During active growth of Escherichia coli, majority of the transcriptional activity is carried out by the housekeeping sigma factor (σ70), whose association with core RNAP is generally favoured because of its higher intracellular level and higher affinity to core RNAP. In order to facilitate transcription by alternative sigma factors during nutrient starvation, the bacterial cell uses multiple strategies by which the transcriptional ability of σ70 is diminished in a reversible manner. The facilitators of shifting the balance in favour of alternative sigma factors happen to be as diverse as a small molecule (p)ppGpp (represents ppGpp or pppGpp), proteins (DksA, Rsd) and a species of RNA (6S RNA). Although 6S RNA and (p)ppGpp were known in literature for a long time, their role in transcriptional switching has been understood only in recent years. With the elucidation of function of DksA, a new dimension has been added to the phenomenon of stringent response. As the final outcome of actions of (p)ppGpp, DksA, 6S RNA and Rsd is similar, there is a need to analyse these mechanisms in a collective manner. We review the recent trends in understanding the regulation of σ70 by (p)ppGpp, DksA, Rsd and 6S RNA and present a case for evolving a unified model of RNAP redistribution during starvation by modulation of σ70 activity in E. coli.
A great deal of regulation of protein expression in bacteria occurs at the level of transcription of genes carried out by RNA polymerase (RNAP) by binding to specific regions on the chromosome (promoters). The specificity of RNAP binding to promoter DNA sequences is provided by sigma factors, which associate with core RNAP to form a holo enzyme (referred to as holo RNAP or Eσ). For transcribing genes, all the sigma factors are dependent on a single species of core RNAP with the following subunit composition: α2, β, β′ and ω. The rate of transcription initiation from various promoters has been known to be regulated by a number of DNA-binding proteins, which either bind to specific sequences on DNA (activators or repressors) (for reviews, see Browning & Busby, 2004; Grainger & Busby, 2008; Balleza et al., 2009) or rather bind in a nonspecific manner (nucleoid-associated proteins) (for reviews, see McLeod & Johnson, 2001) or by mechanisms that do not involve direct interaction of transcription factors with DNA (reviewed by Haugen et al., 2008). Another parameter determining the transcriptional efficiency of a particular gene promoter is the amount of cognate Eσ availability, which is influenced by a number of factors discussed below.
Transcriptional switching: sigma factors compete for limited amounts of core RNAP
The presence of a large number of sigma factors in bacteria helps in switching on transcription in response to a larger variety of environmental signals. The following seven sigma factors, designated either by numbers or letters, are present in Escherichia coli: σ70 (σD), σ54 (σN), σ38 (σS), σ32 (σH), σ28 (σF), σ24 (σE) and σ18 (σFecI). The sigma factors can be broadly classified into two groups: the σ70 family of sigma factors, which share structural similarities, and the σ54 group, which are structurally different from the σ70 family (Wösten, 1998). Despite the overall similarity in their structures, the σ70 family of sigma factors recognizes different classes of promoters (Gruber & Gross, 2003). The core RNAP-binding surface on sigma factors involves multiple regions and is conserved across different classes (Sharp et al., 1999). Except for the essential housekeeping sigma factor, the majority of the other sigma factors (the alternative sigma factors) are required only for certain physiological roles and thus the genes encoding these factors can be inactivated without affecting the viability of the organism, although inactivation of certain alternative sigma factors (e.g. σ24 of E. coli) has also been shown to result in loss of viability (De Las Peñas et al., 1997).
The number of core RNAP molecules inside the cell is limited and this number remains almost constant under various physiological states (Grigorova et al., 2006; Piper et al., 2009). Similarly, the intracellular concentration of σ70, σ54 and σ28 has been found to be constant in actively growing and stationary-phase E. coli cells (Jishage et al., 1996). However, the levels of some of the alternative sigma factors (e.g. σ32 and σ38) vary considerably upon entry into altered physiological states (e.g. stationary phase or heat shock). During the exponential phase, the level of the housekeeping sigma factor σ70 in E. coli is the highest, followed by σ28 and σ54 (50% and 10% of σ70, respectively); σ38 levels are below the detection limits during this phase of growth (Jishage et al., 1996). During the stationary phase, the levels of σ38 go up to 30% of the σ70 levels (Jishage et al., 1996; Piper et al., 2009). In addition to the intracellular concentration, another factor that determines the probability of a sigma factor's association with core RNAP is the affinity with which it binds to core RNAP. As shown in Table 1, in vitro binding experiments using purified E. coli core RNAP and sigma factors have shown that the housekeeping sigma factor, σ70, has the highest affinity for core RNAP and σ38 has the lowest (Maeda et al., 2000). A similar situation exists in Bacillus subtilis, in which σA has a higher affinity to core RNAP compared with σH (Fujita & Sadaie, 1998) or sporulation-specific σF or σB (Lord et al., 1999; Rollenhagen et al., 2003). However, during stress or sporulation, the concentrations of σF or σB are not more than twofold higher than σA, suggesting therefore that increased levels of these sigma factors alone cannot compensate for the lack of a higher affinity to core RNAP as compared with σA (Lord et al., 1999).
Binding affinities of sigma factors to core RNAP of Escherichia coli (Maeda et al., 2000) and Bacillus subtilis (Fujita & Sadaie, 1998; Lord et al., 1999)
Affinity (Kd) in nM
The concept of sigma factor competition for limited amounts of core RNAP has been experimentally proven by varying the level of a particular sigma factor by regulated expression or by generating low-affinity sigma factor mutants. For example, a mutant of the heat shock sigma factor, σ32, which had a lower affinity for core RNAP, showed reduced ability to express heat shock genes in the presence of σ70, implying that the mutant σ32 was unable to compete with σ70 for limited amounts of core RNAP (Zhou et al., 1992). Also, a mutant σ70 with a fivefold reduction in transcriptional efficiency in vivo was found to suppress the heat shock defect of the σ32 mutant (Zhou & Gross, 1992). Similarly, in a study involving a σ38 mutant in vivo, it was found that a σ70-driven gene uspA was induced, while uspB, which is transcribed by σ38, was repressed. Overexpression of σ70 had a similar effect, while overexpression of σ38 had the opposite effect, in which uspA was repressed while uspB was induced (Farewell et al., 1998). Varying the levels of the principal sigma factor in B. subtilis, σA, results in the association of lower or higher amounts of σH with core RNAP (Hicks & Grossman, 1996). In summary, the amounts of sigma factor and affinity to core RNAP influence the outcome of sigma factor competition for association with core RNAP.
Modulation of σ70activity during stress and starvation: redistribution of RNAP to maintenance-related promoters
During the transition from nutrient abundance to nutrient starvation, the limited amount of RNAP is redistributed to transcribe genes involved in maintenance and survival. Although the amounts of certain alternative sigma factors increase during the shift from ‘feast to famine’ (Nystrom, 2004), the increased levels alone cannot compensate for the lower affinity of alternative sigma factors to the core RNAP. Hence, the bacterial cells have evolved strategies to subdue the dominant σ70 in a reversible manner so that the alternative sigma factors can compete more effectively for association with core RNAP. The effectors that facilitate the switchover of transcription during starvation are (p)ppGpp, DksA, Rsd and 6S RNA (Fig. 1). The latter two are involved in inhibiting σ70-driven transcription; ppGpp and DksA, on the other hand, carry out their functions by direct inhibition or activation or indirectly by altering the core binding competitiveness of sigma factors. Because the actions of (p)ppGpp, DksA, Rsd and 6S RNA facilitate the transcription of genes involved in the maintenance of cell functions, understanding the contributions of each one of them will unravel a larger picture of transcriptional regulation during starvation.
Transcriptional switching from proliferation related to maintenance- and survival-related functions during nutrient starvation in Escherichia coli. During active growth, σ70 dominates the transcription activity, while during starvation, alternative sigma factors contribute to major transcriptional activity, which is facilitated by (p)ppGpp, DksA, Rsd and 6S RNA. This is only a diagrammatic representation of transcriptional switching and does not indicate the abundance of any of the molecules shown.
(p)ppGpp and DksA
Amino acid starvation in bacteria results in shutting down of the synthesis of rRNA and tRNA by a process known as a stringent response (reviewed by Magnusson et al., 2005; Potrykus & Cashel, 2008). The primary mediator of this response is (p)ppGpp, whose levels are regulated by RelA and SpoT in Beta- and Gammaproteobacteria, while in others, including many Gram (+) bacteria, a single Rel Spo homologue (RSH) protein, Rel (Mittenhuber et al., 2001), performs a similar function (reviewed by Chatterji & Ojha, 2001; Potrykus & Cashel, 2008). Depletion of iron, phosphorus, carbon source and fatty acids also induces the synthesis of (p)ppGpp (Srivatsan & Wang, 2008 and references therein). A mutant E. coli lacking both RelA and SpoT is unable to generate (p)ppGpp (ppGpp0) and hence does not show any stringent response and continues generating rRNA during amino acid starvation (relaxed response), which has also been confirmed by proteome analysis and global transcription profiles of the mutant strains (Magnusson et al., 2003; Durfee et al., 2008). Initially, (p)ppGpp was thought to be the sole effecter of the physiological events seen during the stringent response; however, with the discovery and characterization of DksA, the story of stringent response has undergone a major makeover. DksA was isolated by virtue of its ability to suppress the thermotolerance defect of a dnaK mutant of E. coli (Kang & Craig, 1990). Subsequently, its role in the expression of σ38 during the stationary phase and in stringent response came to light (Brown et al., 2002; Paul et al., 2004, 2005).
(p)ppGpp and DksA target transcription machinery
Binding experiments and structural studies have shown that (p)ppGpp interacts directly with RNAP (Chatterji et al., 1998; Artsimovitch et al., 2004). Suppressor mutants of E. coli that can reverse the (p)ppGpp0 phenotype have been mapped to σ70, β or β′ subunits (Hernandez & Cashel, 1995; Szalewska-Palasz et al., 2007). Mutants of DksA, which can suppress the auxotrophic requirement of (p)ppGpp0 cells, were found to have increased activity in vitro and in vivo by virtue of their higher affinity to RNAP (Blankschien et al., 2009). Studies on the role played by another RNAP subunit, ω, in a stringent response has not yielded very clear answers. Although in vitro transcription assays indicated that ω was required for (p)ppGpp action in vitro (Vrentas et al., 2005) and relA gene was downregulated in strains lacking ω (Chatterji et al., 2007), the stringent response was not affected in ΔrpoZ cells (Gentry et al., 1991), pointing towards the redundancy of the ω function in vivo. Although purified RNAP lacking the ω subunit does not respond to (p)ppGpp in vitro, this unresponsiveness could be restored by DksA, which suggested a possible link between ω and the stringent response (Vrentas et al., 2005). Based on the structures of DksA and RNAP–(p)ppGpp, a model was proposed that suggested that DksA and (p)ppGpp interact with each other in the secondary channel of RNAP and indicated that DksA might stabilize the (p)ppGpp–RNAP complex (Perederina et al., 2004). However, recent mutational studies of the predicted (p)ppGpp-binding amino acid residues in E. coli RNAP do not support some of the conclusions derived from the earlier findings (Vrentas et al., 2008). By isolating mutants of E. coli RNAP, which show altered effects on rRNA promoters in the presence of DksA, Rutherford et al. (2009) proposed that DksA by binding to the secondary channel of RNAP disrupts its interactions with promoter DNA allosterically without making any direct contact with DNA. The binding affinity of DksA to free RNAP (core or holo) was found to be 10 times higher than to RNAP in an open complex (Lennon et al., 2009), suggesting that DksA is likely to bind RNAP before the formation of open complex. Additionally, DksA has also been shown to facilitate the binding of Eσ70 to promoter DNA (Aberg et al., 2008; Lyzen et al., 2009).
(p)ppGpp and DksA can exert complementary and opposing effects
Both (p)ppGpp and DksA exert their transcriptional regulatory effects by interactions with RNAP. In general, (p)ppGpp and DksA enhance the transcriptional activity from promoters driven by alternative sigma factors, while in the case of σ70 promoters, (p)ppGpp and DksA function by differential regulation in which growth-related σ70-driven genes (e.g. rRNA) are negatively regulated, whereas those involved in maintenance (e.g. amino acid biosynthesis) are positively regulated (Paul et al., 2004, 2005; Gourse et al., 2006; Potrykus & Cashel, 2008). Among the alternative sigma factors, the activities of σ32, σ38, σ24 and σ54 have also been shown to be dependent on (p)ppGpp and DksA (Kvint et al., 2000; Szalewska-Palasz et al., 2007; Costanzo et al., 2008). Additionally, both (p)ppGpp and DksA have been shown to be involved in increased synthesis of σ38 during the stationary phase (Gentry et al., 1993; Brown et al., 2002). Although the exact characteristics of stringently regulated promoters are not very clear, some features that seem to influence responsiveness to (p)ppGpp are −35 and −10 sequence, spacer length, presence of a GC- or an AT-rich sequence between the transcription start site and −10 sequence and the supercoiling status (Travers et al., 1980; Figueroa-Bossi et al., 1998; Potrykus & Cashel, 2008).
With the elucidation of the transcriptional regulatory activity of DksA (Paul et al., 2004, 2005), some of the discrepancies observed between in vitro and in vivo effects of (p)ppGpp could be explained as DksA was shown to influence (enhance) both negative and positive effects of (p)ppGpp on rRNA and amino acid synthesis promoters, respectively. Further, the complementary effects of (p)ppGpp and DksA were also seen in σ54-driven transcription in which both DksA and (p)ppGpp were found to contribute towards σ54 transcription in an additive manner and in the absence of (p)ppGpp and DksA, transcription from σ54-dependent promoters in vivo was completely abolished (Bernardo et al., 2006). Because of its stimulatory effects on (p)ppGpp-mediated transcription inhibition or activation, DksA was thought to be a cofactor of (p)ppGpp (Paul et al., 2004, 2005; Potrykus & Cashel, 2008). However, subsequently, it was demonstrated that overexpression of DksA can reverse a number of phenotypes of a (p)ppGpp0E. coli mutant (Magnusson et al., 2007), which suggested that DksA can also play an independent (complementary to (p)ppGpp) role during a stringent response. Recent studies on comparison of phenotypes of (p)ppGpp0 and DksA mutants in E. coli suggest that these two effectors do have common and complementary roles; however, the presence of opposing phenotypes points towards the possibility of distinct regulatory functions as well. For example, the overexpression of DksA was able to reverse the amino acid auxotrophy, motility defect and filamentation phenotypes of a (p)ppGpp0E. coli mutant, although the effects on adhesion were found to be opposing (Magnusson et al., 2007). Similarly, analysis of fimbriation in E. coli lacking (p)ppGpp and DksA showed contrasting phenotypes, although in vitro transcription from the σ70-driven fimB P2 promoter could be stimulated by (p)ppGpp and DksA independently as well as when added together. The apparent contradiction observed in in vitro and in vivo effects was attributed to the increased association of secondary channel-binding proteins GreA and GreB to RNAP in the absence of DksA, leading to the stimulation of transcription from fimB. The contrasting effect was also reflected in the transcription profile analysis of (p)ppGpp0, ΔdksA and (p)ppGpp0/ΔdksA mutants, wherein it was found that although a number of genes were affected similarly by the absence of either of these, some genes (e.g. chemotaxis and flagellum synthesis) showed opposite effects as well (Aberg et al., 2008, 2009). Opposite effects of (p)ppGpp and DksA on transcriptional regulation have also been demonstrated on a stringently regulated bacteriophage Lambda pR promoter (Lyzen et al., 2009). In conclusion, the effects of DksA can be independent, additive, synergistic or even antagonistic to the (p)ppGpp-mediated transcription regulation in E. coli.
Conditions disfavouring the association of σ70 with core RNAP can compensate for the loss of (p)ppGpp or DksA
Escherichia coliσ38, σ32 and σ24 and Pseudomonas putidaσ54 regulons are transcribed poorly in the absence of (p)ppGpp. The improvement in the transcriptional ability of these alternative sigma factors could be brought about by processes that reduce the transcription activity of σ70 by disfavouring its association with core RNAP, suggesting that the presence of (p)ppGpp increases the competitiveness of alternative sigma factors against σ70 in binding to core RNAP. Evidence for this hypothesis came from studies of Jishage et al. (2002), who showed that the ability of σ32 and σ38 to compete with σ70 was diminished in a (p)ppGpp0 strain. It was also shown that the presence of a σ70 with a reduced affinity to core RNAP, underexpression of σ70 or overexpression of a σ70 antagonist, Rsd, could restore the transcriptional activity of σ38 in the absence of (p)ppGpp in vivo. The effect of (p)ppGpp and DksA on favouring the association of alternative sigma factors with core RNAP has been studied extensively using σ54-driven (p)ppGpp-dependent transcription of the Po promoter of P. putida (Sze & Shingler, 1999; Laurie et al., 2003; Bernardo et al., 2006; Szalewska-Palasz et al., 2007). In the absence of (p)ppGpp and DksA transcription from the Po promoter in vivo is considerably reduced, although these regulators do not affect the σ54-driven transcription in vitro (Sze & Shingler, 1999; Bernardo et al., 2006). Expression of two σ70 mutants in E. coli that do not compete effectively with σ54 for binding to limited amounts of core RNAP in vitro could stimulate transcription from Po promoters in the absence of (p)ppGpp and DksA (Laurie et al., 2003; Bernardo et al., 2006). Also, underexpression of σ70 or overexpression of Rsd could restore the transcriptional activity of σ54-driven Po in the absence of (p)ppGpp (Laurie et al., 2003), although, contrary to (p)ppGpp's ability to favour σ32 over σ70 in binding to core RNAP in vitro and in vivo, competition assays between σ70 and σ54, in vitro, showed that the presence of (p)ppGpp does not influence the outcome of the competition between σ70 and σ54 (Laurie et al., 2003). Suppressor mutants obtained based on their ability to reverse the auxotrophic nature of (p)ppGpp0E. coli cells were found to map to β or β′ subunits of RNAP and some of them were found to support (p)ppGpp-independent transcription from the Po promoter (Sze & Shingler, 1999). The transcription-enhancing effect of these β or β′ mutants on the Po promoter was also seen in ΔdksA mutants (Szalewska-Palasz et al., 2007). In vitro biochemical characterization suggested that these mutant RNAP forms favour the association of σ54 over σ70 with core RNAP and form unstable open RNAP–promoter complexes (Szalewska-Palasz et al., 2007). Another example of (p)ppGpp/DksA-mediated increase in the transcription of an alternative sigma factor is stress-regulated σ24 of E. coli. Expression of low-affinity σ70 mutants in E. coli makes σ24 activity constitutive in a (p)ppGpp0 strain, whereas overexpresssion of Rsd stimulated σ24-driven transcription during the stationary phase in the same strain (Costanzo et al., 2008). As seen with σ54, a mutant β′ that suppresses the (p)ppGpp0 phenotype in E. coli could restore the increase in σ24 activity upon entry into the stationary phase in a (p)ppGpp0 background (Costanzo et al., 2008). Interestingly, the presence of DksA and (p)ppGpp stimulated transcription from σ24 promoters even in vitro, suggesting a direct mode of activation (Costanzo et al., 2008). An inverse relationship between (p)ppGpp levels and growth-related σ70 transcriptional activity at rRNA promoters has become clearer with the recent observations that (p)ppGpp0 cells show increased levels of Eσ70 and overproducing Eσ70 results in increased transcription from rRNA promoters, whereas genes involved in stress and amino acid biosynthesis are downregulated (Gummesson et al., 2009). Thus, it can be argued that the processes that disfavour the association of σ70 with core RNAP can result in the redistribution of RNAP from the housekeeping genes to the maintenance-associated promoters, which essentially is the physiological outcome of (p)ppGpp and DksA actions.
Anti-σ70 factor Rsd
Anti-sigma factors are proteins, which, by binding to the cognate sigma factors, inhibit their transcription activity (reviewed by Hughes & Mathee, 1998). Examples of some of the biochemically and structurally well-characterized anti-sigma factors include SpoIIAB and RsbW of B. subtilis, FlgM of Salmonella typhimurium, RseA of E. coli and RsrA of Streptococcus coelicolor (reviewed by Campbell et al., 2008). A majority of the anti-sigma factors regulate the activity of alternative sigma factors; however, an E. coli protein, Rsd (repressor of sigma D), has been shown to bind to σ70. Rsd was initially identified as a stationary-phase-induced protein, which could be copurified with σ70 of E. coli (Jishage & Ishihama, 1998). Pseudomonas aeruginosa-encoded AlgQ shows a high degree of amino acid homology to Rsd and also seems to be a functional homologue of Rsd as an AlgQ mutant of P. aeruginosa could be functionally complemented by Rsd (Ambrosi et al., 2005).
Initial studies had observed that the maximum intracellular level of Rsd during the stationary phase was only 20% of the σ70 level (Jishage & Ishihama, 1998), although a recent report suggests that the numbers of Rsd and σ70 molecules in E. coli during the stationary phase are similar (Piper et al., 2009). A significantly higher amount of σ70 was found to be associated with Rsd during the stationary phase compared with the exponential phase, although the amount of Rsd had only doubled during this period (Piper et al., 2009). An E. coli mutant lacking Rsd did not show impaired growth during the stationary phase (Jishage & Ishihama, 1999). Similarly, a deletion of rsd gene did not show a significant difference in the transcript levels of E. coli during the stationary phase as detected by DNA microarray analysis (Mitchell et al., 2007). However, overexpression of Rsd resulted in upregulation of some genes, the majority of which were known to be σ38 regulated (Mitchell et al., 2007). These observations suggested that the phenotypic effects of Rsd could only be seen in strains overexpressing Rsd and higher levels of intracellular Rsd favour stationary-phase transcription.
The rsd gene has been shown to be regulated by two promoters, P1 and P2, driven by σ38 and σ70, respectively (Jishage & Ishihama, 1999). Promoter mutation studies in a wild type and σ38 mutant background suggested that the P2 promoter makes a major contribution towards the synthesis of Rsd (Piper et al., 2009). Using a rsd: lacZ fusion in a (p)ppGpp0 background, some reduction in the β-gal levels was observed, indicating that this alarmone might play a role in Rsd expression (Jishage & Ishihama, 1999).
Rsd inhibits σ70 activity and favours transcription by alternative sigma factors
Although most of the studies on Rsd have focused on its ability to favour σ38-driven transcription, there are indications that overexpression of Rsd favours transcription mediated by other alternative sigma factors such as σ24, σ32 and σ54 (Jishage et al., 2002; Laurie et al., 2003; Costanzo et al., 2008). The evidence for a role of Rsd in promoting transcription by σ38 has come from studies involving the transcription of σ38-specific promoters in mutant E. coli lacking Rsd, or σ38 or by overexpressing Rsd or σ38. In transcriptional studies in vivo using lacZ fusions to σ38-driven promoters, it was observed that an Rsd mutant of E. coli showed a reduction in β-gal activity compared with the wild type (Jishage & Ishihama, 1999), while overexpression of Rsd or σ38 from a multicopy plasmid led to an increase in β-gal activity (Jishage & Ishihama, 1999; Mitchell et al., 2007), suggesting that increased levels of Rsd favour σ38-mediated transcription during the stationary phase. Does Rsd inhibit σ70-mediated transcription in vivo? By monitoring the effect of Rsd on a σ70-specific ompF promoter in a ompF: lacZ fusion, an E. coli strain lacking Rsd showed increased β-gal activity, which was suppressed by overexpression of Rsd or σ38 (Jishage & Ishihama, 1999). These effects were only seen during the stationary phase and not during the log phase. Expression of a mutant Rsd, which bound to σ70 with a higher affinity, was shown to enhance transcription from σ38 promoters, indicating that Rsd facilitates the redistribution of RNAP to σ38-specific promoters by sequestering σ70 (Mitchell et al., 2007). Overexpression of Rsd has been shown to favour transcription mediated by alternative sigma factors such as σ24, σ32 and σ54 in the (p)ppGpp0 background (Jishage et al., 2002; Laurie et al., 2003; Costanzo et al., 2008); however, information on the effects of Rsd on these alternative sigma factors in wild-type cells is lacking. Compared with the strong inhibition of σ70-driven transcription by the T4 bacteriophage-encoded anti-sigma factor AsiA, Rsd shows only modest effects on σ70 transcription in vitro and in vivo (Jishage & Ishihama, 1998; Pineda et al., 2004; Sharma & Chatterji, 2008). Rsd was shown to inhibit in vitro transcription from certain σ70 promoters (alaS), while there was no effect or a mild effect on other promoters such as lacUV5 or puvsX (Jishage & Ishihama, 1999; Pineda et al., 2004). The transcriptional profile of an Rsd overexpressing strain showed increased levels of transcripts only in a subset of σ38 promoters while other σ38 specific transcript levels remained unaffected (Mitchell et al., 2007). The characteristic features of promoters inhibited by Rsd are not well understood.
Mechanism of transcription inhibition by Rsd
In vitro and in vivo studies have suggested that Rsd binds to region 4 of σ70, although, unlike AsiA, the binding region of Rsd is not limited to region 4, but includes regions 2 and 3 as well (Patikoglou et al., 2007; Sharma & Chatterji, 2008; Yuan et al., 2008). Structural studies of Rsd-σ70 region 4 suggested that binding of Rsd to σ70 interferes with the core binding and recognition of −35 of E. coli promoters (Patikoglou et al., 2007). Rsd's ability to interact with multiple regions on σ70, including core binding regions, probably allows it to prevent the binding of σ70 to core RNAP, which in turn provides a better opportunity for stationary-phase-specific σ38 to form the Eσ38 complex. In this aspect, Rsd behaves like other anti-sigma factors (e.g. FlgM, RseA, SpoIIAB, etc.), which bind to multiple regions of the cognate sigma factors, thus preventing the holo enzyme formation (Campbell et al., 2008).
Although there is enough evidence to suggest that Rsd enhances σ38 transcription by sequestering σ70, poor inhibition of in vitro transcription, lack of an observable phenotype, no detectable change in the transcription profile of rsd mutants and reported weak interaction of Rsd with core and holo RNAP (Ilag et al., 2004) suggest that much more needs to be learnt to decipher the precise role and mechanism of inhibition by Rsd in regulating transcription during the stationary phase.
6S RNA: The RNAP-binding RNA
Although 6S RNA (coded by ssrS1) was discovered in 1970, it took 30 years to decipher its function and physiological role in E. coli (reviewed by Wassarman et al., 2007). It was observed that 6S RNA forms a specific complex with Eσ70 and not with E or σ70 (Wassarman & Storz, 2000; Wassarman, 2007). Bioinformatics analysis of bacterial genomes has predicted the presence of 6S RNA coding genes in a number of enterobacterial species (Barrick et al., 2005; Trotochaud & Wassarman, 2005). Additionally, 6S RNA from other bacterial species were identified by biochemical approaches involving total RNA analysis and coimmunoprecipitation with RNAP (Trotochaud & Wassarman, 2005). Partial sequencing of an abundant 185 nucleotide RNA from Bordetella pertussis led to identification of 6S RNA (Trotochaud & Wassarman, 2005). Two species of 6S RNA in B. subtilis were identified by coimmunoprecipitation with RNAP (Trotochaud & Wassarman, 2005). Both species of B. subtilis 6S RNA showed binding to the EσA holo enzyme.
Role of 6S RNA in survival during the late stationary phase
The lack of any observable phenotype associated with inactivation of ssrS1 in E. coli cells during growth in the log or the early stationary phase and also at temperatures ranging from 23 to 42 °C in a variety of carbon sources was partly responsible for difficulties in assigning any role to 6S RNA in bacterial physiology (Lee et al., 1985; Trotochaud & Wassarman, 2004). However, in growth competition experiments, the ssrS1 cells were unable to compete with the wild-type cells beyond 1 day, showing that the lack of 6S RNA leads to a growth disadvantage (Trotochaud & Wassarman, 2004). The ssrS1 gene is expressed throughout the life cycle of E. coli, but 6S RNA accumulates during the stationary phase and the number of 6S RNA molecules increases from 1000 to 10 000 (10-fold) during the stationary phase (Wassarman & Storz, 2000). The corresponding increase in 6S RNA in B. subtilis is 5–18 fold (Barrick et al., 2005). As against the (p)ppGpp and Rsd, which are upregulated during transition from active growth to the stationary phase or the early stationary phase (Jishage & Ishihama, 1998; Piper et al., 2009), 6S RNA accumulates maximally during the late stationary phase (Wassarman & Storz, 2000). This would suggest that 6S RNA probably regulates transcription during the late stationary phase, which is also reflected in its growth phenotype (Trotochaud & Wassarman, 2004). Based on the measurements of 6S RNA concentration in mutant E. coli cultures lacking stress regulators such as FIS, H-NS, LRP and StpA and results obtained from gel retardation and in vitro transcription assays, Neusser et al. (2008) concluded that these proteins might play a role in controlling transcription from ssrS1 promoter in a negative manner. It was also shown that (p)ppGpp does not regulate the generation of 6S RNA in E. coli under in vitro conditions (Neusser et al., 2008). As in the case of the rsd gene, ssrS1 is transcribed by σ70- and σ38-specific P1 and P2 promoters yielding 6S RNA of different lengths, which are cleaved by RNAseE and RNAseG to yield the mature 6S RNA (Kim & Lee, 2004). The differential transcription and processing is supposed to help in modulating the levels of 6S RNA in E. coli (Kim & Lee, 2004). Because the upregulation of 6S RNA during the stationary phase is not affected in a σ38 mutant (Wassarman & Storz, 2000), it shows that ssrS1 gene transcription is mainly regulated by σ70. Interestingly, 6S RNA bound to RNAP is capable of acting as a template to synthesize a product RNA, called pRNA (Wassarman & Saecker, 2006; Gildehaus et al., 2007). The process of pRNA generations is thought to be similar to transcription involving DNA templates in terms of abortive initiation, falling off of the sigma factor, etc. By observing the formation of 6S RNA–pRNA hybrids, the synthesis of pRNA was shown to take place during outgrowth of cells from the stationary phase, indicating that transcription from 6S RNA helps in its release from Eσ70, thus relieving the inhibition. The signal for this release is probably the higher concentration of NTP inside the cell during nutrient availability (Wassarman & Saecker, 2006). Recently, it was shown that upon pRNA synthesis, the 6S RNA–Eσ70 complex disintegrates and 6S RNA and pRNA undergo degradation (Wurm et al., 2010).
6S RNA inhibits σ70-mediated transcription and favours the transcription of stationary-phase genes
Binding analysis by cosedimentation and immunoprecipitation showed that 6S RNA interacts specifically with Eσ70, while UV crosslinking indicated that it binds to regions of σ70, β and β′ (Wassarman & Storz, 2000; Trotochaud & Wassarman, 2005; Gildehaus et al., 2007). In studies involving overexpression of 6S RNA, a mutant 6S RNA and 5S RNA in 6S RNA-deficient E. coli, it was shown that only 6S RNA coimmunoprecipitated with Eσ70 and inhibited transcription from a σ70-specific rsd P2 promoter in vivo and in vitro (Trotochaud & Wassarman, 2005). Although there have been indications of 6S RNA binding to and inhibiting σ38-driven in vitro transcription in purified RNAP preparations (Gildehaus et al., 2007), the physiological relevance of these observations is not clear. Overexpression of the ssrS1 gene in E. coli resulted in inhibition of transcription from the rsd P2 promoter, suggesting that 6S RNA inhibits σ70-driven transcription, while ssrS1 mutants showed higher transcription from the σ70-specific P2 promoter without any change in P1 transcription (Wassarman & Storz, 2000). Examination of transcription from other stationary-phase-specific promoters (e.g. bolA, hya and rsd P1) showed reduced transcription from some of these promoters in the absence of 6S RNA (Trotochaud & Wassarman, 2004). However, in some cases, 6S RNA overexpression-mediated reduction of transcription from σ70 promoters did not lead to a concomitant increase in σ38 activity (Kim et al., 2004), suggesting the involvement of other regulators in the transcription of certain σ38-specific promoters. Microarray data analysis has shown that almost half the genes expressed in E. coli are regulated by 6S RNA, thus making 6S RNA a global regulator of transcription (Cavanagh et al., 2008). Most of the inhibited promoters belong to the category of an extended −10 class (promoters with a TGn motif upstream to the −10 consensus sequence) or those with a poor −35 consensus sequence.
From the estimated high number of 6S RNA inside an E. coli cell (Fig 2), it can be predicted that the majority of the Eσ70 complexes will be trapped by 6S RNA. In this kind of situation, how does the σ38 transcription increase? There is no evidence to suggest that binding of 6S RNA to Eσ70 leads to destabilization of holo RNAP or an increase in the affinity of σ38 to core RNAP.
Abundance of various molecules involved in the transcriptional regulation and redistribution of RNAP in Escherichia coli. (a) Intracellular abundance (steady-state levels) of various molecules during active growth and during starvation or the stationary phase. The sources of the data are as follows: E, σ70, σ38 and Rsd, Piper et al. (2009); σ24 and σ32, Grigorova et al. (2006); σ54, Maeda et al. (2000); (p)ppGpp, Ryals et al. (1982); Cashel et al. (1996); DksA, Rutherford et al. (2007); 6S RNA, Wassarman & Storz (2000). When required, the number of molecules per cell was calculated as described by Grigorova et al. (2006). (b) A diagram depicting a possible scenario involving the redistribution of RNAP to maintenance and survival gene promoters during starvation in E. coli. Only σ70 and some of the alternative sigma factors (σ38, σ32 and σ24 and σ54) whose activity has been shown to be influenced by ppGpp, DksA, Rsd or 6S RNA have been shown. For the sake of simplicity, RNAP and the other molecules have been shown in approximate stoichiometric amounts. ND, not detectable.
Mechanism of inhibition: 6S RNA competes for binding to region 4.2 of σ70
The secondary structure of 6S RNA, which is conserved across bacterial species, has been shown to be critical for its binding activity as mutants lacking the structure were found to be deficient in interacting with Eσ70 (Trotochaud & Wasserman, 2005). All σ70 promoters are not equally sensitive to 6S RNA inhibition. The prominent feature of the promoters of 6S RNA regulated genes is the presence of a weak −35 region and an extended −10 region (Cavanagh et al., 2008). Yet, the inhibition of transcription by 6S RNA does not seem to be limited to weak promoters alone and has been seen in promoters with varying strengths (Cavanagh et al., 2008). It has been demonstrated that 6S RNA binds to the active site of RNAP and inhibits transcription from certain promoters by preventing the access of RNAP to the promoter regions on DNA (Wassarman & Saecker, 2006; Gildehaus et al., 2007). In vitro binding studies with RNAP bound to truncated σ70 fragments showed that region 4.2 is required for binding to 6S RNA (Cavanagh et al., 2008), although earlier studies had indicated binding to other regions of RNAP as well (Wassarman & Storz, 2000). 6S RNA thus competes with the −35 promoter region for binding to region 4.2 of σ70. Further characterization of this interaction revealed that the 6S binding region consists of positively charged residues in the helix–turn–helix motif of region 4.2 of σ70, thus suggesting that promoter recognition residues and 6S RNA-binding residues of region 4.2 are distinct (Klocko & Wassarman, 2009). Many of the 6S RNA-binding residues of σ70 have been known to interact with protein activators that bind to region 4 of σ70 (Decker & Hinton, 2009). The availability of structural information on the 6S–RNAP complex will unveil the exact binding mechanism of 6S RNA.
Other factors influencing σ70 activity
Bacterial DNA is present in a negatively supercoiled state under normal growth conditions. It has been observed that there is a change in the superhelicity of DNA upon changes in environmental conditions, which can lead to the modulation of the transcription profile of the organism (Travers & Muskhelishvili, 2005). Transcription mediated by different sigma factors is affected differentially by the supercoiling status of DNA, with negative supercoiling generally favouring σ70-mediated transcription and relaxed DNA supporting σ38-mediated transcription (Travers & Muskhelishvili, 2005). Nucleoid-associated proteins (e.g. Fis, IHF and H-NS, etc.) can bring about changes in local DNA topology, thus neutralizing the effects of the overall changes in supercoiling, resulting in a variety of effects in different regions of the chromosome (McLeod & Johnson, 2001). In addition to this, the presence of a large number of transcription activators and repressors adds to the complexity of transcription initiation in bacteria (Grainger & Busby, 2008; Balleza et al., 2009). Potassium glutamate, which accumulates inside the cell in response to high osmolarity, exerts differential effects on transcription and strongly inhibits transcription from salt-sensitive ribosomal promoters (Gralla & Vargas, 2006). Acetate has also been shown to modulate transcription from σ70- and σ38-driven promoters to varying extents (Rosenthal et al., 2008).
Concluding remarks and future directions
Because the majority of the core RNAP during active growth is bound to σ70, upon entry into the stationary phase, E. coli cells use multiple strategies to redistribute RNAP to promoters of maintenance- and survival-related genes. This primarily involves shifting the focus away from σ70-driven proliferation-related transcription. Although (p)ppGpp and DksA regulate transcription by direct inhibition and activation, there is enough evidence to suggest that a significant increase in the transcription activity of alternative sigma factors occurs because of the increased availability of core RNAP as a result of destabilization of Eσ70 complexes or by making σ70 less competitive by (p)ppGpp and DksA. To prove whether the presence of (p)ppGpp and DksA can affect the affinity of σ70–core RNAP interaction will require studies on kinetics of binding of σ70 to core RNAP in the presence of (p)ppGpp and DksA. Rsd, on the other hand, seems to prevent the binding of σ70 to core RNAP, thus enabling transcription by σ38 during the stationary phase. Although 6S RNA inhibits transcription at a large number of σ70 promoters, the mechanism of 6S RNA-mediated increase in stationary-phase transcription is not understood. Because 6S RNA traps σ70 bound core RNAP during the stationary phase, it is unlikely that the increase in transcription by σ38 is because of the availability of increased amounts of core RNAP. In contrast to the growth phenotype shown by (p)ppGpp0 or dksA mutants, the phenotypes of rsd and ssrS1 are quite mild. Does it mean that (p)ppGpp and DksA are the major regulators involved in the redistribution of RNAP during starvation, while Rsd and 6S RNA are required for fine tuning of the (p)ppGpp effects? The interdependence and interactions of these effectors with each other are not understood. However, upregulation of these effectors during different phases of growth, ranging from the late exponential to the late stationary phase, suggests the existence of a mechanism of temporally regulated transcriptional switching. The relative increase in the total number of E, sigma factors and these molecules during starvation has been depicted in Fig. 2a.The outcome of modulation of σ70 activity by these effectors, leading to a shift towards transcription of maintenance-related genes, has been diagrammatically represented in Fig. 2b.
Both 6S RNA and (p)ppGpp have been shown to be present in Gram (+) and Gram (−) bacteria, whereas the presence of Rsd seems to be restricted to some Gram (−) bacteria only. As bioinformatics approaches may not be able to predict anti-σ70 factors in Gram (+) bacteria, experimental approaches are required to detect these proteins. Understanding the role of DksA-like proteins and other secondary channel binding proteins in the regulation of transcription in various bacteria also needs further investigation. The study of ssrS1 and rsd mutant or overexpression phenotypes and transcription profiles in strains lacking (p)ppGpp and or DksA will help in understanding the relative contributions of these effectors towards the modulation of σ70 activity. An integrated approach for studying the role of (p)ppGpp, DksA, Rsd and 6S RNA in transcriptional regulation will unravel their contribution towards adaptation of bacteria to environmental changes.
We are grateful to Drs Anand Kumar and Kakoli Mukherjee for critically reading the manuscript. We greatly appreciate the inputs provided by the anonymous reviewers for improving the manuscript.
(2007) Studies on the function of the riboregulator 6S RNA from E. coli: RNA polymerase binding, inhibition of in vitro transcription and synthesis of RNA-directed de novo transcripts. Nucleic Acids Res 35: 1885–1896.
(2008) Differential mechanisms of binding of anti-sigma factors Escherichia coli Rsd and bacteriophage T4 AsiA to E. coli RNA polymerase lead to diverse physiological consequences. J Bacteriol 190: 3434–3443.
(2008) Still looking for the magic spot: the crystallographically defined binding site for ppGpp on RNA polymerase is unlikely to be responsible for rRNA transcription regulation. J Mol Biol 377: 551–564.