Dissection of recognition determinants of Escherichia coli σ³² suggests a composite −10 region with an ‘extended −10’ motif and a core −10 element

Summary

σ³² controls expression of heat shock genes in Escherichia coli and is widely distributed in proteobacteria. The distinguishing feature of σ³² promoters is a long −10 region (CCCCATNT) whose tetra-C motif is important for promoter activity. Using alanine-scanning mutagenesis of σ³² and in vivo and in vitro assays, we identified promoter recognition determinants of this motif. The most downstream C (−13) is part of the −10 motif; our work confirms and extends recognition determinants of −13C. Most importantly, our work suggests that the two upstream Cs (−16, −15) constitute an ‘extended −10’ recognition motif that is recognized by K130, a residue universally conserved in β- and γ-proteobacteria. This residue is located in the α-helix of σDomain 3 that mediates recognition of the extended −10 promoter motif in other δs. K130 is not conserved in α- and δ-/εproteobacteria and we found that σ³² from the α-proteobacterium Caulobacter crescentus does not need the extended −10 motif for high promoter activity. This result supports the idea that K130 mediates extended −10 recognition. σ³² is the first Group 3σ shown to use the ‘extended −10’ recognition motif.

Introduction

Bacterial transcription initiation is mediated by RNA polymerase holoenzyme, consisting of a 5-subunit core RNA polymerase (α₂ββ’ω) and a dissociable subunit, σ. σs contain many of the promoter recognition determinants (Gross et al., 1998). Most transcription in exponentially growing cells is initiated by RNA polymerase holoenzyme carrying the housekeeping σ, called σ⁷⁰ in Escherichia coli. Alternative ss mediate transcription of regulons activated by specific stress conditions, or by growth or developmental transitions. The σ⁷⁰ family of proteins has been divided into four groups, based on their phylogenetic relationships and modular structure (Lonetto et al., 1992; Gruber and Gross, 2003). Group 1 σs are the essential housekeeping ss and contain four domains, of which the first is unique to the Group 1 σs. Group 2 ss are most closely related to Group 1 σs, but are not essential. Most Group 3 σs contain Domains 2–4, and the Group 4 σs contain only Domains 2 and 4 (Gruber and Gross, 2003).

Promoters recognized by σ⁷⁰ and other Group 1 σs have been extensively studied. Promoters are multipartite, and can have the following potential elements to mediate recognition: (i) the ‘UP element’, generally located from −61 to −41 relative to the transcription start point (Ross et al., 1993; Gourse et al., 2000), (ii) the −35 region, a hexamer (TTGACA) centred about 35 bp upstream of the transcription start point (Gross et al., 1998), (iii) the ‘extended −10 motif’ (TG) located immediately upstream of the −10 region (Mitchell et al., 2003), (iv) the −10 region, a hexamer (TATAAT) centred about 10 bp upstream of the transcription start (Helmann and deHaseth, 1999) and (v) the discriminator region (GGGa) located downstream of the −10 region of the promoter (Feklistov et al., 2006; Haugen et al., 2006). With the exception of the UP element, recognized by the C-terminal domain of α, promoter elements are recognized by the σ subunit. The −35 element is recognized by Region 4.2 in sDomain 4; the extended −10 element by an α-helix in σDomain 3; the −10 by Region 2.4 in σDomain 2 and the discriminator by σRegion 1.2 (Campbell et al., 2002; Feklistov et al., 2006; Haugen et al., 2006). The number of elements in promoters is variable.

Promoters recognized by the housekeeping σ in E. coli can be roughly divided into those with an extended −10 motif (‘extended −10 promoters’) and those without this motif. Extended −10 promoters have a lower match to the consensus −35 motif than do those lacking the extended −10 motif (Mitchell et al., 2003). Moreover, at least in vitro, extended −10 promoters can function without σRegion 4.2, the −35 recognition region (Kumar et al., 1993; Ponnambalam et al., 1986; Minakhin and Severinov, 2003; Young et al., 2004). Together these results imply that the extended −10 may functionally substitute for the −35 promoter region. Interestingly, in Bacillus subtilis and other Gram-positive bacteria, this distinction does not hold: > 50% of the promoters have high consensus to all three elements (−35, extended −10 and −10) (Graves and Rabinowitz, 1986; Bashyam and Tyagi, 1998; Helmann, 1995; Sabelnikov et al., 1995).

The ‘discriminator’ region serves as an ancillary element in promoters utilized by the E. coli housekeeping s. It is most important for stable RNA promoters, which characteristically form unstable complexes with RNA polymerase; their instability is largely attributable to a very disfavoured discriminator sequence (Haugen et al., 2006). However, for the thermophilic bacterium Thermus aquaticus housekeeping σ, the discriminator, like the extended −10 motif, can substitute for a −35 element (Feklistov et al., 2006; Barinova et al., 2008). The UP element enhances promoter activity, especially in weaker promoters (Ross et al., 1998; Miroslavova and Busby, 2006).

Promoters of the Group 3 σs and their σ recognition determinants have not been nearly as well studied. However, one contrast between many Group 3 σ promoters and those recognized by housekeeping σs has been noted. Promoters of the Group 3 σs often have a relatively long −10 promoter region, whose upstream end is GC-rich, separated by only 13 or 14 bp from the −35 region of the promoter (Table 1). This configuration suggests the possibility that the long −10 promoter region includes an ‘extended −10’ region adjacent to the −10 region of the promoter. That known ‘extended −10's include either G or C nucleotides (TG: Group 1 σs; T/GC σ^S, a Group 2 σ) would be consistent with this idea (Helmann, 1995; Becker and Hengge-Aronis, 2001; Mitchell et al., 2003). However, studies to date, on promoters recognized by Group 3 σs provided no evidence for this idea. Neither of the promoters recognized by σ^E and σ^H, two B. subtilis Group 3 σs, had been found to have an extended −10 motif (Daniels et al., 1990; Tatti et al., 1991). The major goal of this work was to determine whether an extended −10 recognition element is used by a Group 3 σ.

Table 1.

Consensus sequence of Group 3 s promoters.

	Promoter sequence
σs	–35	–10	Reference
E. coli σ70	-TTGACA--------------TG-TATAAT		Gross et al. (1998)
E. coli σ32	-TTGAAA--------------CCCCAT-T-		Nonaka et al. (2006)
E. coli σ28	--TAAAGTTT-----------GCCGATAA-		Zhao et al. (2007)
B. subtilis σD	--TAAAG--------------GCCGATAT		Serizawa et al. (2004)
B. subtilis σB	-GTTTAA--------------GGGTAT---		Petersohn et al. (2001)
B. subtilis σF	-GTATWW--------------GG--A---T		Amaya et al. (2001)
B. subtilis σE	-TCATATT--------------CATA-A-T		Eichenberger et al. (2003)
B. subtilis σH	RNAGGAWWW-----------R- - GAATWW-		Britton et al. (2002)

R,Aor G; W,Aor T.

We chose to determine whether E. coli σ³² contains an extended −10 element both because we can build on the extensive studies already performed on σ³² and because it is widely distributed in proteobacteria (Nakahigashi et al., 1995). Several global studies of promoters recognized by σ³² suggested that the consensus sequence of σ³² promoter is TTGAAA – N_13-14 – CCCCATNT (Zhao et al., 2005; Nonaka et al., 2006; Wade et al., 2006). A careful mutational analysis of P_groE, a strong σ³² promoter, conducted by deHaseth and collaborators, validated that the conserved −35 region and the entire −10 region, including the tetra C-motif (−16C to −13C), are important for promoter activity (Wang and deHaseth, 2003). More recently, deHaseth has examined promoter recognition determinants in σ³², selecting residues for substitution based on known structures of various ss and by comparison with known promoter recognition determinants in σ⁷⁰ (Kourennaia et al., 2005). These studies showed that recognition of the −35 region was mediated by Region 4.2 residues in σ³² analogous to those used by s⁷⁰. In contrast, residues in Region 2.4 analogous to those in σ⁷⁰ did not mediate recognition of the −10 element. However, W108, at the border between Regions 2.3 and 2.4, located at the end of the α-helix proposed to mediate −10 recognition, did affect recognition of −13C. This work established that the most downstream C is part of the −10 region of the promoter, but leaves unsettled whether the upstream Cs are an extended −10 motif.

σ³² controls the heat shock response and is highly regulated to maintain homeostasis (Guisbert et al., 2008). This presents a challenge when investigating promoter recog- nition of σ³². Both the amount and activity of σ³² can be adjusted by negative feedback loops when cellular amounts of key members of the regulon change. As a consequence, when a mutant σ³² defective in promoter recognition produces insufficient amounts of key regulon members, the regulatory system compensates by overex-pressing the defective σ³², thereby obscuring promoter recognition defects. To address this problem, we developed a new assay system, using an allele of σ³² that is no longer subject to regulation (Yura et al., 2007). Using this assay system, newly constructed libraries of promoter mutants and an alanine scanning mutagenesis library encompassing Region 2.4 in Domain 2 and Region 3.0 in Domain 3 of σ³², we examined recognition of the −10 region. Our studies indicate that the −10 region is a composite element. The two upstream Cs are recognized by Domain 3 and thus likely correspond to an extended −10 element, whereas the downstream C is recognized by two Region 2.4 residues and is therefore part of the −10 element. Thus, in contrast to the σ⁷⁰ paradigm, E. coli σ³² promoters require both a canonical −35 region and the extended −10 promoter region for high activity. Interestingly, in some groups of proteobacteria, the σ³² residue in Domain 3 that recognizes the upstream Cs is missing. Investigation of one such σ³² reveals that it no longer needs the extended −10 promoter region for high activity.

Results

In vivo assay of σ³² activity

A robust in vivo assay system for the activity of mutant σ³² proteins must be solely dependent on the properties of the mutant protein, and impervious to the regulatory loops that can adjust the amount of σ³² to compensate for its activity deficit. To achieve these goals, the previous assay system used a host background with two attributes: (i) a chromosomal rpoH locus that encoded a σ³² with very reduced function and (ii) a partially disabled regulatory circuit [by substituting dnaK756 for wild-type (wt) dnaK] (Wang and deHaseth, 2003). This system allowed assay of altered function σ³² above a background of expression mediated by the endogenous locus. Our new assay system goes further; it eliminates chromosomal σ³² and completely disables the σ³² regulatory loop. The assay system consists of a host strain lacking σ³² and two plasmids, one supplying mutant σ³² alleles, and the other supplying σ³² promoter mutants of the groE and grpE σ³² promoters fused to the lacZ reporter (Fig. 1A). Overexpression of GroESL is known to permit survival in the absence of σ³² (Kusukawa and Yura, 1988). Thus, the host strain can survive in the absence of σ³² because the σ³² promoter of the groESL operon is replaced with the regulatable σ⁷⁰ P_BAD promoter and GroESL can be over-expressed constitutively by addition of arabinose in medium. The plasmid-encoded mutant σ³² alleles are under the control of an IPTG-inducible promoter and were constructed in a σ³² backbone that contains the I54N substitution, which renders σ³² impervious to regulation of its activity or stability (Yura et al., 2007). In this assay system, the wt groE and grpE promoters (Fig. 1B, P_groE, P_grpE, containing sequences from −44 to −9 of each promoter) had a basal β-galactosidase activity of < 1 and 3 Miller units respectively. Following induction of I54N σ³², the activity of these promoters were ~130 and 90 Miller units respectively. Thus, this assay system has excellent signal to noise ratio. Moreover, as discussed below, the I54N σ³² variant does not alter promoter recognition.

Fig. 1.

Components of in vivo assay systems used in this study.

A. Activity of σ³²-dependent promoters in the presence of σ³² variants was determined by β-galactosidase assay from the ΔrpoH strain CAG57101 carrying the plasmids pSAKT32 and pQF50K. σ³² expression was induced from pSAKT32 and derivatives (pBK116–pBK151), which has a p15A replication origin and rpoH under control of the lac promoter. Note that most of the σ³² derivatives contain the additional substitution I54N that improves stability and decreases negative regulation. The reporter plasmid, pQF50K and derivatives (pBK501–pBK549), has a pMB1 replication origin and lacZ under control of σ³² promoter.

B. Sequences of σ³²-dependent promoters used in this study. Only wt sequences of each promoter are shown: single- or double-base changed promoter mutants were constructed from these parent promoters. The native sequences of the groE and grpE promoter regions are shown in capital letters; vector sequences are in lower case; BglII and XbaI cloning sites are underlined; −35 and −10 regions, and the

Comparison of base dependence of σ³² promoters with different spacing

Virtually all σ³² promoters have a spacing of either 13 or 14 bp between the −35 and −10 region, with a 14 bp spacing being optimal (Wang and deHaseth, 2003; Nonaka et al., 2006). Before analysing the effects of mutant σ³²s on −10 promoter recognition, it was important to determine whether promoters with each spacing showed similar dependence on positions in the −10 region. We used both a bioinformatics and experimental approach to examine this issue. We classified the 50 σ³² promoters validated in our previous work (Nonaka et al., 2006) by their spacer length and then aligned them to see whether spacer length affected consensus. Our new alignment showed only one difference between promoters with a different spacer length: the upstream CC motif was highly conserved in the 16 promoters with a 13 bp spacer; the second C of this motif is less conserved in 27 promoters with a 14 bp spacer (Fig. 2A). To examine this issue experimentally, we created two identical libraries that are mutant at every conserved position in the −10 region, one in the grpE promoter (13 bp spacing) and a second in the groE promoter (14 bp spacing) (Fig. 1B) and examined the relative transcription of each promoter variant in both backgrounds (Fig. 2B). This permitted two conclusions: (i) the relative effects of single mutant changes in the −10 region are similar regardless of spacing and (ii) the promoter preference of I54N σ³² on the groE library (Fig. 2B) is virtually identical to that reported by deHaseth using wt σ³² (Wang and deHaseth, 2003), eliminating any concern that using the I54N σ³² variant might perturb results.

Fig. 2.

Sequence conservation in the −10 region of E. coli σ³² promoters and effects of single-base mutations at the −10 region tetra-C motif on activities of groE or grpE promoter.

A. Sequence logos (http://weblogo.berkeley.edu; Crooks et al., 2004) denoting sequence conservation within the −10 region of 13 bp spacer and 14 bp spacer promoters. Sequences of 16 σ³² promoters with a 13 bp spacer and 27 promoters with a 14 bp spacer (previously identified by Nonaka et al., 2006) were used to generate sequence logos of the −10 region.

B. Activity of each mutant groE or grpE promoter with wt σ³² is shown as a percentage of measured β-galactosidase activity of its respective wt promoter. Assay strains are as described in Fig. 1A. Promoter mutations and positions are shown on the x-axis. All values are averages of at least three independent experiments; error bars indicate one standard deviation; black bars, groE promoter activities; grey bars, grpE promoter activities.

General strategy for in vivo analysis of mutant σs and overview of results

The major goal of this work was to determine whether part of the tetra-C motif in the −10 promoter region is an extended −10 motif. In both Group 1 and Group 2 σs, recognition of this motif is carried out by an α-helix at the N-terminus of Domain 3. By analogy, this same helix should mediate extended −10 recognition in σ³². We therefore performed extensive alanine substitution mutagenesis in this region of Domain 3 of I54N σ³². In addition, we mutagenized the RpoH box region, a σ³²-specific insertion in Region 3.0 in Domain 3 that is the most highly conserved region among all σ³² orthologues. Finally, we remade the Region 2.4 substitutions initially assayed by deHaseth and tested them in our new assay system to see if the improved resolution of this system provided new results. In addition, we tested a few additional substitutions in this region (Fig. 3A). Assays of transcriptional activity on both wt P_groE and P_grpE revealed that with the exception of R140A, each mutant s showed some impairment of transcriptional capacity, ranging from mild to severe (Fig. 3B).

Fig. 3.

Effects of single amino acid substitutions in Regions 2.4 and 3.0 of σ³² on activity of the groE and grpE promoters.

A. Amino acid sequence of Regions 2.3, 2.4 and 3.0 of σ³². Numbers at each end of the sequence indicate amino acid positions. Single-amino acid substitutions of σ³² used in this study are shown in light grey below the main sequence. The location of the RpoH box is denoted as a grey box (Nakahigashi et al., 1995); positions of the αhelices in each region are shown above the sequence. These data are derived both from structural data indicating that these s subregions are virtually identical in σ⁷⁰, a Group 1 σ, and σ²⁸, a Group 3 σ (Campbell et al., 2002; Sorenson et al., 2004), and secondary structure predictions for σ³² that are completely consistent with these assignments (E. Campbell, data not shown).

B. Activity of the σ³² Regions 2.4 and 3.0 mutants at the wt groE and grpE promoters. β-galactosidase activity driven by each mutant σ³² from the wt groE or grpE promoters is shown as a percentage of that driven by wt σ³². Assay strains are as described in Fig. 1A. The different σ³² amino-acid substitutions and their locations are shown on the x-axis. Note the effects of several σ³² substitutions (W108R, A111G, A111E, W121A and from L135 to K141) were not measured at the grpE promoter. All values are averages of at least three independent experiments; error bars indicate one standard deviation; black bars, groE promoter activities; grey bars, grpE promoter activities.

We next determined whether any of these mutant ss were specifically impaired in their interaction with a particular base in the promoter, using a well-validated allele-specific suppression criterion that has been applied to identify such interactions in both transcription factors and in other σs (Ebright et al., 1984; Ebright, 1986; Hochschild and Ptashne, 1986; Hochschild et al., 1986; Siegele et al., 1989). Basically, if the residue in σ that recognizes the consensus base at position X in the promoter is altered, then expression mediated by this mutant σ should be relatively unaffected by base changes at position X, but as sensitive as wt σ to changes at all other positions in the promoter. In the best-case scenario, promoters having a mutation at position X will be transcribed equivalently to the wt promoter by the s mutant lacking the residue mediating recognition.

To this end, we assayed the wt σ and all mutant ss on a P_groE library in which all positions except N in the −10 region CCCCATNT had been changed to every other base and on a P_grpE library in which each of the four Cs had been changed to every other base. Wt and mutant ss showed equivalent decreases at almost all promoters assayed, as expected (data not shown). However, several base-specific interactions with the tetra-C motif were detected: two amino acids in Region 2.4 recognized the downstream C (−13; Fig. 4) and an amino acid in the N-terminal helix of Domain 3 recognized the two upstream Cs (−16, −15; Fig. 5). Thus the tetra-C motif is a hybrid element, whose upstream end is an extended −10 motif and whose downstream end is part of the core −10 region.

Fig. 4.

Effects of substitutions in Region 2.4 of σ³² in suppressing promoter mutations at positions −14 and −13. A. Effect of W108A σ³² and W108R σ³² on the groE promoter with single-base changes at the −14 or −13 positions.

B. Effect of E112A σ³² on groE and grpE promoters with single-base changes at the −13 position.

C. Effect of E112A σ³² on groE promoter with a single-base changes at positions −14, −15 and −16. β-galactosidase activity driven by each σ³² variant from mutant groE or grpE promoters is shown as a percentage of β-galactosidase activity driven by each σ³² variant from the wt promoter. Assay strains are as described in Fig. 1A. Below the x-axis, mutation and position indicate the base change at the different promoter positions; promoter indicates the groE or grpE promoter; σ³² indicates the σ³² wt or mutant derivative. All values are averages of at least three independent experiments; error bars indicate one standard deviation.

Fig. 5.

Effect of substitution K130A in Region 3.0 of σ³² in suppressing promoter mutations at positions −16 and −15. A. Effect of wt and K130A σ³² on promoters with single-base changes at position −16 or −15.

B. Effect of K130A σ³² on groE and grpE promoters with single-base changes at other positions in the promoter.

C. Effect of wt and K130A of σ³² on promoters with double-base changes at positions −16 and −15. β-galactosidase activity driven by each σ³² variant on mutant groE or grpE promoters is shown as a percentage of β-galactosidase activity of each σ³² variant from the wt promoter. Assay strains are as described in Fig. 1A. Below the x-axis, mutation and position indicate the base change at the different promoter positions; promoter indicates the groE or grpE promoter; σ³² indicates the σ³² wt or mutant derivative. All values are averages of at least three independent experiments; error bars indicate one standard deviation.

Identification of base-specific interactions between Region 2.4 of σ³² and the −10 region of promoter

A previous study implicated W108 in recognition of −13C (Kourennaia et al., 2005). However, somewhat unexpectedly, they found that C-13T was not suppressed by W108A (Kourennaia et al., 2005). Results in our new assay system paint a simpler picture. W108 recognizes −13C: W108A σ³² is almost completely insensitive to changes at −13 (Fig. 4A). Suppression is specific as it is not manifest on promoters with changes at −14C (Fig. 4A) or at any other promoter variant tested (data not shown). Interestingly, the W108A mutant showed almost twofold higher activity on the C-13A and C-13T promoters than on the wt promoter. Possibly, W108A may make adventitious contacts at C-13A/T promoters that are not possible in wt σ³², leading to the increased activity. Consistent with this idea, substituting a bulky residue at position 108 (W108R) led to suppression but not increased activity at C-13A/T (Fig. 4A). Both the W108A and R substitutions also specifically suppressed −13C changes in grpE promoter variants (data not shown).

Our results also implicate E112 in −13C recognition. E112A decreased expression from the wt promoter only about 20–40% (Fig. 3B), but partially suppressed promoter mutations at −13C, especially C-13A and C-13G (Fig. 4B). Suppression is specific as E112A did not suppress promoter mutations at C-14, −15 or −16 (Fig. 4C). E112 may participate with W108 in recognizing −13C. Alternatively, E112 may inhibit other nearby residues from interacting with the −13C position. 112A might allow these adventitious interactions and thereby suppress the deleterious effect of base changes at −13C.

Identification of a base-specific interaction between Region 3.0 of σ³² and the −10 region of promoter

Of the 14 residues mutated in Domain 3, only K130 was identified as mediating a base-specific interaction. K130A had a significant defect in transcribing both P_groE and P_grpE (about 25% activity of wt σ³²; Fig. 3B) and suppressed all mutational changes at −16C and −15C to a variable extent (Fig. 5A). C to A changes at both positions were completely suppressed in both promoters, exhibiting the same activity as the wt promoter. C to G changes were either completely or almost completely suppressed and C to T changes were moderately suppressed. Suppression is specific as K130A did not suppress promoter mutations at −13C or −14C (Fig. 5B). Taken together, these results suggest that the K130 residue makes base-specific contacts with both −16C and −15C. To test this idea, we determined whether K130A was able to suppress promoters with base changes at both −16C and −15C (Fig. 5C). K130A suppressed these mutations effectively in both groE and grpE promoters. Importantly, the CC to AA change was almost completely suppressed, consistent with the fact that each individual change was also completely suppressed (Fig. 5C). As expected, the double mutant promoters were expressed very poorly in cells with wt σ³² (Fig. 5C). The fact that the activity of the double mutants was less than that of the K130A mutant could result from negative interactions of the changed bases with K130; changing K130 to A could alleviate these clashes. Alternatively, other residues might be able to partially substitute for K130. K130 is located in the N-terminal α-helix of Domain 3. Residues in this helix mediate extended −10 recognition in all Group 1 and Group 2 ss examined (Barne et al., 1997; Becker and Hengge-Aronis, 2001; Sanderson et al., 2003). This leads us to conclude that −16, −15 CC constitute an extended −10 recognition motif.

Do all proteobacteria have σ³² promoters with an extended −10 motif?

σ³² orthologues are widely distributed in proteobacteria and one would imagine that residues mediating recognition would be highly conserved. Indeed, W108 is universally conserved in all proteobacteria, consistent with its proposed role in −13C recognition. In contrast, K130 is universally conserved in β- and γ-proteobacteria, but is not conserved in the α- and δ-/ε-groups (Fig. S1). It is noteworthy that the amino acids positioned homologously to K130 in the α- and δ-/ε-groups of σ³² are alanine, serine or glutamine, which are unlikely to substitute for K130 in mediating recognition. Intriguingly, it has been reported that CC motif of the upstream −10 region is less well conserved in the two α-proteobacteria examined, Caulobacter crescentus and Rhodobacter sphaeroides (Green and Donohue, 2006; McGrath et al., 2007), than in the γ-proteobacteria. Together, these results raised the possibility that α- and δ-/ε-proteobacteria do not utilize an extended −10 motif. We tested this idea by examining the −10 promoter recognition properties of C. crescentus σ³², which has an alanine at the position comparable to K130 (Fig. 6A). The assay system used was the one described in Fig. 1A except that C. crescentus σ³² was the sole source of σ³² and it was not modified with the I54N substitution. Instead, we disabled the regulatory loops that control the amount and activity of σ³² by using an assay strain in which wt dnaK was replaced by dnaK756. This alteration in DnaK prevents it from participating in feedback inhibition and degradation of σ³², thereby disabling the regulatory loops (Georgopoulos et al., 1979; Straus et al., 1990; da Silva et al., 2003). C. crescentus σ³² was produced at a level comparable to that of E. coli σ³² (data not shown) and is generally less sensitive than E. coli σ³² to mutations at the −10 region (compare Figs 6B and 2B). Most striking is that C. crescentus σ³² has little or no sequence selectivity at the −16 and −15 positions, with even a promoter mutated at both −16C and −15C showing only slightly less activity than the wt promoter (Fig. 6B). Thus, loss of the conserved K130 residue correlates with loss of extended −10 recognition.

Fig. 6.

Recognition of the −10 region of groE promoter by C. crescentus σ³².

A. Amino acid sequence alignment of E. coli σ³² Regions 2.4–3.0 with the corresponding region of C. crescentus σ³². Amino acids shown to interact with specific promoter bases in E. coli σ³² are shown in bold.

B. Activities of C. crescentus σ³² on mutant E. coli groE promoters are shown as a percentage of measured β-galactosidase activity for wt groE promoter driven by C. crescentus σ³². CAG57102 strain was used for this assay with plasmid pBK151 expressing C. crescentus σ³² and derivatives of pQF50K_groE carrying the groE promoter lacZ reporter. Promoter mutations and positions are shown on the x-axis. All values are averages of at least three independent experiments; error bars indicate one standard deviation.

It is also interesting to note that E112, implicated by our results in Region 2.4 recognition of −13C, is universally conserved in β- and γ-proteobacteria, but not in the α- and δ-/ε-groups (Fig. S1). The amino acid corresponding to E112 is serine in C. crescentus σ³² and C. crescentus σ³² showed the same activity on the C-13T promoter variant as on the wt promoter (Fig. 6B). Possibly a serine at this position permits recognition of both T and C at the −13 position.

In vitro effect of the K130A substitution in E. coli has little or no sequence ³² and C. crescentus s³² on transcription

We performed single-round in vitro transcription assays to examine the promoter preferences of E. coli and C. crescentus σ³² as well as the K130A variant of E. coli has little or no sequence ³². The linear templates were based on P_groE and were identical to those used for the in vivo assay. In addition to the test template, each assay also contained a competitor template to serve as an internal control. The competitor template encodes wt P_groE and gave a longer transcript than the test template (Fig. 7A). Whereas wt E. coli σ³² initiates transcription from the wt promoter about sixfold better than a promoter mutated with A at positions −16 and −15, the K130A mutant initiates transcription from each promoter equivalently (Fig. 7B). Additionally, C. crescentus shows only a slight defect on the doubly mutant promoter as compared with the wt promoter (Fig. 7B). That the in vivo expression phenotype is reproduced in vitro argues that it is a direct effect of the particular σ³² present during transcription.

Fig. 7.

Effect of substitution at K130 of σ³² on promoter recognition in vitro. Single-round in vitro transcriptions were performed on linear DNA templates containing wt or −16,−15 AA mutant groE promoters by RNA polymerase containing E. coli wt σ³², K130A σ³² or C. crescentus s ³²(CCRσ³²).

A. In vitro transcription from groE promoter variants with σ³² variants. Upper bands in each lane are from long template of wt groE promoter (served as internal control); lower bands are from short template of either wt or −16,−15 AA mutant groE promoter (see Experimental procedures); bottom bands (end-labelled 35 nt oligomer) are shown as loading control.

B. Quantification of the in vitro transcripts to show the effect of substitution at K130 position on promoter recognition. The bars indicate the relative transcription from −16, −15 AA mutant promoter as a percentage transcript from wt promoter for each σ³² variant. Each value was calculated as follows: (i) each short transcript was expressed as a fraction of total transcript in the lane (long + short transcript) and called normalized short transcript and (ii) normalized short transcript from −16, −15 AA promoter was divided by normalized short transcript from the wt promoter. All values are averages of three independent experiments; error bars indicate one standard deviation.

Effect of UP element on promoter recognition by σ³²

The studies described thus far have been performed on P_groE and P_grpE core promoters (−44 to −7). The UP element can often compensate for defects in the core promoter (Miroslavova and Busby, 2006). Both bioinformatic analysis (Nonaka et al., 2006) and experimental studies (V.A. Rhodius, unpublished) suggest that many σ³² promoters contain UP elements. We therefore repeated key experiments in a P_groE variant that contained the UP element (−66 to −7). Expression from complete P_groE was about fourfold higher than from core P_groE, validating that the groE promoter contains a functional UP element (data not shown). Importantly, addition of the UP element partially suppressed single-base changes at every position in the tetra-C motif of the −10 region of the promoter (Fig. 8A). Moreover, expression from double-base changed −16,−15 AA promoter was increased to 20% of wt promoter. This finding led us to test whether the UP element also suppressed the deleterious effects of mutant σ³²s implicated in recognizing this motif. Indeed, the deleterious effects of both the K130A and W108A σ³² mutants were significantly suppressed (Fig. 8B). Whereas the K130A mutant had a fourfold defect when initiating transcription from the core groE promoter, it was 80% as effective as wt when initiating transcription from the complete groE promoter. Likewise, W108A was about threefold more effective in initiating transcription from the complete promoter than the core promoter. We also tested the two variants of the grpE promoter. We find that grpE does not have a functional UP element, as the activities of both the complete and core promoter variants were the same (data not shown). Consistent with this, the complete grpE promoter does not suppress the deleterious effects of K130A σ³² (Fig. 8C). However, addition of the functional UP element from the groE promoter to the grpE promoter did suppress K130A (Fig. 8C), demonstrating that UP suppression of defective recognition of the tetra-C motif is likely to be a general phenomenon.

Fig. 8.

Effect of A/T-rich UP element on promoter recognition.

A. Effect of UP element on mutations at the −10 region of σ³² promoter. Activity of each mutant groE promoter with/without UP element are shown as a percentage of measured β-galactosidase activity of each wt promoter with/without UP element driven by wt σ³². Grey bars, P_groE-long (with UP element); clear bars, P_groE (without UP element).

B. Effect of UP element on promoter recognition mutant σ³². The bars show the activity ratios of mutant σ³² versus wt σ³². Grey bars, P_groE-long (with UP element); clear bars, P_groE (without UP element).

C. Suppression of promoter recognition mutant σ³² at grpE promoter by UP element from groE promoter (P_{groE_UP-grpE}). Assay strains are as described in Fig. 1A; promoter sequences are shown in Fig. 1B. Promoter mutations and variants, and σ³² derivatives are indicated below the x-axis. All values are averages of at least three independent experiments; error bars indicate one standard deviation.

Discussion

Sigma factors are composed of a variable number of domains, each with a particular function in promoter recognition. Thus, Domain 4 recognizes the −35 region, Domain 3 the extended −10 region and Domain 2 the −10 region of the promoter. Most Group 3 σs possess sDomain 3, and the promoters recognized by these ss have longer than normal −10 regions, often a hallmark of a composite region composed of both the extended −10 and −10 recognition units. However, no Group 3 σ had been shown to carry out recognition of the extended −10 region. The principal accomplishment of this work was to provide strong evidence that E. coli σ³² promoters have an extended −10 motif that is recognized by sDomain 3. Additionally, we provide information concerning residues in Region 2.4 recognizing the −10 region. Our findings are represented in schematic from in Fig. 9A. A structural model of the interactions we propose, based on the T. aquaticus holoenzyme/fork-junction promoter DNA complex, is presented in Fig. 9B.

Fig. 9.

Structural model of the recognition of the −10 region of promoter by σ³².

A. Comparison of the recognition of the −10 region of promoters by σ⁷⁰ and σ³². Amino acid sequences of Regions 2.3, 2.4 and 3.0 of σ⁷⁰, σ^A and σ³² are shown. The −10 consensus promoter sequences for each σ is indicated above or below the amino acid sequences. Thick arrows indicate contacts suggested by the previous studies for σ⁷⁰ and in this study for σ³². Arrow with dotted line indicates contact to increase selectivity.

B. Shown is the structure of σ^A Regions 2.1–3.0 and the −10 element region of the fork-junction promoter DNA from the T. aquaticus holoenzyme/fork-junction promoter DNA complex (Murakami et al., 2002). The rest of the protein and DNA are not shown. The σ^A Region 2.4 is coloured green and Region 3.0 is magenta. Most of the 3.0 helix is transparent for clarity. The positions corresponding to residues of σ³² that make base-specific contacts with the promoter are shown as α-carbon spheres, and colour-coded along with the corresponding DNA base pair: W108 (W257 in T. aquaticus σ^A)/−11CG, blue; E112 (A261)/−11CG, blue; and K130 (M279)/−13CG/−14CG, purple. This figure was prepared using PyMol (http://pymol.sourceforge.net/).

Two independent lines of evidence support the proposition that the upstream edge of the tetra-C motif (positions −16C and −15C) constitutes an extended −10 motif in σ³² promoters. First, allele-specific suppression analysis identifies a single residue that mediates recognition of the −16, −15 CC position. That residue, K130, is located in sDomain 3, in a position similar to extended −10 recognition determinants identified for Group 1 and Group 2 ss (Becker and Hengge-Aronis, 2001; Sanderson et al., 2003). Briefly, when K130 is present, the consensus C is preferred at positions −16 and −15; all substitutions decrease promoter activity. When K130 is altered to A130, expression from the consensus promoter decreases, but base changes at −16C and −15C no longer significantly decrease promoter activity. This behaviour constitutes genetic evidence that K130 mediates base-specific recognition of −16,−15 CC. Lysine has two hydrogen bond donors. Therefore, K130 could interact with the major groove hydrogen bond acceptors, −16G and −15G, located in the template strand. This explains how a single amino acid could mediate dual recognition. This kind of interaction has been shown in other DNA binding proteins (Luscombe et al., 2001). Second, phylogenetic analysis supports the assignment of K130 as the recognition determinant for −16, −15 CC. As reported in the supplementary figure, K130 is universally conserved in β- and γ-proteobacteria, but is substituted by alanine, serine or gluatmine in the α- and δ-/ε-groups. For example, σ³² from the α-proteobacteria, C. crescentus and R. sphaeroides are not conserved for K130. Both promoter analysis of C. crescentus σ³² (McGrath et al., 2007) and R. sphaeroides σ³² (Green and Donohue, 2006) and, more importantly, the direct analysis of the promoter recognition properties of C. crescentus σ³² (Fig. 6B) indicate that these σ³² do not recognize the extended −10 motif. Thus, both allele-specific suppression analysis and the concordance between distribution of K130 and requirement of −16,−15 CC for promoter function support the idea that −16, −15 CC is an extended −10 motif whose recognition is mediated by sDomain 3. We note Domain 3 of σ^E and σ^H, the two B. subtilis Group 3 ss studied previously, is rather discrepant from other Group 3 σs (Gruber and Gross, 2003), providing a rationale for why these σs do not have extended −10 recognition.

Parenthetically, our results rule out a role for the RpoH box in −10 recognition. This nine-amino-acid insertion in sDomain 3 is the most conserved portion of σ³² (Nakahigashi et al., 1995). Although alanine substitutions of these residues result in transcription defects, we could identify no promoter-specific defects associated with altering the RpoH box. Previous work indicated that at least one residue in the RpoH Box is involved in interaction with RNA polymerase, and that may be the function of this motif (Joo et al., 1998).

The most downstream C in the tetra-C motif, −13C, has been shown by deHaseth to be part of the −10 element, recognized by W108 located in Region 2.4. Our data extend the observations of deHaseth in two respects. First, we provide allele-specific suppression evidence that clarifies the role of W108. When W108 is present, the consensus −13C is preferred and substitutions at that position significantly decrease promoter activity. When W108 is altered to A108, expression from the consensus promoter decreases, but base changes at −13C no longer significantly decrease promoter activity, providing genetic evidence that W108 mediates recognition. The corresponding position in s⁷⁰, W434, mediates promoter melting process as well as base recognition (Fenton et al., 2000; Tomsic et al., 2001); it is possible that W108 also has a role in both processes. Second, we suggest that E112, likely to protrude from the same surface as W108 (Fig. 9B and Kourennaia et al., 2005), has an auxiliary role in recognizing −13C. Although A112 did not significantly decrease expression from the wt groE or grpE promoters, it did suppress promoter mutations at the −13C position. This kind of interaction between a Group 3 σ and its promoter is reminiscent of that reported by Moran and colleagues for the M124 residue of B. subtilis σ_E M124A does not significantly decrease expression from wt promoters, but does significantly suppress the deleterious effects of base mutations at the 5′ end of the −10 element, leading Moran to propose interaction of M124 with bases at this position (Tatti et al., 1991; Diederich et al., 1992). Possibly, the function of M124 of s^E and of E112 of σ³² is to increase the specificity of promoter recognition. These residues may inhibit non-specific interaction of the ss with the promoter region rather than directly interacting with the base pair themselves.

We report here the first studies of the interplay between the UP element and the composite −10 tetra-C motif. The presence of the UP element diminishes but does not eliminate the effect of effects of base substitutions at positions −16, −15, −14 and −13C, suggesting that initial recognition by a can partially substitute for recognition by s. We repeated these particular assays in wt cells to make sure that compensation still occurs under physiological conditions where the levels of σ³² are very low. UP element compensation for defects in the tetra-C motif is slightly diminished in wt cells, but still occurs (data not shown). We note that most σ³² promoters retain a good consensus tetra-C motif even when they have UP elements, suggesting that incomplete compensation is deleterious to cell growth.

One of the most intriguing findings in these studies is that E. coli σ³², in common with σ³²s from other γ-and β-proteobacteria, is constructed so that high promoter activity requires a consensus extended −10 motif in addition to consensus at the −35 and −10 elements of the promoter. This stands in contrast to the paradigm derived from extensive study of σ⁷⁰, the E. coli housekeeping s, and is reminiscent of promoter requirements in Gram-positive bacteria, which utilize all three elements. In the case of σ³², we know that the alteration in promoter requirements is completely contained with the σ³² itself, as σ³² from the α-proteobacteria, C. crescentus, in combination with E. coli core RNA polymerase efficiently utilizes σ³² promoters lacking the extended −10 region. Our current studies are aimed at identifying the determinants in E. coli σ³² responsible for this phenotype. Understanding this distinction is likely to provide important insights into how alternative ss are constructed.

Experimental procedures

Strains, plasmids and growth conditions

Strains and plasmids used in this study are listed in Table S1. Escherichia coli strains were grown at 30°C or 37°C in Luria–Bertani media (Sambrook and Russell, 2001) supplemented with appropriate antibiotics such as ampicillin (100 mg ml⁻¹), chloramphenicol (30 mg ml⁻¹), kanamycin (20 mg ml⁻¹), spectinomycin (50 mg ml⁻¹) and tetracycline (10 mg ml⁻¹). 0.1% of L-(+)-arabinose was supplemented for growth of rpoH strains to induce expression of GroESL. C. crescentus strain, CB15N, was grown at 30°C in PYE medium consisting of 0.2% peptone, 0.1% yeast extract, 0.8 mM MgSO₄ and 0.5 mM CaCl₂ (Johnson and Ely, 1977).

Construction of assay strains

To construct the rpoH mutant strains CAG57099, CAG57101 and CAG57102, as a first step CAG57097 (DY330 P_BAD-groESL) was constructed by P1 transduction of P_BAD-groESL from CAG48176 (McLennan and Masters, 1998; Guisbert et al., 2004) and selecting for the downstream kanamycin marker. This marker was then replaced by cat (CAG57098). Finally, CAG57099 (DY330 P_BAD-groESL DrpoH::aadA) was constructed by replacing the rpoH open reading frame by the spectinomycin-resistant gene, aadA, using electroporated linear DNA amplified by PCR and E. coli DY330 as described previously (Yu et al., 2000). aadA was amplified from pDH50 with the following primers: forward primer, 5′-CAGTTGTTGCTACCACTGAAGCGCCAGAAGATATCGATTGAGAGGATTTGAACCGTGGAAACGGATGAAGGC-3′ and reverse primer, 5′-AAACAAAAACCCCGGACTCTCATCCAGGGTTC TCTGCTTAATAGCGGAAAAGGGCTTATTATGCACGCTTAA-3′ (underlined sequences are complementary to the upstream and downstream non-coding sequences immediately adjacent to the rpoH open reading frame). The assay strain used in this study, CAG57101, was constructed by P1 transduction of ΔrpoH::aadA from CAG57100. CAG57102 was constructed by P1 transduction of dnaK756 (Georgopoulos et al., 1979) into CAG57101. The presence of each gene disruption was confirmed by resistance to appropriate antibiotics and by PCR.

Plasmid construction

Genomic DNA from E. coli and C. crescentus were prepared by Blood and Cell Culture DNA Midi kit (Qiagen) and used for templates of PCR for cloning of wt rpoH gene and promoter fragments.

The plasmid pSAKT32 encoding wt E. coli σ³², and pSAKT32 derivatives pBK116–pBK150 encoding E. coli σ³² variants were used for σ³² expression to determine their activities in vivo. The pSAKT32 plasmid derivatives carry the rpoH gene under control of the isopropyl-β-D-thiogalactopyranoside-inducible P_lac promoter as well as the lac repressor gene under the control of a strong mutant promoter(i^q) (Wang and deHaseth, 2003). All E. coli σ³² variants were constructed using site-directed mutagenesis. To express C. crescentus σ³² in CAG57102, pBK151 was constructed by replacing E. coli rpoH gene of pSAKT32 with the C. crescentus rpoH gene.

Derivatives of pQF50K (pQF50K derivatives and pBK501–pBK549 plasmid series) carrying promoter fragments cloned in the BglII-XbaI sites upstream of lacZ were used for measuring promoter strength and activities of ss (Wang and deHaseth, 2003). Promoter mutant derivatives were constructed using PCR by amplifying the promoter regions using mutagenic primers that encompassed either the BglII or XbaI site.

Derivatives of the pET21a vector (Novagen) encoding σ³² and variants were used to overproduce σ³² protein for purification. The vectors were constructed using PCR by amplifying from the pSAKT32 series using primers to create flanking NdeI and HindIII sites and inserted as NdeI-HindIII fragments into pET21a NdeI-HindIII vector. The resultant ss contain an N-terminal His₆-tag and PreScission cleavage site (Patikoglou et al., 2007).

β-Galactosidase assay

Assay systems for in vivo transcription were constructed by transforming derivatives of pSAKT32 and pQF50K into appropriate strains sequentially using electroporation. Overnight cultures in Luria–Bertani media supplemented with 100 mg ml⁻¹ ampicillin and 30 mg ml⁻¹ kanamycin (strains CAG57101 and CAG57102 were also supplemented with 0.1% L-(+)-arabinose) at 30°C were diluted 1:100 into fresh media and grown for 2 h with aeration. Cultures were then induced with 1 mM IPTG and grown for a further 3 h before harvesting (final ODs at 600 nm was approximately 0.6–0.8). β-galactosidase assays were performed as described previously (Miller, 1972).

Overproduction and purification of σ³²

Escherichia coli BL21(DE3)/pLysS cells harbouring pET21a encoding wt σ³² or variants were used to overproduce σ³² with an N-terminal His₆-tag and PreScission cleavage site for purification. Overnight cultures were diluted 1:100 into fresh media containing 100 mg ml⁻¹ ampicillin and 30 mg ml⁻¹ chloramphenicol, and grown aerobically at 30°C. When the cultures reached OD₆₀₀ of 0.4, expression of σ³² was induced by addition of IPTG to a final concentration of 1 mM and growth continued at 30°C with aeration for 2 h. Five hundred millilitres of each culture was then harvested by centrifugation and the cell pellets resuspended in buffer A1 (20 mM HEPES, pH 8.0, 300 mM KCl, 1 mM β-mercaptoethanol, 5 mM imidazole, 0.66% isoleucine, 0.83% phenylalnine, 1 mM AEBSF). The cells were disrupted by sonication and the lysates cleared of cell debris by centrifugation at 100 000 g for 90 min. N-terminal His₆-σ³² was purified from the soluble fraction by metal affinity chromatography using Talon metal affinity resin (Clontech). The N-terminal His₆-tag was then removed using PreScission protease (GE Healthcare) during dialysis into buffer Q (10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM DTT) at 4°C for 16 h. The sample was further purified by a second, subtractive metal affinity chromatography step to remove uncleaved His₆-ss and the His₆-tag, followed by ion exchange chromatography (HiTrap Q Sepharose; GE Healthcare). As a final step, DnaK-bound σ³² was removed by gel filtration chromatography (Superdex 75; GE Healthcare) and free σ³² fractions were taken and concentrated in buffer B (20 mM Tris, pH 7.5, 200 mM KCl, 1 mM DTT, 0.01% Triton X-100, 50% glycerol).

Purity of all His₆-tag free σ³²s were > 95% judged by SDS-PAGE and native-PAGE. Protein concentration was determined by the bicinchoninic acid protein assay (Pierce).

In vitro transcription

Run-off single-round transcription was performed at 30°C. Linear DNA templates were generated by amplification of promoter region of pQF50K_groE (wt groE promoter) or pBK501 (−16,−15 AA groE promoter) plasmids with following primers: forward primer, IT-F, 5′-ACTCTCAGTACAATCTG CTCTGATG-3′ and reverse primers, IT-R-S, 5′-GCTATTA CGCCAGCTGGCGAAAGG-3′ and IT-R-L, 5′-AAAGCGCC ATTCGCCATTCAGGCTG-3′. Primers IT-F and IT-R-L were used to generate control template (420 bp: 220 nt upstream and 200 nt downstream of the transcription start site) from pQF50K_groE. Short templates (355 bp: 220 nt upstream and 135 nt downstream of the transcription start site) were amplified from pQF50K_groE and pBK501 with IT-F and IT-R-S and used for comparing transcription from wt or mutant promoters.

RNA polymerase core enzyme was purified as described in previous study (Sharp et al., 1999). Holoenzyme was reconstituted by incubation of core RNA polymerase and a threefold excess of σ³² on ice for 30 min in binding buffer (20 mM Tris pH 8.0, 100 mM KCl, 10 mM MgCl₂, 0.1 mM EDTA, 1 mM DTT, 100 mg ml⁻¹ BSA, 5% glycerol and 0.05% Tween-20). Holoenzyme was diluted in binding buffer and incubated 30°C for 5 min. After incubation of holoenzyme and DNA template at 30°C for 5 min, transcription was initiated by adding transcription mix and incubated at 30°C for 6 min. Total reaction mixture was given as a 9 ml of 50 nM holoenzyme, 5 nM template DNA (2.5 nM of long template and 2.5 nM of short template), 200 mM ATP, 200 mM GTP, 200 mM UTP, 10 mM CTP, 100 nM [α-³²P]-CTP (3000 Ci mmol⁻¹), 100 mg ml⁻¹ heparin in binding buffer. Reactions were stopped by the addition of 7 ml of gel loading solution containing 20 mM EDTA, 80% deionized formamide, 0.1% of bromophenol blue and xylene cyanol FF, and 35 nt ³²P end-labelled oligomer. After incubation at 90°C for 2 min, 9 ml of each sample was subjected to electrophoresis in a 6% acrylamide/7M urea/TBE gel. The transcripts were visualized and analysed using Molecular Dynamics Storm 560 PhosphoImager scanning system and ImageQuant 5.2 densitometry software. Sizes of transcripts from long and short templates are 200 and 135 nt respectively.