Universal functions of the σ finger in alternative σ factors during transcription initiation by bacterial RNA polymerase

ABSTRACT

The bacterial σ factor plays the central role in promoter recognition by RNA polymerase (RNAP). The primary σ factor, involved in transcription of housekeeping genes, was also shown to participate in the initiation of RNA synthesis and promoter escape by RNAP. In the open promoter complex, the σ finger formed by σ region 3.2 directly interacts with the template DNA strand upstream of the transcription start site. Here, we analysed the role of the σ finger in transcription initiation by four alternative σ factors in Escherichia coli, σ³⁸, σ³², σ²⁸ and σ²⁴. We found that deletions of the σ finger to various extent compromise the activity of RNAP holoenzymes containing alternative σ factors, especially at low NTP concentrations. All four σs are able to utilize NADH as a noncanonical priming substrate but it has only mild effects on the efficiency of transcription initiation. The mediators of the stringent response, transcription factor DksA and the alarmone ppGpp decrease RNAP activity and promoter complex stability for all four σ factors on tested promoters. For all σs except σ³⁸, deletions of the σ finger conversely increase the stability of promoter complexes and decrease their sensitivity to DksA and ppGpp. The result suggests that the σ finger plays a universal role in transcription initiation by alternative σ factors and sensitizes promoter complexes to the action of global transcription regulators DksA and ppGpp by modulating promoter complex stability.

KEYWORDS:

• RNA polymerase
• alternative σ factors
• ECF σ factors
• σ finger
• promoter complex stability
• DksA
• ppGpp

Introduction

The initiation of DNA transcription is one of the main regulatory steps in the control of gene expression. Bacterial promoters are recognized by the RNA polymerase (RNAP) holoenzyme consisting of the catalytic core enzyme and a variable σ factor (reviewed in [1,2]). This provides an elegant way for switching the RNAP specificity depending on the bound initiation factor. All known bacteria have at list one primary σ factor responsible for transcription of housekeeping genes (σ⁷⁰ in Escherichia coli, σ^A in others). Alternative σ factors control transcription of genes during adaptation to various growth conditions, stress response or specific developmental programmes [1–4].

The six alternative σ factors encoded in the genome of E. coli belong to two unrelated families, σ⁷⁰ and σ⁵⁴. The members of the σ⁷⁰ family include σ³⁸ (σ^S), σ³² (σ^H), σ²⁸ (σ^F or FliA), σ²⁴ (σ^E) and σ¹⁹ (FecI) factors. RNAP containing σ³⁸ transcribes genes involved in stress response, survival at the stationary phase, regulation of secondary metabolism and biofilm formation [5–8]. σ³² is responsible for transcription of genes during heat shock response [9–11]. σ²⁸ controls genes involved in flagella synthesis and cell mobility [12–14]. σ²⁴ and σ¹⁹ belong to the most diverse ECF (extra cytoplasmic functions) family of σ factors that regulate transcription in response to signals in the periplasm or in the cell membrane [3,15,16]. σ²⁴ is involved in the membranous/periplasmic heat shock response [17–19], while σ¹⁹, the shortest σ in E. coli, transcribes only one operon necessary for transport of iron(III) dicitrate in iron starvation conditions [20,21]. Finally, the σ⁵⁴ factor is a regulator of nitrogen metabolism and a variety of other cellular processes. It shares no sequence homology with the σ⁷⁰ family and uses a distinct mechanism of transcription initiation (reviewed in ref. [22]).

Members of the σ⁷⁰ family contain up to four conserved regions further divided into subregions [2,4,23]. Promoters recognized by σ factors of the σ⁷⁰ family usually contain two conserved motifs, the −10 and −35 elements, interacting with σ regions 2 and 4, respectively [1,2,24]. In addition, promoters of the primary σ⁷⁰ factor can contain the extended −10 motif (TG upstream of the −10 hexamer) and a specific discriminator motif downstream of the −10 element, recognized by region 3.0 (aka region 2.5) and region 1.2, respectively [23–28]. σ³⁸ recognizes promoter elements similar to promoters of σ⁷⁰ [29,30], while promoters of other σs considerably differ from σ⁷⁰ [14,31–33]. However, recent analyses of the open promoter complexes (RPo) formed by RNAP holoenzymes containing σ⁷⁰, σ³⁸, σ²⁸ and σ²⁴ factors revealed that they all have a common architecture (Fig. 1). The presence of two conserved DNA binding domains, formed by σ regions 2 and 4, and the similar organization of promoter complexes suggests a universal mechanism of promoter recognition for all alternative σs from the σ⁷⁰ family [1,2,34–38].

Figure 1. Structure of the σ finger in the principal and alternative σ factors in E. coli. (A-E) Structures of promoter complexes formed by RNAP holoenzymes containing the principal and alternative σ factors: σ⁷⁰ (A, PDB 4YLO [25]), σ⁷⁰ with DksA and ppGpp (G4P) (B, PDB:5VSW [55]), σ³⁸ (C, PDB:5IPL [34]), σ²⁸ (D, PDB 6PMI [38], σ²⁴ (E, PDB 6JBQ [35]). The σ factor is shown in ochre, the σ finger (region 3.2) is orange, deletions in the σ finger analysed in this study are brown. The template DNA strand (T) is light blue, the nontemplate DNA strand (NT) is dark blue, the RNA primer is black, the bridge helix (BH) is green, magnesium ions (Mg) bound in the active site are magenta. (F) Scheme of transcription initiation. Changes in the σ finger during RNA priming are indicated with arrows. (G) Sequences of the σ finger (shown in the orange box, corresponding to the region highlighted in orange in panels A-E) in the principal and alternative σs in E. coli. Deleted amino acid residues are shown in bold. Position of the σ finger in σ²⁴ is based on the published structure of the RNAP σ²⁴ holoenzyme [35]

The primary σ factor contains a flexible linker between regions 2 and 4, formed by region 3.2 and called the σ finger. The σ finger deeply penetrates into the active site cleft and directly interacts with the template DNA strand upstream of the active site in the open promoter complex (Fig. 1A). These contacts are important for the positioning of the template strand in the active site and for binding of initiating NTPs and priming of RNA synthesis [24,39–44]. At later steps of initiation, the σ finger clashes with the growing RNA transcript, leading to abortive initiation and transcriptional pausing (Fig. 1). Further displacement of the σ finger from the RNA exit channel facilitates disassembly of the promoter complex and dissociation of the σ factor during promoter escape [39–43,45-48].

Surprisingly, despite its significant role in transcription initiation by the primary σ factor, the sequence of the σ finger is not conserved in σ³⁸, σ³² and σ²⁸ except for the presence of a few negatively charged amino acid residues. Moreover, σ²⁴ and σ¹⁹ completely lack region 3 which is replaced by a short linker of unrelated sequence (Fig. 1G) [4]. Nevertheless, recent structural analysis demonstrated that in promoter complexes formed by σ³⁸ and σ²⁸, the σ finger occupies a similar position to σ⁷⁰ in the active site cleft (Fig. 1D,E) [34,38]. Even in the case of σ²⁴, the linker between domains 2 and 4 inserts into the active site cleft in a way similar to the holoenzymes of other σs containing region 3 (Fig. 1F) [35]. At the same time, beyond this structural homology the role of the σ finger during transcription initiation by alternative σs remains largely unknown.

Transcription factor DksA and the alarmone ppGpp are key regulators of transcription during the stringent response [49–51]. The major result of the synergetic action of DksA and ppGpp on the σ⁷⁰ RNAP holoenzyme is a reduction of ribosome synthesis, mainly due to the decreased activity of promoters of ribosomal RNA and proteins [50,52,53]. DksA binds RNAP within the secondary channel and destabilizes promoter complexes formed by the σ⁷⁰ holoenzyme (Fig. 1B) [50,53–55]. The alarmone ppGpp binds to two sites on RNAP, site 1 at the β’-ω interface outside of the DNA binding cleft and site 2 at the interface between RNAP and DksA in the secondary channel, thus potentiating the action of DksA (Fig. 1B) [49,52,55,56]. DksA and ppGpp together strongly suppress the activity of promoters that form unstable complexes with RNAP, including ribosomal promoters [49,50,53,56,57]. Previously, we demonstrated that deletions and substitutions in the σ finger in σ⁷⁰ increase the stability of promoter complexes formed by RNAP and thus make them resistant to the action of DksA and ppGpp, suggesting that the σ finger plays an important role in transcription regulation during stress response [58]. DksA and ppGpp also have major effects on the expression of genes in the regulons controlled by alternative σ factors, many of which are activated possibly due to re-distribution of RNAP from the inhibited σ⁷⁰ promoters [49,59]. At the same time, the ability of DksA and ppGpp to directly regulate transcription by RNAP holoenzymes containing alternative σ factors remains poorly investigated. Furthermore, a possible role of the σ finger in alternative σ factors in the control of promoter complex stability and its sensitivity to DksA and ppGpp is unknown.

In this study, we have investigated the effects of deletions in the σ finger in σ³⁸, σ³², σ²⁸ and σ²⁴ on transcription initiation by E. coli RNAP. We have shown that the σ finger is important for efficient transcription initiation and that it modulates promoter complex stability for all σ factors. DksA and ppGpp destabilize promoter complexes of RNAP holoenzymes containing alternative σ factors. For σ³², σ²⁸ and σ²⁴, deletions in the σ finger increase the stability of promoter complexes and make them less sensitive to the action of DksA and ppGpp. Therefore, the σ finger plays a conserved function in transcription initiation by various σ factors, by promoting transcription initiation but at the same time making promoter complexes susceptible to transcription regulation by the stringent response factors.

Results

Deletions in the σ finger compromise the transcriptional activity of RNAP holoenzymes containing alternative σ factors

For analysis of the functional role of the σ finger in transcription initiation, we obtained deletions in region 3.2 of the alternative σ factors in positions roughly corresponding to the position of the previously studied deletion Δ513-519 in σ⁷⁰ (Fig. 1G) [42,43]. For σ³², the size of the deletion was larger because of an insertion of several residues in region 3.2 in this σ in comparison with σ⁷⁰ and σ³⁸. In the case of σ²⁴, the position of the deletion was selected based on the positioning of the σ2-σ4 linker in the published promoter complex structure (Fig. 1E) [35]. For analysis of the activity of RNAPs variants containing alternative σ factors, we selected several previously characterized promoters that are recognized by corresponding RNAP holoenzymes [60]: adhEp1 [61,62], talAp2 [63] and ecnBp [64,65] for σ³⁸, dnaKp1 [66] for σ³², flgMp [13] and vesP [67] for σ²⁸, and yieEp [68] for σ²⁴ (Fig. S1).

Since mutations in the σ finger in σ⁷⁰ were previously shown to affect the binding of initiating NTPs and thus impair RNA priming [42,43], we measured RNAP activity in the presence of different NTP concentrations (‘high’ and ‘low’, 200 μM and 30 μM, respectively). It should be noted that these concentrations are lower than measured NTP pools in exponentially growing E. coli cells, but are comparable to the cellular NTP concentrations at the stationary phase of growth (e.g [69].). Most experiments were performed with the adhEp1, dnaKp1, flgMp and yieEp promoters, which were highly active with corresponding RNAP holoenzymes in vitro (Fig. 2). It was shown that deletions in the σ finger markedly decreased the efficiency of RNA synthesis (quantified in percent of full-length RNA relative to the wild-type σ) for each RNAP holoenzyme, with the exception of σ²⁸ for which the deletion had only a minor effect on the total RNAP activity (Fig. 2). At the same time, the deletion decreased the activity of the σ²⁸ Δfinger RNAP holoenzyme on another tested promoter, vesP (Fig. S2). Furthermore, for the σ³⁸ adhEp1 and σ³² dnaKp1 promoters, the defects of mutant RNAPs were much more pronounced at low NTP concentrations (Fig. 2A,B, compare lanes 1,4 with lanes 7,10). This suggested that deletions of the σ finger may impair RNA priming during transcription initiation by alternative σ factors.

Figure 2. Transcription initiation by RNAP holoenzymes containing alternative σ factors with deletions in the σ finger. (A) σ³⁸ on the adhEp1 promoter. (B) σ³⁸ on the dnaKp1 promoter. (C) σ³⁸ on the flgMp promoter. (D) σ²⁴ on the yieEp promoter. The reactions contained high (200 μM ATP, GTP, CTP and 10 μM UTP) or low (30 μM ATP, GTP, CTP and 10 μM UTP) nucleotide concentrations. Primers added to the reactions are indicated as X (dinucleotide primers corresponding to each promoter, Fig. S1) and N (NADH). For each RNAP, RNA products were separated in one or two parallel polyacrylamide gel of different densities (15 and 23%) for analysis of abortive products (AP) and full-length (RO, run-off) RNA products. Positions of full length products initiated in the absence (A) or in the presence of NADH are indicated. For most promoters, additional RNA products observed below the RO product likely correspond to RNAP stalling before the end of the linear DNA fragment (see Fig. S3). The marker (M) lanes contain abortive RNA products of known lengths synthesized during transcription by the σ⁷⁰ RNAP holoenzyme on the T7A1cons promoter [43]. RNAP activities calculated in percent of full-length RNA products relative to the wild-type RNAP are indicated below the gels (means from 2 independent measurements). The scanning profiles of full-length and short RNA products synthesized in reactions performed at high nucleotide concentrations (lanes 1,2 and 7,8 in each panel) are shown on the right in each panel. The profiles were drawn for reactions separated on the same gel and were normalized to the level of RO synthesis. The ratio of RO RNA to the sum of short RNA products (indicated in the profiles) in the same reactions is shown at the bottom

We also tested the effects of short dinucleotide primers, corresponding to the transcription start site in each promoter (Fig. S1), on transcription initiation. The primers had only minor effects on the pattern of RNA products synthesized on most promoters, except for dnaKp1, in the case of which the activity of RNAP was significantly stimulated at low NTP concentrations (Fig. 2B, compare lanes 4 and 5, 10 and 11).

The decrease in the full-length RNA synthesis caused by the σ finger deletions was accompanied by changes in the quantity and pattern of abortive RNA products synthesized during transcription. In particular, the amounts of short RNAs were significantly decreased for the σ³⁸ Δfinger mutant, as a result of overall defects in transcription initiation (Fig. 2A, lanes 7–12). At the same time, the relative amounts of short RNA products (8–12 nt) synthesized by RNAP containing the mutant σ³⁸ subunit were comparable or even higher than in the case of wild-type σ³⁸, when the profiles of synthesized RNAs were normalized to the amounts of full-length RNA. Similar effects were observed for reactions performed with and without the dinucleotide primer (Fig. 2A, compare scanning profiles for lanes 1, 2 and 7, 8). The ratio of the amounts of the full-length transcript to 8–12 nt products (RO/AP) was also somewhat higher for wild-type RNAP than for the mutant (Fig. 2A). Deletions of the σ finger also increased the efficiency of short RNA synthesis relative to full-length RNA in the case of σ³² and σ²⁴ RNAPs (Fig. 2B and Fig. 2D), as is evident from both comparison of the scanning profiles of RNA products and the RO/AP ratio (lanes 1, 2 and 7, 8). In contrast, deletion of the σ finger in σ²⁸ did not significantly affect the efficiency of promoter escape but it changed the pattern of short RNA products synthesized by σ²⁸ RNAP on both the flgM and ves promoters. In paritcular the amounts of RNAs shorter than 8–9 nts were significantly decreased for this RNAP variant (Fig. 2C and Fig. S2). Overall, these results indicated that the mutations in the σ finger impair promoter escape by the mutant variants of the RNAP holoenzyme.

Incorporation of noncanonical priming nucleotides with alternative σ factors

Bacterial RNAP containing the principal σ factor was previously shown to incorporate noncanonical priming substrates during transcription initiation, including NAD⁺/NADH, FAD, desphospho-CoA and others [70–74]. However, the ability of RNAP holoenzymes containing alternative σ factors to initiate transcription with noncanonical priming nucleotides has never been studied, except for σ³⁸ [71]. Since changes in the σ finger affect the binding of initiating nucleotides, it is also possible that deletions of the σ finger in alternative σ factors might affect utilization of noncanonical priming substrates such as NADH, one of the most common capping nucleotides. We therefore compared the activities of the wild-type and mutant RNAP variants containing alternative σ factors in the absence and in the presence of NADH.

It was found that all four alternative σs could incorporate NADH during transcription initiation, as revealed by the appearance of a lower mobility RNA transcript in comparison with control reactions (Fig. S3 and Fig. 2). When NADH was present in excess over ATP, the canonical priming substrate for all four promoters, most transcripts were initiated from NADH (Fig. S3). Notably, secondary channel factors GreB or DksA had no effect on the relative efficiency of transcription initiation from ATP or NADH under these conditions (Fig. S3).

We then compared the effects of NADH on transcription performed with wild-type or mutant σ variants at different NTP concentrations. Similarly to wild-type RNAPs, all mutant RNAPs incorporated NADH during initiation, resulting in shifts in the position of abortive and full-length RNAs in the gel (Fig. 2, compare lanes 3,6 and 9,12 in each panel). NADH had a mild stimulatory effect on the activity of mutant RNAP variants, which was more pronounced at low NTP concentrations (~1.5–2 fold increase in RNA synthesis of σ³⁸, σ³² and σ²⁴ Δfinger mutants; Fig. 2A,B,D). The effect of NADH on abortive transcription was most obvious for σ³⁸, for which the whole range of short RNA transcripts was shifted by one nucleotide, apparently as a result of incorporation of NADH instead of initiating ATP (Fig. 2A, lanes 3,6,9,12).

Sensitivity of RNAP holoenzymes containing alternative σ factors to DksA and ppGpp

DksA and ppGpp were previously shown to have a crucial role in the regulation of RNAP activity during stringent response (see Introduction). To test whether promoters analysed in this study can be directly regulated by DksA and ppGpp, we measured the activity of RNAP holoenzymes containing alternative σs in the presence of either one of these factors or both. DksA or ppGpp alone had only minor effects on the overall RNAP activity for all four σs under the conditions of our experiments (Fig. 3). However, when both factors were present together, the activity of alternative RNAP holoenzymes was decreased ~2-fold in comparison with reactions containing no factors (with the exception of σ²⁴, for which the effect was smaller than for other σs) (Fig. 3, wild-type σs).

Figure 3. Effects of DksA and ppGpp on the activity of RNAP holoenzymes containing wild-type and mutant σ factors. For each σ factor, wild-type or mutant, transcription was performed in the absence or in the presence of DksA and ppGpp and normalized to the activity in the absence of regulators (see Materials and Methods for details). Means and standard deviations from four independent measurements are shown. Significant differences in activities are shown with brackets above the plots (p-values < 0.05)

To measure the effects of DksA and ppGpp on the stability of RNAP-promoter complexes, we incubated preformed promoter complexes with the competitive inhibitor of DNA binding heparin and measured the residual activity of RNAP after increasing time intervals, by monitoring full-length RNA synthesis (Fig. 4). For promoter complexes analysed in this study, the half-life times for wild-type RNAPs varied between >10 min and <5 s (810 s for σ³⁸ adhEp1, 480 s for σ²⁴ yieEp, 78 s for σ²⁸ flgMp and <5 s for σ³² dnaKp1) (Fig. 4, Table 1).

Table 1. Promoter complex half-life times for RNAP holoenzymes containing wild-type and mutant σ factors, in the absence and in the presence of DksA and ppGpp

Promoter	Factor	t1/2, s
		WT	∆3.2
adhEp1 σ^38	–	803 ± 83 1	954 ± 254 1
	DksA	260 ± 26 0.32	308 ± 71 0.32
	DksA+ppGpp	19 ± 2.3 0.02	34 ± 7.1 0.04
dnaKp1 σ^32	–	<5	170 ± 20 1
	DksA	<5	187 ± 22 1.1
	DksA+ppGpp	<5	42 ± 7.3 0.25
flgMp σ^28	–	67 ± 8.1 1	745 ± 66 1
	DksA	40 ± 4.5 0.59	850 ± 80 1.1
	DksA+ppGpp	7.8 ± 1.6 0.12	285 ± 25 0.38
yieEp σ^24	–	428 ± 95 1	> 3600
	DksA	230 ± 57 0.54	> 3600
	DksA+ppGpp	30 ± 11 0.07	3570 ± 740

Means and standard deviations from 3–4 independent measurements are shown. The numbers in bold indicate fold-changes in the promoter complex stabilities relative to the reactions in the absence of DksA and ppGpp.

Figure 4. Stabilities of promoter complexes formed by wild-type and mutant σ factors in the absence and in the presence of DksA/ppGpp. Preformed promoter complexes were incubated for increasing time intervals with heparin, followed by the addition of NTPs (in the control samples heparin was added together with NTPs). Positions of the run-off RNA transcripts are indicated. The plots show the kinetics of the decay of RNAP activity (averages from 3–4 independent experiments). The data were fitted to a single-exponential equation. The half-life times for each promoter complex are shown in Table 1

DksA and ppGpp decreased promoter complex stabilities for all four σ factors (the effect for σ³² was barely detectable because the dnaKp1 promoter complex was already unstable even in the absence of these factors) (Fig. 4). In all cases, DksA by itself moderately decreased the promoter complex half-life, from 1.5 to 2.8 fold for various σ factors (Table 1). However, the effects were much larger when both DksA and ppGpp were present in the reaction (from 4 to 25 fold) (Fig. 4, Table 1), indicating that ppGpp potentiates the action of DksA for various alternative σ factors. Together, these results indicate that DksA and ppGpp can decrease promoter complex stability and suppress the activity of RNAP holoenzymes containing alternative σ factors.

Deletions in the σ finger stabilize promoter complexes and decrease their sensitivity to DksA and ppGpp

In the last set of experiments, we analysed the effects of deletions in the σ finger on the stability of promoter complexes formed by RNAP holoenzymes containing alternative σs and on their sensitivity to DksA and ppGpp. Deletions of the σ finger significantly increased promoter complex stabilities for all σ factors except σ³⁸ (10-fold for σ²⁸ and σ²⁴, ~40-fold for σ³²) (Fig. 4, Table 1). Since no effect was observed for σ³⁸ on the adhEp1 promoter, we analysed two additional promoters for this σ, talAp2 and ecnBp (Fig. S1). The stabilities of complexes formed with the Δfinger variant of σ³⁸ on these promoters were also similar to the wild-type σ factor (Fig. S4).

We then tested the effects of DksA and ppGpp on the stability of promoter complexes formed by mutant RNAPs. RNAP holoenzyme containing the Δfinger variant of σ³⁸ was similarly sensitive to DksA and ppGpp to wild-type RNAP (t_1/2 in the presence of both factors was decreased ~25-fold for both WT and Δfinger RNAPs, Table 1). At the same time, the other three σ factors revealed markedly decreased sensitivity to the regulatory factors. DksA by itself had essentially no effect on t_1/2 for promoter complexes formed by the σ³², σ²⁸ and σ²⁴ Δfinger RNAPs (Fig. 4, Table 1). DksA together with ppGpp decreased half-lives of promoter complexes formed by the mutant RNAPs, but the extent of this effect was smaller than in the case of corresponding wild-type RNAP variants, and the resulting t_1/2 values measured for the mutants were much larger than for wild-type RNAPs (Table 1). For example, for σ²⁸ the stability of promoter complexes was increased as much as 36-fold in comparison with wild-type RNAP (t_1/2 of 7.8 and 285 s for wild-type and Δfinger RNAP variants). Similarly, for σ³² t_1/2 was increased from <5 to 42 s, and for σ²⁴ from 30 to 3600 s (Table 1).

In agreement with the observed effects of the σ finger deletions on promoter complex stabilities, DksA and ppGpp did not inhibit the activity of RNAPs containing the mutant variants of σ³², σ²⁸ and σ²⁴ during full-length RNA synthesis (Fig. 3B-D). Moreover, the activity of the σ²⁴ Δfinger RNAP was even stimulated in the presence of these factors, indicating that they may facilitate a rate-limiting step in promoter complex formation by the mutant RNAP holoenzyme (Fig. 3D). At the same time, the activity of the σ³⁸ Δfinger RNAP was still inhibited by DksA and ppGpp, consistent with the absence of the effects of this deletion on the promoter complex stability (Fig. 3A). Overall, we conclude that deletions of the σ finger in all σs except σ³⁸ can greatly stabilize promoter complexes formed by RNAP holoenzymes containing alternative σ factors from various groups and make them resistant to the action of DksA and ppGpp.

Discussion

Recent structural analysis of promoter complexes formed by RNAP holoenzymes containing alternative σ factors of the σ⁷⁰ family showed that they have a common general structure, with the σ finger occupying similar positions inside the DNA binding cleft and contacting the template DNA strand upstream of the active site (Fig. 1) [34–38,41]. The sequences of the σ finger are relatively well conserved in primary σ factors but are very diverse in alternative σs (Fig. 1G) [4,23]. Nevertheless, we have demonstrated that changes in the σ finger have similar effects on the transcriptional properties of alternative σ factors. Deletions of the σ finger impair the activity of RNAP holoenzymes containing σ³⁸, σ³², σ²⁴, and σ²⁸ factors, at least on a subset of promoters tested in this study. Deletions of the σ finger also change the pattern of abortive RNA products synthesized during transcription initiation, suggesting that the σ finger is involved in first steps of RNA synthesis, similarly to the primary σ⁷⁰ factor [42,43]. In particular, the relative amounts of short RNAs synthesized by the mutant RNAPs were increased in comparison with full-length RNA, indicative of problems in promoter escape for these RNAP holoenzymes. In all analysed σ factors, the σ finger contains negatively charged residues (Fig. 1G) that may directly interfere with the 5ʹ-triphosphate group during RNA extension and thus promote release of abortive RNAs or displacement of σ during promoter escape by RNAP [39–43]. Deletion of the σ finger also changes the pattern of abortive RNA transcripts in the case of σ⁷⁰ [42,43], and increases the retention of σ within the transcription elongation complex for both σ⁷⁰ and σ³⁸ [43,65], indicating that the σ finger facilitates dissociation of the σ factor during promoter escape by RNAP.

The σ⁷⁰ and σ³⁸ RNAP holoenzymes were previously shown to incorporate noncanonical initiating substrates into the 5ʹ-end of RNA, including NAD⁺/NADH, 3′-desphospho-CoA, FAD, UDP-glucose and UDP-N-acetylglucosamine [70–72,74]. Here, we have demonstrated that NADH can be incorporated into the nascent RNA by all four analysed alternative σ factors. Deletions of the σ finger to various extent impair transcription initiation from NADH for three out of four alternative σs, except for σ²⁸. This contrasts the previous report showing that incorporation of noncanonical substrates by the σ⁷⁰ RNAP holoenzyme on an RNA I promoter was not strongly affected by the σ finger deletion in σ⁷⁰ [71]. Therefore, the role of the σ finger in incorporation of various initiating substrates may vary for different σ factors and for different tested promoters.

Similarly to σ⁷⁰, deletions of the σ finger in three out of four analysed alternative σs, except for σ³⁸, stabilize promoter complexes formed by the RNAP holoenzyme. This indicates that the collision of the σ finger with the template DNA strand may have a general destabilizing effect on RNAP-promoter interactions. This effect may likely be stronger for promoters that form intrinsically unstable complexes with RNAP, such as rrnB P1 for σ⁷⁰ that has a suboptimal spacer between the −10 element and the transcription start site [58,75] or dnaKp1 for σ³² (Table 1). Interestingly, the deletion of the σ finger in σ³⁸ has no significant effect on the promoter complex stability on three tested promoters. While comparison of available structures of the open promoter complexes of the σ³⁸ and σ⁷⁰ RNAP holoenzymes does not reveal major differences in the positioning of the σ finger (Fig. 1) [25,34], the result suggests that it does not destabilize promoter complexes during transcription initiation by the σ³⁸ holoenzyme, in contrast to σ⁷⁰ and other alternative σ factors.

Importantly, we have demonstrated that the σ finger in the ECF σ²⁴ factor likely plays the same functions in transcription initiation as in the case of other σ factors, consistent with its similar positioning in the open promoter complex [35,37,41]. Previously, deletion or replacement of the whole σ finger in E. coli σ²⁴ with an unrelated sequence was shown to dramatically impair promoter complex formation and full-length RNA synthesis by RNAP [35]. In our study, deletion of the central part of the σ finger in σ²⁴ has a more specific effect on transcription initiation, by decreasing the overall RNAP activity, increasing promoter complex stability and changing RNAP sensitivity to the stringent response factors.

We have shown that DksA and ppGpp strongly destabilize promoter complexes formed by all four alternative σs. They also decrease the efficiency of transcription initiation from the analysed promoters, although the effects are smaller than in the case of highly regulated σ⁷⁰ promoters. The concentration of DksA is relatively constant during cell growth, and its functional activity is dependent on the varying concentration of ppGpp [50]. Consistently, we have observed that the effects of DksA on the promoter complex stability and activity of RNAP holoenzymes containing alternative σ factors are strongly potentiated by ppGpp. Deletions in the σ finger oppose the effects of DksA and ppGpp, likely by increasing the stability of promoter complexes. Similarly, mutations in the σ finger in σ⁷⁰ were shown to change the regulation of the rrnB P1 promoter by the stringent response factors, by increasing the stability of promoter complexes, decreasing their apparent affinity to DksA, and relieving the inhibitory effects of DksA in vitro and in vivo [58]. From the structural point of view, stabilization of promoter complexes by the σ finger deletion may counteract destabilization of downstream RNAP-DNA contacts caused by DksA/ppGpp [55,57,58].

During stringent response, DksA and ppGpp have complex effects on the expression of regulons controlled by alternative σ factors, first of all, σ³⁸. In most cases, they were proposed to control transcription indirectly, through re-distribution of RNAP freed from the σ⁷⁰ promoters [51,53,76]. DksA and ppGpp were also shown to activate the σ³⁸ regulon by direct stimulation of σ⁷⁰ promoters responsible for transcription of key regulators of translation and proteolysis of σ³⁸ [77]. Such direct activation of σ⁷⁰ transcription by DksA and ppGpp may result from their effects on isomerization of intermediate complexes formed during promoter recognition [56,78]. These effects likely overwhelm the possible inhibitory action of DksA and ppGpp on σ³⁸ promoters that we have observed in vitro.

In contrast, DksA and ppGpp were shown to inhibit flagellar transcription, by directly inhibiting σ⁷⁰-dependent expression of σ²⁸ itself and of its master regulator FlhDC [79]. Consistently, deletion of the dksA gene results in activation of the σ²⁸ regulon [76]. Previous studies did not observe direct effects of DksA on transcription by the σ²⁸ RNAP holoenzyme on the fliA and fliC promoters [79]. We have demonstrated that DksA and ppGpp can inhibit the activity of the σ²⁸ RNAP holoenzyme and decrease the stability of promoter complexes on the flgM promoter suggesting that DksA/ppGpp may also directly regulate the activity of σ²⁸ promoters in vivo.

Finally, the activity of the σ²⁴ regulon was shown to be increased by ppGpp and DksA upon entry to the stationary phase [80]. Some promoters recognized by the σ²⁴ RNAP holoenzyme (rpoHp3 and rybBp) could be activated in vitro by the combined action of DksA and ppGpp, despite their negative effect on the promoter complex stability [81]. Similarly, the DksA homolog TraR encoded in the F factor was shown to activate σ²⁴ promoters independently of ppGpp [82]. In contrast, we show that DksA/ppGpp do not activate the yieEp promoter recognized by the σ²⁴ holoenzyme and strongly decrease the stability of its complexes with RNAP. A possible role for the observed inhibition of the σ²⁴ transcription by DksA/ppGpp for gene regulation in vivo remains to be investigated.

It was previously proposed that DksA/ppGpp and TraR may target particular kinetic intermediates during promoter complex formation by the σ⁷⁰ holoenzyme resulting in different outcomes depending on the promoter sequence [56,78,83]. Furthermore, we demonstrated that mutations in the σ finger in σ⁷⁰ have different effects on the RNAP activity and its sensitivity to DksA/ppGpp depending on the stability of promoter complexes [58]. Thus, the reported differences in the effects of DksA/ppGpp on various promoters of σ²⁴ and other σ factors may likely result from different kinetic pathways of open complex formation on these promoters.

In conclusion, our data suggest that the σ finger play important functions during transcription initiation by bacterial RNAP holoenzymes containing various alternative σ factors. It likely participates in RNA priming and promoter escape but at the same time destabilizes promoter complexes thus making them amenable for transcription regulation by the stringent response factors. These functions of the σ finger are apparently conserved even in the most diverse ECF σ factors of the σ⁷⁰ family that completely lack conserved region 3 [4,84]. Moreover, an unrelated region RII.3 in σ⁵⁴ was proposed to play a similar role in RNA priming during transcription initiation [22,85]. Notably, the B-reader motif in the general transcription initiation factor TFIIB of eukaryotic RNAP II [86] and its counterparts in RNAP I [87] and RNAP III [88,89] are also positioned in the open promoter complex in a similar fashion to the σ finger in bacterial RNAP, and may play similar functions in transcription initiation. The requirement for the contacts of an initiation factor with the template DNA strand and growing RNA in the promoter complex may therefore be a fundamental feature of transcription initiation in various systems, likely connected to the ability of RNAPs to initiate RNA synthesis de novo.

The involvement of the σ finger in several steps of transcription initiation opens additional opportunities for the regulation of gene expression. In various σ factors, the σ finger may likely modulate the action of other transcription regulators in addition to DksA and ppGpp. Indeed, recent structural analysis of promoter complexes of M. tuberculosis RNAP demonstrated that actinobacteria-specific factor RbpA directly contacts the σ finger and may rely on these contacts to affect initial steps of RNA synthesis [90]. The contacts of rifamycin antibiotics with the σ finger observed in the E. coli σ⁷⁰ RNAP holoenzyme [91,92] may explain the effects of rifampicin on the binding of initiating NTPs, prior to its steric occlusion of the RNA exit [43,73,91]. Antibiotic fidaxomycin binds RNAP near the σ finger and these contacts possibly contribute to its effects on transcription [90,93,94]. The σ finger may therefore serve as a target for novel antibacterial compounds and antibiotics.

Materials and methods

Proteins

The E. coli genes encoding the σ³⁸, σ³², σ²⁸ and σ²⁴ factors, respectively, were amplified from genomic by PCR and cloned into the pET29 (rpoS, rpoH, rpoF) or pLATE52 (rpoE) vectors as described previously [65,95]. Mutant genes with deletions of the σ finger (Δ228-234, Δ188-197, Δ149-156, Δ101-111 for σ³⁸, σ³², σ²⁸ and σ²⁸, respectively) were obtained by PCR and cloned in the same way. The wild-type and mutant σ³⁸, σ³², σ²⁸ variants were expressed and purified from E. coli strain C3013 (New England Biolabs) by MonoQ anion exchange chromatography as described previously [65,95]. σ²⁴ containing an N-terminal His₆ tag was purified using Co²⁺-affinity chromatography as described [95]. The DksA protein was purified as in [45]. Wild-type E. coli core RNAP was purified from E. coli BL21(DE3) (Novagen) containing the expression plasmid pVS10 as described previously [96], including polyethylenimine precipitation, heparin affinity chromatography, Ni-affinity chromatography and MonoQ anion exchange chromatography steps.

In vitro transcription

Wild-type promoter fragments were obtained by PCR from genomic DNA (see Supplementary Fig. S1 and Table S1 for promoter and primer sequences). Promoter fragments with substitutions around the transcription start site were obtained by PCR from oligonucleotide templates, cloned into the pJET1.2 vector using CloneJET PCR Cloning Kit, and amplified by PCR using the same primers (Table S1).

Promoter complexes were assembled by mixing 200 nM core enzyme, 500 nM σ factors and 30 nM DNA fragments in 10 μl of transcription buffer containing 40 mM Tris-HCl pH 7.9, 40 mM KCl, 10 mM MgCl₂, and incubating the mix for 10 min at 37°C. The reaction was started by adding NTP substrates at high (200 μM ATP, GTP, CTP, 10 μM UTP, 2.5 μCi of α-[³²P]-UTP; most experiments) or low (30 μM ATP, GTP, CTP, 10 μM UTP, 2.5 μCi of α-[³²P]-UTP; experiments in Fig. 2 and Fig. S2) concentrations, together with 10 μg/ml heparin. To detect the ability of RNAP holoenzymes with alternative σs to initiate transcription starting from non-canonical substrates, dinucleotide primers corresponding to positions −1/+1 (20 μM) or NADH (2 mM) were added together with the NTP mix. The primers used for each promoter fragment are shown in Fig. S1. When indicated, DksA (2 μM) and/or ppGpp (100 μM, TriLink BioTechnologies, Inc.) were added and the samples were incubated for additional 5 min. To measure half-life times of promoter complexes, heparin (10 μg/ml) was added for different time intervals prior to NTPs. The reactions were stopped after 3 min (for half-life time measurements) or after 7 min (for all other experiments) by adding 10 μl of stop-buffer containing 8 M urea, 20 mM EDTA, 2^×TBE. RNA products were analysed by denaturing 15–23% PAGE and quantified by phosphorimaging using a Typhoon 9500 scanner (GE Healthcare). The activity of RNAP in the absence of any factors was set as 100%, all other measurements were normalized to this value. The difference across experimental conditions was evaluated with the ANOVA test followed by the post-hoc Tukey test for multiple comparison of means with the standard significance level of 0.05. The half-life times (t_1/2) of promoter complexes were calculated by nonlinear least-squares estimation of the parameters of a single-exponential decay model, y = 100×exp(-k_obs×t), where t_1/2 equals ln2/k_obs.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by the Russian Science Foundation grant [19-14-00359] to DE (role of the σ finger in transcription initiation) and [17-14-01393] to AK (analysis of DksA functions).