to affect the ability of ppGpp to function in positive and negative regulation in vitro and in vivo (Vrentas et al., 2005; Rutherford et al., 2007), we constructed a ΔdksAΔrpoZ strain and tested the effects on transcription by EσE. Growth phase-dependent regulation of σE activity was disrupted in the ΔdksAΔrpoZ strain. Transcription by ΕσE in this strain was low throughout the growth curve, increasing only slightly between exponential and stationary phase (Fig. 2A and C).
σE activity was altered in the single-mutant ΔrpoZ strain, suggesting a role for ω in σE-dependent transcription. In wild-type cells, σE activity increased fourfold from early exponential phase to mid-exponential phase, while in the ΔrpoZ strain σE activity only increased twofold at this point in the growth curve (Fig. 2A and C). The regulator responsible for the increase in σE activity between early and mid-exponential phase has not been identified, but these results suggest that ω contributes to this regulation. This is one of the few phenotypes that have been detected from deletion of the rpoZ gene. It is possible that the effects of the rpoZ deletion are in part due to polar effects on the spoT gene immediately downstream. However, the rpoZ allele used in this study has only a weak effect on SpoT expression (Gentry et al., 1991).
ppGpp affects the activity, not production of σE
How does ppGpp regulate σE activity? ppGpp could affect the production of σE, transcription by EσE, or both production and activity. If ppGpp activates transcription by EσE by increasing the production of σE itself, then the level of σE in the cell during entry into stationary phase should increase concomitantly with the increase in EσE activity, and the increase in σE levels should be dependent on ppGpp. To address this issue, σE levels were examined by Western blotting in wild-type strains and in strains unable to produce ppGpp (ppGppo) owing to deletion of the relA and spoT genes. σE levels were similar in both the wild-type and ppGppo strains (Fig. 3). Furthermore, the levels of σE did not increase following gratuitous production of ppGpp owing to overexpression of the constitutive relA variant from pALS13 (data not shown). Therefore, the ppGpp-dependent increase in σE activity is not caused by an increase in the overall amount of σE in the cell.
To further demonstrate that ppGpp regulates transcription by EσE and not the transcription of the rpoE gene, we constructed a strain in which the chromosomal copy of the rpoE operon (rpoE, rseA, rseB and rseC) was deleted and the rpoE gene was expressed from a plasmid under the control of the σ70-dependent pTrc promoter. The pTrc promoter is repressed by the Lac repressor, but is somewhat leaky. The uninduced level of σE expression was sufficient to maintain cell viability (rpoE is essential). σE activity was higher in this strain than in a wild-type strain, because the anti-sigma factor, rseA, was deleted as part of the rpoE operon. However, σE activity was still regulated with respect to growth phase and changes in σE levels cannot account for this regulation, indicating that transcriptional regulation of the rpoE operon promoter is not required for the growth phase-dependent increase in transcription by EσE (data not shown).
Regulation of σE-dependent transcription by ppGpp
As ppGpp does not affect the level of σE, it must alter its activity. ppGpp binds to core RNAP and could directly affect transcription by EσE with or without the assistance of DksA. Alternatively, ppGpp could indirectly affect σE-dependent transcription. Indirect regulation could be achieved by several mechanisms. For example, ppGpp could regulate the expression of a co-activator protein or molecule that is required specifically for transcription by EσE. ppGpp could alter the competition among sigma factors, including σE, for core RNAP, as suggested by studies on the regulation of transcription by the alternative sigma factors σS and σN by ppGpp (Magnusson et al., 2005). The model that ppGpp alters the competition among sigma factors for RNAP is supported by the observation that the rpoDP504L and rpoDS506F mutations, which lower the affinity of σ70 for core RNAP (Hernandez and Cashel, 1995; Sharp et al., 1999; Gruber et al., 2001), restore σN activity during entry into stationary phase in cells lacking ppGpp (Sze and Shingler, 1999; Laurie et al., 2003). These mutations were originally isolated as suppressors of growth defects of ppGppo strains on media lacking amino acids, suggesting that they also suppress defects in Eσ70 activity in ppGppo strains (Hernandez and Cashel, 1995). To determine whether these σ70 variants could suppress the defect in σE activity in strains lacking ppGpp, transcription by EσE was monitored in ppGppo strains carrying these mutations. In the ppGppo strains with either the rpoDS506F or rpoDP504L alleles, not only was transcription by EσE restored during entry into stationary phase, but it was also nearly constitutive throughout the growth curve. σE activity increased by early exponential phase to a level comparable to that observed during entry into stationary phase in the wild-type strain (Fig. 4A, D and E). Similar results were obtained with the σE-dependent fkpA promoter in rpoDS506F ppGppo and rpoDP504L ppGppo strains (data not shown). These results indicate that ppGpp itself is not absolutely required for σE activity. The σ70 variants not only suppress the requirement of ppGpp for σE activity, but also nearly eliminate the growth phase-dependent regulation of transcription by EσE throughout the growth curve.
The observation that mutations in σ70 that reduce its affinity for core RNAP render σE activity nearly constitutive suggests that σE activity is higher because it can compete better against σ70 for binding to core RNAP, and ppGpp is no longer needed. Therefore, other mechanisms to reduce the ability of σ70 to bind to RNAP should also restore σE activity in a strain lacking ppGpp. Rsd is a σ70-specific anti-sigma factor that binds to σ70 and blocks its association with core RNAP (Jishage and Ishihama, 1998). To determine if overexpression of rsd could also restore σE activity in a strain lacking ppGpp, transcription by EσE was monitored in a ppGppo strain carrying a plasmid with the rsd gene placed under the control of an IPTG-inducible promoter. Overexpression of rsd restored the growth phase-dependent increase in σE activity during entry into stationary phase (Fig. 4B, D and E). These results provide further evidence that ppGpp itself is not absolutely required for this activation. Rsd overproduction will also restore σS and σN activity in a ppGppo background, indicating that altering the competition among sigma factors for core RNAP by disabling σ70 is a general mechanism to restore alternative sigma factor activity in the absence of ppGpp (Jishage et al., 2002; Laurie et al., 2003).
We demonstrated that the growth phase-dependent regulation of transcription by EσE is disrupted in a ΔdksAΔrpoZ strain (Fig. 2A and C), presumably because RNAP cannot respond to ppGpp. As overexpression of rsd suppressed the requirement for ppGpp, it should also suppress the defects in σE activity in the ΔdksAΔrpoZ strain, unless there are additional defects unrelated to ppGpp that affect σE activity in this strain. Overexpression of rsd did restore σE activity during entry into stationary phase (Fig. 2B and C), providing further evidence that at least some aspects of the ΔdksAΔrpoZ strain phenocopy a ppGppo strain. The expression of rsd itself is regulated by ppGpp, and increases during entry into stationary phase (Jishage and Ishihama, 1999). This observation, along with the results that overexpression of rsd can restore the growth phase-dependent regulation of σE activity, suggested that ppGpp may indirectly affect σE activity by increasing the expression of rsd. If this were true, then the pattern of EσE activity during the growth of cells lacking rsd should resemble that of cells lacking ppGpp, i.e. no activation during entry into stationary phase. However, no defects in the growth phase-dependent regulation of transcription by EσE were observed in a Δrsd strain (data not shown). Therefore, either rsd is not required for ppGpp to act, or the regulation is redundant and another regulator controls σE activity in the absence of rsd.
The above results demonstrate that the requirement for ppGpp to increase σE activity can be complemented by conditions in the cell that decrease the ability of σ70 to compete for core RNAP. Many mutations that suppress the growth defects in ppGpp-deficient cells have also been mapped to the β and β′ subunits of RNAP (Bartlett et al., 1998; Zhou and Jin, 1998). Several of these mutations will also complement defects in σN activity in ppGppo strains (Sze and Shingler, 1999). We examined the ability of one such mutation, a deletion of amino acids 215–220 of the β′ subunit, to restore EσE activity in a ppGpp deficient strain (Bartlett et al., 1998). This deletion in β′ is not at the sigma/core interface and is unlikely to affect the affinity of σ70 for core RNAP. The increase in σE-dependent activity during entry into stationary phase was restored in the ppGppo strain containing the rpoCΔ215-220 mutation, indicating that alterations in β′ can also complement the defects in σE activity observed when ppGpp is lacking (Fig. 4C, D and E). σE activity was also elevated during mid-exponential phase in this strain, approximately threefold compared with the wild-type strain, but not to the same extent as in the strains with mutations in rpoD (Fig. 4E). Again, similar results were obtained from the fkpA promoter in the ppGpporpoCΔ215-220 strain (data not shown). This mutation in β′ most likely restores σE activity indirectly via its destabilizing effects on ppGpp-sensitive σ70-dependent rRNA promoter complexes, thereby increasing the availability of core RNAP for binding to σE (see Discussion). We have not formally ruled out that this mutation could also directly favour transcription initiation by EσE, thus compensating for the loss of activation in the absence of ppGpp.
To explore further the mechanism(s) responsible for activation of σE activity, direct effects of ppGpp and DksA were tested on promoter complexes formed by EσE. ppGpp and DksA destabilize RNAP complexes formed on all promoters examined to date (Barker et al., 2001a,b; Paul et al., 2004a; 2005; Bernardo et al., 2006; Mallik et al., 2006), but their effects on EσE promoter complexes have not been tested previously. Complexes were formed on two promoters, rpoH P3, the same EσE-dependent promoter used in the in vivo experiments described above, and rybB, the promoter for a σE-dependent small RNA whose expression is also regulated by ppGpp/DksA (Thompson et al., 2007). The fraction of complexes in the presence of ppGpp, DksA, or the two together that remained at different times after addition of the competitor heparin was measured by in vitro transcription (Barker et al., 2001). Both promoters formed relatively stable complexes with EσE under these solution conditions (t½ = 72 min for rpoH P3 and t½ = 26 min for rybB; Fig. 5A). DksA by itself had little effect on the lifetimes of competitor-resistant complexes at either promoter, while ppGpp by itself destabilized complex stability by approximately 2.5-fold (Fig. 5A). In contrast, DksA and ppGpp together greatly decreased the lifetimes of the promoter complexes, by sixfold for the rybB promoter and 15-fold for the rpoH P3 promoter (Fig. 5A). Future studies will be needed to address why DksA alone had little effect on the half-lives of these promoters under these conditions; nevertheless, these results show that ppGpp and DksA can together function on promoter complexes containing EσE RNAP.
Although ppGpp and DksA destabilize open complexes on all promoters yet examined, their effect on overall transcription depends on the properties of the individual promoters (Barker et al., 2001a,b; Paul et al., 2004a; 2005; Bernardo et al., 2006; Mallik et al., 2006). Therefore, effects of ppGpp and DksA on transcriptional output were measured by multiround transcription from rpoH P3 (representative assays are shown at the left in Fig. 5B, and the results from multiple assays are quantified at the right in Fig. 5B). As observed previously with certain σ70-dependent amino acid biosynthesis promoters (Paul et al., 2005), ppGpp by itself exerted little or no effect (Fig. 5B), and as also observed previously, DksA by itself activated transcription up to approximately twofold (Fig. 5B). When 200 μM ppGpp was also included in the reactions, transcription increased slightly more (up to approximately threefold) and was observed at lower concentrations of DksA (Fig. 5B). Similar effects were obtained on transcription from the rybB promoter, although the magnitude of the activation was smaller. In preliminary experiments DksA alone (2 μM) increased transcription from PrybB approximately 1.5-fold and DksA/ppGpp (2 μM/100 μM respectively) increased transcription from PrybB 1.8-fold, while ppGpp alone had no effect (data not shown). We conclude that ppGpp and DksA can directly activate transcription by EσE, and, as with Eσ70-dependent promoters (Paul et al., 2005), different promoters are activated to different extents.
The data presented here are consistent with the model that ppGpp works as a general alarm system to redistribute RNAP from promoters for genes required for rapid growth to promoters for genes that are important for survival during stress, shifting the spectrum of genes transcribed by Eσ70 and activating genes transcribed by alternative sigma factors (Nystrom, 2004; Magnusson et al., 2005; Costanzo and Ades, 2006; Gourse et al., 2006). Our data also support the model that ppGpp, along with its cofactor DksA, changes the distribution of RNAP holoenzymes on promoters, both through direct effects on promoter complexes sensitive to ppGpp and DksA and also through indirect effects achieved primarily by reducing transcription of rRNA genes (Zhou and Jin, 1998; Barker et al., 2001a; Jishage et al., 2002; Paul et al., 2005; Szalewska-Palasz et al., 2007). In this manner, ppGpp and DksA provide a general mechanism to co-ordinately activate alternative sigma factors.
ppGpp and the regulation of σE expression
In addition to co-ordinately regulating alternative sigma factors by altering the distribution of RNAP holoenzymes, ppGpp and DksA exert specific effects on individual alternative sigma factors. ppGpp is thought to increase the production of σS, but not of σN (Gentry et al., 1993; Sze and Shingler, 1999; Kvint et al., 2000). We found that the production of σE is not dependent on ppGpp. Although one of the two promoters that directs transcription of the rpoE operon, rpoE P2, is dependent on σE (Rouviere et al., 1995), not all promoters transcribed by EσE are necessarily subject to control by ppGpp/DksA. Only specific σ70-dependent promoters have the kinetic characteristics that make them sensitive to ppGpp/DksA, and this is likely to be true of σE-dependent promoters as well (see below). Alternatively, as the rpoE gene is transcribed from two major promoters, one dependent on σ70 and the other dependent on σE, increased expression from the σ70-dependent rpoE P1 promoter might compensate for decreased expression from rpoE P2 in the absence of ppGpp. Conversely, if transcription from rpoE P2 increases in the presence of ppGpp, transcription from rpoE P1 may decrease, thereby maintaining a constant amount of σE in the cell.
Direct regulation of σE-dependent transcription by ppGpp
ppGpp and DksA can have positive, negative, or no effects on overall transcription, even though these factors reduce the stability of open complexes on all promoters tested to date, including the two σE-dependent promoters tested here. This observation can be explained by the fact that ppGpp/DksA affect transcription from only a subset of promoters, those with specific kinetic characteristics. Studies with rrnB P1 promoter variants strongly suggest that ppGpp/DksA inhibit transcription by further destabilizing promoter complexes that are intrinsically short-lived. Decay of the open complex formed on these promoters is rate-limiting for transcription initiation (Haugen et al., 2006). In contrast, promoters that form stable open complexes are not inhibited by ppGpp/DksA, even though the open complexes are destabilized, because dissociation of the open complex is not rate-limiting for transcription initiation (Haugen et al., 2006). Consistent with this model, the open complexes formed by EσE are relatively long-lived and ppGpp/DksA do not inhibit transcription from these promoters. It was somewhat surprising that effects of ppGpp alone but not DksA alone were observed on the lifetime of the open complex, while effects of DksA alone but not ppGpp alone were observed in multiround transcription assays. However, we emphasize that ppGpp/DksA together had parallel effects in the two assays, and this is the condition most relevant to cells.
The mechanism of positive control of transcription by ppGpp/DksA is much less well understood than negative control. Experiments with the σ70-dependent PargI promoter suggest that ppGpp/DksA activate transcription by increasing the isomerization rate from the closed to open complex (Paul et al., 2005). A model based on the results with PargI and other amino acid biosynthetic promoters has been proposed explaining how DksA/ppGpp could reduce the free energy of a transition state on the pathway to open complex formation, and this could result in positive effects on transcription (Paul et al., 2005). The kinetics of transcription initiation for EσE have not yet been studied in detail. In fact these studies present the first measurements of open complex stability for EσE on any promoter. Studies on the mechanics of transcription initiation at EσE-dependent promoters will be needed to elucidate how pGpp/DksA acts on these complexes.
Indirect regulation of alternative sigma factor transcription by ppGpp
EσE activity increased fivefold during entry into stationary phase in the in vivo reporter assays with the rpoH P3 promoter. This increase was greater than that observed by ppGpp and DksA in vitro. Although the solution conditions may not adequately mimic the conditions in vivo, a likely explanation for the disparity between the magnitude of the observed effects in vivo and in vitro is that a component of the activation of σE-dependent promoters results from indirect effects of ppGpp and DksA. Previous models proposed to explain how ppGpp positively regulated transcription by EσS and EσN invoked alteration of the competition among sigma factors for RNAP (Hernandez and Cashel, 1995; Zhou and Jin, 1998; Jishage et al., 2002; Magnusson et al., 2005; Szalewska-Palasz et al., 2007).
Several hypotheses can explain how ppGpp alters the competition among sigma factors for RNAP. The observation that both mutations in rpoD, which reduce the affinity for core RNAP, and overexpression of the σ70-inhibitor rsd bypass the requirement of ppGpp for EσE, EσS and EσN activity in vivo suggests that ppGpp could affect the association of sigma factors with RNAP favouring alternative sigma factors (Hernandez and Cashel, 1995; Sharp et al., 1999; Sze and Shingler, 1999; Jishage et al., 2002; Laurie et al., 2003). However, to our knowledge it has not been demonstrated that ppGpp and/or DksA directly alters the affinity of any sigma factor for core RNAP. The model also does not explain suppression of the defects in σE activity by the rpoC variant. The rpoC mutation, a deletion of amino acids 215–220, is not at the sigma/core interface and therefore is less likely than the rpoD mutants to alter the affinity of σ70 for core RNAP.
An alternative explanation is that ppGpp alters the competition by increasing the amount of free RNAP available to bind to all sigma factors via its effects on transcription of ribosomal RNAs. In rapidly growing cells, up to 70% of the RNAs transcribed in the cell are stable rRNAs encoded in long operons (Bremer and Dennis, 1996). When ppGpp levels increase, transcription of these operons decreases and the core RNAP that was actively engaged in transcription will be released upon transcription termination (Cashel et al., 1996; Barker et al., 2001b). This release of core RNAP will increase the size of the free pool of the enzyme available to bind all sigma factors. This model is consistent with phenotypes of the rpoC mutation and with additional properties of the rpoD mutations, which appear to functionally mimic the effect of ppGpp/DksA on transcription initiation from the rRNA promoters (Bartlett et al., 1998; 2000; Barker et al., 2001a; Laurie et al., 2003; Szalewska-Palasz et al., 2007). RNAP containing the β'Δ215-220 deletion reduced transcription from the rrnB P1 promoters, and the open complexes formed by the variant holoenzyme on this promoter are extremely unstable (Bartlett et al., 1998; 2000; Barker et al., 2001a). The σ70 variants, in addition to having a lower affinity for core RNAP than WT σ70, also further destabilize competitor-resistant open complexes formed on the rrnB P1 promoter in vitro (Laurie et al., 2003; Szalewska-Palasz et al., 2007).
The combination of direct and indirect effects of ppGpp on gene expression provide a powerful means of transcriptional regulation. On a global level, ppGpp co-ordinately alters the activity of individual sigma factors, redistributing RNA polymerase among the sigma factors in the cell. This general regulatory mechanism works in conjunction with the ppGpp-independent signalling pathways that are specific for each alternative sigma factor and determine the overall amounts of the different alternative holoenzymes available to respond to changing ppGpp levels. For example, the overall amount of free σE available to interact with core RNAP is determined primarily by the amount and proteolytic stability of the anti-sigma factor RseA (Ades, 2004; Costanzo and Ades, 2006). Finally, the unique thermodynamic and kinetic properties of individual promoters ultimately determine the extent to which the direct effects of ppGpp on transcription and the changing amounts of alternative holoenzymes modulate gene expression.
Strains and growth conditions
Strains used in this work are listed in Table 1. Mutant alleles were moved into the appropriate strains using P1 transductions according to standard techniques (Miller, 1972). The rpoD and rpoC mutant alleles with tightly linked zgh-3075::Tn10 (rpoD alleles) or thi39::Tn10 (rpoC allele) insertions (Hernandez and Cashel, 1995; Bartlett et al., 1998) were transferred into strain SEA2010 by P1 transduction. Transductants were selected on tetracycline and the presence of the correct mutation was verified by sequencing. Experiments with the rpoD, rpoC and dksA rpoZ mutant strains were performed with at least two independent transductants to ensure that the results were not affected by spontaneous suppressor mutations. All ppGppo strains were verified as being unable to grow on minimal media lacking amino acids, and the rpoD and rpoC suppressor strains were verified to have reverted this auxotrophy. Strain SEA2023 was made by a targeted disruption of the lacYA genes in the ΦλrpoHP3::lacZ reporter of SEA001. The genes were deleted according to the procedure of Datsenko and Wanner (2000) and the drug marker removed by FLP recombinase. Strain SEA2043 was made by transformation of SEA001 with pLC245 expressing the rpoE gene (Rhodius et al., 2005) followed by P1 transduction of the ΔrpoE-rseC::kan allele. Strains were grown in LB broth at 30°C with aeration unless otherwise noted.
Plasmids used in this study are listed in Table 1. The plasmid pRsd was constructed by PCR amplifying the rsd gene from the chromosome and cloning it into the expression plasmid pTrc99a. Plasmids used as transcription templates are derivatives of pRLG770 (Ross et al., 1990), containing the σE-dependent rpoH P3 (pSEB015) or rybB (pSEB014) promoters. The rpoH P3 promoter was amplified with the primers 5′-GGGCCGGAATTCGCCTTGATGTTACCCGAGAG-3′ and 5′-GGGCCAGGTGGAGACCCCCAGTCACGACGTTGTAAA-3′ from plasmid p2rpoHP3, which has the isolated σE-dependent rpoH P3 promoter cloned into the PstI-HindIII sites of pUC19. The rpoH P3 promoter sequence, from the −35 region to the +1 for transcription, is identical to that in the chromosomal lacZ fusion (λrpoHP3::lacZ) used to study σE activity (Mecsas et al., 1993). The resulting PCR product was digested with EcoRI and BsaI then cloned into the EcoRI and HindIII of pRLG770 to make pSEB015. The rybB promoter region and rybB gene, including the native transcription terminator, were amplified from genomic DNA by PCR with the primers: 5′-GGGCGGGAATTCGTTGTTCGGCGCAATGAT-3′ and 5′-GGGCCAAGCTTGTTGAGAGGGTTGCAGGGTA-3′ and cloned into the EcoRI and HindIII of pRLG770 to make pSEB014.
Whole-cell extracts were prepared as described (Gentry et al., 1993). Briefly, cells were lysed in protein sample buffer, proteins were precipitated with acetone, and re-suspended in 2% SDS. Protein concentrations were determined using the BCA Protein Assay (Pierce). Ten micrograms of total protein from each sample was loaded onto 12% polyacrylamide-SDS gels and transferred to HybondTM-P, PVDF membrane (GE Healthcare). Bands containing σE were detected by probing the blots with rabbit polyclonal antibodies raised against σE (gift from CA Gross), then with alkaline phosphatase conjugated goat anti-rabbit IgG (Vector Laboratories, Burlingame, CA). The secondary antibody was visualized with the ECF reagent from GE Healthcare according to the manufacturer's instructions. Blots were scanned using Typhoon 8600 Imager in fluorescence mode.
Overnight cultures were diluted to an OD600 of 0.02 and grown with shaking in a gyratory water bath at 30°C. Samples (0.5 ml) were collected throughout the growth curve. The β-galactosidase activity of each sample was measured by the standard assay (Miller, 1972) and is expressed as the OD420 of the reaction mixture divided by the reaction time (o-nitrophenol min−1). The data are presented as differential rate plots in which β-galactosidase activity in each 0.5 ml sample is plotted versus the optical density (OD600) of the sample. β-galactosidase activity (o-nitrophenol min−1) per 0.5 ml sample is plotted, rather than standard Miller Units (o-nitrophenol min−1 OD600−1), therefore the slope of the curve at each time point indicates the change in β-galactosidase activity with increased cell number. The plots illustrate how σE activity changes throughout the growth curve, in recovery from stationary phase, exponential phase and re-entry into stationary phase. A complete explanation of differential rate plots is presented in Costanzo and Ades (2006). Experiments were repeated a minimum of three times with independent cultures.
For measurements of σE activity as a function of growth rate, cultures were grown in MOPS minimal media (Teknova) with 0.4% glycerol, MOPS minimal media (Teknova) with 0.2% glucose, EZ rich media (Teknova) and LB at 37°C. σE activity was determined in early exponential phase to avoid interference from any additional regulation owing to changing ppGpp levels during entry into stationary phase. Activity was determined by the slope of the line on a differential rate plot of β-galactosidase activity in 0.5 ml of culture (as described above) versus OD600. Experiments were repeated a minimum of three times to ensure reproducibility. σE activity in early exponential phase reflects growth in fresh media at a low optical density. σE activity is the same whether it is measured in cultures directly following dilution of a saturated overnight or following repeated dilution of cultures that had reached exponential phase (OD600∼0.3) to ensure that the cells are in so-called steady-state growth (Costanzo and Ades, 2006; and data not shown).
N-terminally His-tagged σE was purified from strain BL21(DE3) slyD::kan pLysS pPER76 as previously described (Campbell et al., 2003). Briefly, cells were grown at 25°C to an OD600 of 0.5 at which point IPTG was added to induce protein production. Following 1.5 h of induction, cells were harvested by centrifugation, re-suspended in lysis buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 20 mM imidazole, 2.5 mM β-mercaptoethanol, 1 mM PMSF). Cells were lysed by sonication and the lysate was cleared by centrifugation. The supernatant containing soluble His-σE was loaded onto a NiNTA column. Bound proteins were eluted with a stepwise gradient of 20, 60, 100 and 200 mM imidazole in column buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 2.5 mM β-mercaptoethanol). Fractions containing σE were pooled and dialysed into 20 mM Tris-HCl pH 8.0, 50 mM NaCl, 2.5 mM β-mercaptoethanol.
His-tagged DksA was purified as described in Paul et al., 2004a or with minor modifications. Briefly, cells were grown at 37°C to an OD600 of 0.4 at which point IPTG was added to induce protein production. Following 3 h of induction, cells were harvested by centrifugation, re-suspended in buffer 1 (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 5 mM imidazole, 5 mM β-mercaptoethanol, 0.1 mM PMSF) and lysed by sonication. The lysate was cleared by centrifugation and the supernatant loaded onto a NiNTA column. Bound proteins were eluted with a stepwise gradient of buffer 1 with 75, 150 and 300 mM imidazole. Fractions containing His-DksA were combined and dialysed into buffer 2 (20 mM Tris-HCl pH 8.0, 100 mM NaCl, 0.1 mM DTT) then loaded onto a Hi-TrapTM Q FF column (GE Healthcare). Bound proteins were eluted with a stepwise gradient of buffer 2 with 200, 300 and 400 mM NaCl. Fractions containing His-DksA were pooled and dialysed into Buffer 2 with 20% glycerol.
RNAP–promoter complex decay assays
Lifetime of the competitor-resistant open complex was measured by single-round in vitro transcriptional assays as described in Barker et al. (2001b) with the exception of the use of EσE (reconstituted at 30°C from 1:2 ratio of native core RNAP: His-σE). Briefly, 10 nM EσE and 1 nM supercoiled plasmid DNA in transcription buffer (40 mM Tris-HCl pH 8.0, 10 mM MgCl2, 50 mM NaCl, 1 mM DTT, 0.1 mg ml−1 BSA), and ppGpp and DksA (as indicated in Fig. 5) were incubated for 10 min at 30°C. Heparin was added to a final concentration of 10 μg ml−1, and aliquots were removed to tubes containing NTPs (500 μM ATP, 200 μM GTP, 200 μM CTP, 10 μM UTP and 1.0 μCi [α-32P]-UTP) at various times after heparin addition. Transcription reactions were stopped after 10 min with an equal volume of urea-based gel loading buffer. Transcripts were separated on a 6% polyacrylamide gel containing 7 M urea and then visualized and quantitated by phosphorimaging using ImageQuant software (Molecular Dynamics).
Multiround transcription assays
EσE holoenzyme was formed by incubating 200 nM core RNAP (Epicentre) with 800 nM His-σE for 10 min at 30°C. Multiround transcription reactions were initiated by addition of EσE, to a final concentration of 80 nM σE and 20 nM core RNAP, to prewarmed (30°C) transcription mix containing 2.5 nM supercoiled plasmid template, 5% glycerol, transcription buffer, 500 μM ATP, 200 μM CTP, 200 μM GTP, 10 μM UTP, 2.5 μCi [α-32P]-UTP, and the appropriate concentrations of ppGpp and/or His-DksA. After 10 min at 30°C, reactions were stopped by the addition of stop solution (80% formamide, 20 mM EDTA, 0.1% xylene cyanol and 0.1% bromophenol blue). Samples were electrophoresed on 6% polyacrylamide gels containing 7.5 M urea, visualized by phosphorimaging and quantified using ImageQuant software (Molecular Dynamics).
We thank Mike Cashel, Carol Gross, Jim Hernandez and Karen Wassarman for sharing strains, Steve Rutherford for preparing the σE used in some experiments, numerous discussions and other contributions, and Pui-Wen Cheung for technical assistance.
This work was supported by Grant MCB-0347302 from the National Science Foundation to S.E.A., Grant R37-GM37048 from the National Institutes of General Medical Sciences to R.L.G., an NIH Graduate Research Fellowship to A.B.B. and a National Science Foundation Graduate Research Fellowship to S.E.B.