Mechanism of Bacterial Transcription Initiation: RNA Polymerase - Promoter Binding, Isomerization to Initiation-Competent Open Complexes, and Initiation of RNA Synthesis

Abstract

Initiation of RNA synthesis from DNA templates by RNA polymerase (RNAP) is a multi-step process, in which initial recognition of promoter DNA by RNAP triggers a series of conformational changes in both RNAP and promoter DNA. The bacterial RNAP functions as a molecular isomerization machine, using binding free energy to remodel the initial recognition complex, placing downstream duplex DNA in the active site cleft and then separating the nontemplate and template strands in the region surrounding the start site of RNA synthesis. In this initial unstable “open” complex the template strand appears correctly positioned in the active site. Subsequently, the nontemplate strand is repositioned and a clamp is assembled on duplex DNA downstream of the open region to form the highly stable open complex, RP_o. The transcription initiation factor, σ⁷⁰, plays critical roles in promoter recognition and RP_o formation as well as in early steps of RNA synthesis.

Introduction

We begin this perspective with a brief overview of transcription initiation by bacterial RNA polymerase (RNAP), summarizing the players and the major steps in the process. Excellent review articles provide a more detailed coverage of many aspects of transcription initiation.^1–9 Here we focus on current advances in understanding the process of isomerization of the initial closed complex to form the stable open complex RP_o and the many crucial roles of the specificity subunit σ⁷⁰ in all steps of initiation.

Initiation of RNA Synthesis in Bacteria

The essential players

The bacterial RNAP “core enzyme” (E) consists of five subunits, ββ′α₂ω (see Fig. 1). The core enzyme is capable of nonspecific DNA binding and initiation of RNA synthesis from DNA ends or nicks, but requires a sigma factor to initiate specific transcription from promoter DNA. Sigma assembles with core to form the “holoenzyme” (or Eσ).^18,19 Sigma factors recognize specific promoter DNA sequences, interact with transcription activators, participate in promoter DNA opening, and influence the early phases of transcription (e.g., Gruber and Gross;⁵ the latter two roles of sigma are further discussed in this review). The vast majority of studies of bacterial initiation have been carried out using Escherichia coli as model system. A model of the structure of the open complex formed by E. coli Eσ⁷⁰ RNAP (shown in Fig. 1) highlights: (i) the positioning of σ⁷⁰ on the core enzyme (Fig. 1a); (ii) the deep, wide cleft formed by β and β′ that binds the transcription bubble (Fig. 1b); and (iii) the flexible domains of β and β′ at the downstream end of the cleft proposed to assemble on the downstream duplex DNA to stabilize the open complex(es) (Fig. 1b).

Fig. 1.

Model of the E. coli RNAP (σ⁷⁰ α₂ββ′ω) open complex RP_o based on Protein Data Bank IDs 3IYD¹⁰ and 3LU0.¹¹ (a) View of RP_o illustrating the interactions between promoter DNA [nontemplate strand (NT), black; template (T), dark green] and σ₂, σ₃, and σ₄ (wheat). Linker σ_3.2 is buried in the RNA exit channel. The N-terminal domains of α (bright green, yellow) form a hinge at the bottom of the cleft. σ_NCD is a folded nonconserved domain connecting σ_1.2 and σ₂. ω is shown in light gray. Missing from the figure are σ_1.1 and the flexibly tethered αCTD (not resolved in any holoenzyme structure to date). (b) View down into the active-site channel highlighting mobile regions on the periphery of the cleft and in the cleft. At the upstream entrance to the cleft, β′ clamp helices (black) tightly interact with σ_1.2 and σ₂. The open transcription bubble (–11 to +3 in this model) binds in the cleft, with the template strand start site (+1) next to the active site Mg²⁺ (red sphere) at the bottom. βSI1 (magenta) and β′SI3 (blue) are species-specific sequence insertions (SIs) present in E. coli. The remaining colored regions are highly conserved in bacteria.¹² Along with βSI and β′SI3, β′ jaw (yellow) and β′ clamp (red) appear positioned to clamp on the downstream duplex DNA after the bubble has opened.^13–15 Flexible elements in the cleft that likely bind and stabilize the DNA single strands in RP_o include the bridge helix (visible under the double-stranded–single-stranded boundary of the downstream DNA; pink), rudder (green), fork loop 2 (teal), and switch 2 (light blue). Other mobile elements shown are the β′ upclamp (hot pink; see Supplementary Fig. 1), which is proposed to interact with upstream DNA in forming I₁ (the first kinetically-significant intermediate at the λP_R promoter),¹⁶ and the trigger loop (orange), which is known to be critically involved in the RNA synthesis steps.¹⁷

All bacteria have a primary sigma factor that suffices for growth under nutrient-rich conditions. In E. coli, the primary sigma factor is σ⁷⁰ (also called σ^D), reflecting its molecular mass of approximately 70,000 Da. In many other bacteria, the analogous primary sigma factor is designated σ^A. Most bacteria also have a complement of “alternative” or “minor” sigma factors (six in E. coli). Holoenzymes containing minor sigma factors recognize promoters of genes that can mitigate the effects of various adverse conditions.^5,6 Most bacterial sigma factors exhibit significant homology to E. coli σ⁷⁰⁶ and, as such, belong the σ⁷⁰ class. Others belong to the σ⁵⁴ class due to their similarity to E. coli σ⁵⁴ (also called σ^N, which is responsible for the expression of genes involved in nitrogen utilization), which has little sequence similarity with σ⁷⁰.³ The evolution of these two distinct lineages of sigma factors is not understood.

The structure of σ⁷⁰ is shown in Fig. 1a. The four regions of sequence conservation common to the σ⁷⁰ class sigma factors²⁰ and the architecture of promoter DNA sequences that they recognize are shown in Fig. 2. (Regions of σ⁷⁰ are designated in this review as subscripts; i.e., σ₂ refers to region 2 of σ⁷⁰.) In addition to the –10 and –35 hexameric recognition sequences²² (Fig. 2), σ⁷⁰ factors also recognize a TG sequence upstream of –10 (together called the extended –10)^23–26 and guanines in the discriminator region (see the text below) at –6 and –5.^27,28 The spacer length (i.e., the number of base pairs separating the – 10 and – 35 elements, optimally 17 bp) and the number of base pairs separating the –10 element from the transcription start site (optimally 7 bp)^29–31 both modulate the interactions of Eσ⁷⁰ with the promoter. Some promoters also include ~20 bp of A/T-rich sequence upstream of the –35 element, referred to as an “UP element” (see Fig. 2). The UP element is recognized by the flexibly tethered α-subunit C-terminal domain (αCTD).²¹ The αCTD also can bind nonspecifically to upstream DNA,^32–35 making contacts up to ~–90.

Fig. 2.

Promoter recognition by amino acids of the α subunit and σ⁷⁰. Orange and blue arrows indicate recognition of promoter regions as double-stranded DNA elements by the α and σ⁷⁰ subunits, respectively. The two red arrows delineate a region of the nontemplate strand DNA recognized by σ⁷⁰ subsequent to strand separation. In the linear representations (not drawn to scale) of both σ⁷⁰ and α, the N-termini are on the right. Only the sequence of the nontemplate strand is shown (5′ end on the left). A typical E. coli promoter does not have all elements shown and exhibits deviations from the consensus sequences shown here for the –10, –35, and UP²¹ elements, as well as the consensus spacer length (17 bp).

Steps of transcription initiation

Specific binding of Eσ⁷⁰ RNAP to promoter DNA, forming an initial closed complex RP_c, triggers a series of conformational changes in both biomolecules. This series of events, often collectively called “isomerization,” opens ~13 bp from the –10 element to just beyond the transcription start site, creating the initiation “bubble” and an unstable open complex.³⁶ In this step or in subsequent steps of forming the final stable open complex (RP_o), the +1 template (T) strand base is placed in the active site of the polymerase, and the nontemplate (NT) strand is placed in its binding track. RP_o is stabilized by the assembly and DNA binding of a downstream jaw/clamp,^13–15,36 which presumably is important for processive transcription (see Fig. 1b and the text below).

During isomerization, contacts between σ₂ and the duplex form of the –10 region in the closed complex are replaced by interactions between conserved aromatic residues in σ₂ and NT strand bases from –11 to –7 during or after DNA opening (see the text below). Work to date indicates that the –10 element (with the exception of –12, which remains base-paired) is primarily recognized as single-stranded DNA.³⁷ Open bases at positions –6 and –5 on the NT strand (the discriminator region; see Fig. 2) interact with σ_1.2,^27,28 as judged by cross-linking experiments.³⁸ Base identity at these positions has very large effects on the rate of dissociation of the open complex at the ribosomal rrnB P1 promoter, but only small effects on the binding and isomerization steps that determine the association kinetics.²⁸ Thus, bases on the NT strand from –11 to –5 appear to be largely recognized in the single-stranded state²⁸ after the opening of the initiation bubble. Importantly, no such recognition occurs on the T strand. The difference in interactions with the T strand versus the NT strand has consequences for the later steps of NTP addition and promoter escape. σ₃ (also called σ_2.5) interacts with the extended –10 TG sequences;²³ it appears to also play a role in the steps after promoter binding (Fig. 2). Although the extended –10 element remains duplex throughout initiation, changes in this sequence primarily affect the rate of isomerization and not closed complex formation.^39,40

Templated RNA synthesis (transcription) involves covalent bond formation between the 3′ OH end of the nascent RNA and the α phosphate of the incoming NTP (nucleoside triphosphate). Phosphodiester bond synthesis results in the extension of the chain by one residue and in the release of pyrophosphate. At most Eσ⁷⁰ promoters, transcription is initiated with ATP or GTP, but promoters at which initiation occurs with CTP or UTP have also been characterized. The initial transcribing complex may go down either productive or abortive paths. As a short RNA chain is synthesized, contacts between the RNAP and the promoter DNA upstream of –11 remain intact, while the promoter DNA is progressively “scrunched”^41,42 and the transcription bubble is expanded⁴³ as additional DNA is pulled in and copied into RNA. This process builds up stress and sets up a competition between extending the RNA chain and increasing the size of the DNA bubble, or releasing both the small product and the stress in the scrunched DNA to revert to RP_o.^41,42 When the nascent RNA reaches a critical length of about 11 nucleotides, the stress is instead relieved by disruption of the contacts between the RNAP and the promoter DNA.

The number and the length of abortive products produced prior to productive initiation are a function of promoter sequence and conditions,^44,45 but the precise “rules” governing this behavior remain unknown. For the single subunit phage T7 RNAP, single-molecule and fluorescence studies demonstrate that the probability of transition from an initiating complex to an elongation complex strongly depends on RNA length.⁴⁶ Abortive initiation was once thought to be an in vitro artifact or an inconsequential aspect of promoter escape. However, abortive RNAs have now been detected in vivo.⁴⁷ This discovery raises the intriguing possibility that the small products (e.g., 2–4 mers) may rebind the open complex and thus prime initiation in vivo and alter gene expression in a concentration-dependent fashion.⁴⁷

σ⁷⁰ is not required for elongation and is typically released from the transcription complex when the RNA reaches a length of 12–15 nt.⁴⁸ Release of σ⁷⁰ is likely triggered by events set in motion when the nascent RNA–DNA hybrid reaches 8–9 bp. Further extension requires displacing the sigma linker connecting σ₂ and σ₄ that lies in the RNA exit channel.^49,50 While the competition with the growing RNA chain is thought to release σ_3.2 and σ₄, it is unclear how the remaining interaction between σ₂ and β′ clamp helices⁵¹ is disrupted. Indeed, when σ⁷⁰ is retained, σ₂ induces promoter-proximal pausing at promoters with a –10-like sequence in the NT strand downstream of the start site.^52–55 Under some conditions, σ⁷⁰ release is delayed beyond the transition from initiation to productive elongation. The events governing σ⁷⁰ release versus retention remain to be defined. They are likely regulated and thus motivate ongoing investigations.

X-ray and electron microscopy structures of Eσ RNAP: Implications for transcription initiation

X-ray structural data for core,⁵⁶ Eσ^A,^50,57 and several nucleic acid–thermophilic RNAP complexes^17,49 have had considerable impact and influence on the understanding of bacterial transcription initiation. In common with other nucleic acid polymerases (but on a larger scale), the active site lies at the bottom of a deep cleft (~70 Å deep and >100 Å long; see Fig. 1b). In addition, the Eσ^A structures detail the extensive interface formed between the highly conserved regions of sigma and core, as first deduced by biochemical and genetic studies.⁵⁸

The highly conserved multisubunit RNAP architecture¹² itself appears to play a key role in discriminating promoter DNA from nonpromoter DNA during initiation. First, the arrangement of sigma on core and, in particular, the positioning of the promoter-recognition regions of sigma relative to the active-site cleft create a series of obstacles for the DNA to overcome to form an open complex. Interactions of σ₂ and σ₄ with the –10 and –35 elements (Fig. 2) define a promoter DNA trajectory in the initial “short footprint” closed complex RP_c (see the text below), which is at 90° with respect to the cleft. Consequently, a sharp bend at –11⁵⁹ and/or DNA opening outside the cleft^50,60,61 must be introduced for DNA to enter the cleft.

Does promoter DNA containing the start site of transcription (+1) enter the cleft as separated single strands or as a double helix? The ongoing debate regarding this question is driven, in part, by the narrow width (>25 Å) of the cleft seen in the Eσ^A crystal structures. This observation motivates the hypothesis that the cleft “screens” the state of the DNA by only allowing single-stranded, but not duplex, DNA entry.^50,60 However, this narrow width appears to be a snapshot of just one conformational state. In general, structures of the bacterial RNAP and eukaryotic RNAP II in various states of ligation and/or crystal forms exhibit a range of cleft widths, with distances (between the β′ clamp and the β pincer lobes) varying from <25 Å (e.g., “open”) to >15 Å (“closed”) (Mukhopadhyay et al.⁶² and references therein). Recent single-molecule fluorescence resonance energy transfer (FRET) experiments confirm that clamp opening and closing occur, demonstrating that the “hinges” in β and β′ at the base of the cleft are flexible in solution (A. Chakraborty and R. Ebright, personal communication).

Comparison of RNAP structures indicates that cleft width is controlled, in part, by the conformation of “switch 2” at the base of β′ (see Fig. 1b) (Mukhopadhyay et al.⁶² and references therein). Recent work has demonstrated that the bacterial transcription initiation inhibitors myxopyronin (myxo), corallopyronin, ripostatin, and lipiarmycin (lpm) target switch 2.^62–65 Although RNAP–myxo crystal structures indicate that myxo binding stabilizes a “partly closed” conformation of the clamp,^62,63 footprinting data on myxo–RNAP–promoter DNA complexes suggest that myxo inhibits melting of the start site region (–2 to +2) but does not prevent the entry of duplex DNA into the cleft.⁶³ Like myxo ternary complexes, lpm–RNAP–promoter DNA complexes are protected downstream to at least +15 from DNase I cleavage.⁶⁵ However, unlike myxo–RNAP– promoter DNA complexes, no permanganate-reactive thymines are detected in the presence of lpm.⁶⁵ These data, along with results summarized below, argue that cleft width does not preclude duplex DNA entry during formation of RP_o.

An additional block to forming the open complex is created by the acidic N-terminal domain of σ⁷⁰. The single-stranded nucleic acid mimic σ_1.1 binds in the cleft, blocking access to the active site.^66,67 For proper orientation of the start site base with respect to the active site Mg²⁺, σ_1.1 must be repositioned and the T strand must descend ~70 Å from its location in the closed complex (see the text below). After NTP binding and short RNA synthesis, the transition from initiation to elongation (promoter escape) requires displacing the flexible linker (σ_3.2) connecting σ₂ and σ₄ from the RNA exit channel,^50,57 and breaking the contacts of σ₂ and σ₄ with the –10 and –35 elements of promoter DNA, respectively.

Although high-resolution structures are not available for E. coli RNAP, a recent 20-Å electron microscopy structure of a ternary complex (E. coli holoenzyme-CRP-DNA)¹⁰ and a complete model of the E. coli core enzyme¹¹ reveal the locations of three large sequence insertions (SIs) in the E. coli β and β′ subunits that are absent in the thermophilic RNAP.¹² Two of these insertions, βSI1 and β′SI3, lie at the downstream end of the cleft (see Fig. 1b). β′SI3 occupies a particularly prominent position: it forms a tethered independent domain with the highly conserved β′ “jaw.” The β′ jaw/SI3 domain is highly mobile⁶⁸ and likely provides an additional steric “block” to prevent nonpromoter duplex DNA from accessing the active site.¹⁶

Steps in RP_c-to-RP_o Isomerization

Mechanistic studies

How is the start site DNA opened, placed in the active site, and stabilized? During RP_o formation, how and when are the obstacles that prevent nonpromoter DNA from accessing the cleft and being opened overcome? For several decades, kinetic mechanistic and footprinting studies have been employed to determine the sequence of conformational changes and the nature of intermediate complexes on the pathway from the initial promoter-recognition complex RP_c to RP_o. At the lac UV5 and λP_R promoters, at least two steps are required to convert the initial closed complex into the final stable open complex RP_o.^69–71 However, the “isomerization” intermediates that separate the closed complex and RP_o are relatively unstable and short lived (<1 ms to 1 s; see Fig. 3). To date, they have resisted characterization by crystallography, cross-linking, FRET, and single-molecule approaches. Their size currently precludes NMR characterization.

Fig. 3.

Summary of the proposed isomerization steps that form the initiating complex (RP_init) after recruitment of RNAP to form an initial complex at the promoter (RP_c). Formation of the closed complex RP_c triggers a series of subsequent large-scale conformational changes. The RNAP molecular machine places start-site duplex DNA in the active-site cleft in I₁, opens it to form I₂, and stabilizes the open form by assembling a clamp in I₃ and RP_o (model based on Gries et al.,³⁶ Kontur et al.,^13,15 Davis et al.,¹⁶ and Craig et al.⁷²). Once promoter DNA is open, NTPs can bind, and transcription initiates. I₂ and I₃ are open complexes; current studies are addressing whether they can bind NTPs and initiate transcription.

While methods for characterizing transient isomerization intermediates were being developed, attention was focused on initial promoter recognition (forming RP_c) and its regulation by promoter sequence and by activator or repressor proteins. Information about RP_c and other potential intermediates has been obtained by equilibrium footprinting either at low temperatures (0–15 °C) (e.g., Kovacic,⁷³ Cowing et al.,⁷⁴ and Schickor et al.⁷⁵) or with variant RNAPs⁴⁰ unable to effect promoter DNA melting. These complexes, all closed, exhibit different hydroxyl radical (·OH) or DNase footprint boundaries at different temperatures or promoters (see the text below) and have been given different names.^{59,69,70,72,76–78} Because of the challenges in characterizing kinetically significant but unstable intermediates in real time, the mechanism of forming RP_o has often been condensed into two steps: R +P→RP_c→RP_o. This mechanism collapses all of isomerization, including DNA opening and placement of the start site base of the T strand in the active site, into a single step.

In the association direction of the three-step mechanism for the lac UV5 and λP_R promoters,^70,71 the first kinetically-significant intermediate (designated I₁ at the λP_R promoter) is found to be closed¹⁶. I₁ is more “advanced” than RP_c, protecting DNA to +20. While RP_c likely forms first, it never accumulates at the λP_R promoter. I₁ isomerizes in the rate-determining step to a second intermediate (designated I₂ at the λP_R promoter and found to be open³⁶), which rapidly converts to RP_o. In the dissociation direction, the reverse direction of this same step (I₂→I₁) is rate determining. Thus, the I₁→I₂ and I₂→ I₁ steps are the bottleneck steps in each direction. In the forward direction, use of high concentrations of RNAP creates a “burst” in the population of the closed complex immediately preceding this rate-limiting isomerization step.^77–79 To create bursts of intermediates formed after I₁, RP_o is destabilized by using salts and solutes that do not destabilize these intermediates. To trap the elusive second intermediate, temperature downshifts were attempted.^70,72,80,81 However, because I₂ is an open complex, this approach also destabilized it rather than leading to its accumulation.^14,36

DNA footprinting studies of intermediate complexes

Methods for investigating late intermediates of isomerization have been developed recently.^14,36 These methods, combined with rapid-quench (<2 ms) mixing, allow one to perform “real-time” kinetic and chemical footprinting experiments on the timescale of the formation and disappearance of transient intermediates.^36,77,78 To date, all structural information about complexes known to be on-pathway intermediates in RP_o formation has come from chemical and enzymatic DNA footprinting methods.

Recent advances in describing the large-scale conformational changes that occur after recruitment of RNAP to the promoter and initial specific binding are summarized in Fig. 3. In addition to opening of the promoter DNA, strong evidence is obtained for DNA wrapping and for coupled folding and domain repositioning of RNAP. Because all of these conformational changes are driven by binding free energy, the motions in the RNAP machinery are linked to the DNA sequence and structure in the interfaces that form, as well as to solution conditions. Below we detail our current understanding of the progression of conformational changes, the structures and stabilities of the intermediates, and the controversies and questions that remain.

RP_c formation: Initial recognition of duplex promoter DNA sequences

Initial specific interactions of RNAP with promoter DNA form a closed complex in which the promoter DNA is fully duplex. Based on the low-temperature footprinting data (see the text above) and the structures for holoenzyme^50,57 and for a complex of RNAP with a “fork junction” promoter fragment,⁴⁹ a model of such an initial closed complex has been proposed. In this model, duplex promoter DNA interacts with σ₂ and σ₄, and a continuous DNA duplex extends downstream of the –10 element, projecting away from the active-site cleft and therefore is cleavable by DNase I or ·OH.^60,82

Real-time ·OH footprinting after mixing high concentrations of RNAP and T7A1 promoter DNA shows progression in the protection of the downstream boundary after mixing.^77,78 In these studies, a series of closed intermediates, which initially establish protection between –80 and –55 and then progressively extend it downstream, has been proposed. These snapshots suggest that interactions involving αCTDs and the T7A1 UP element and/or other far-upstream DNA contacts with RNAP are established first (see also Borukhov and Nudler⁸). The footprint then extends downstream as contacts between σ₂, σ₃, and σ₄ and the –10 element, and between the spacer and the –35 element of the promoter DNA, respectively, are established to form a RP_c complex with a downstream boundary of –5.^77,78

More advanced closed complexes, including the upstream-wrapped closed intermediate I₁ with downstream duplex in the active-site cleft

Downstream boundary and its implications for closed complexes

More “advanced” closed complexes with downstream boundaries extending to +15–20 have also been characterized.^{16,72,74,80,83–86} Conversion from an “RP_c”-like complex into one that protects one to two turns of the DNA helix downstream of the start site (+1) has been observed by increasing the temperature to 16–20 °C^74,80,83 or, in the case of the rrnB P1 promoter, by increasing the temperature to 37 °C in the absence of NTPs.^84,86 In all cases, chemical probes (dimethyl sulfate and permanganate) do not detect open/unstacked bases in these “advanced” complexes, and the periodic patterns of protection observed from ~–55 to –12 transition to full protection of both strands of the helix between –11 and +15.

At the λP_R promoter, both the transient, kinetically significant intermediate closed complex (I₁) that accumulates early in the time course of open complex formation¹⁶ and the low-temperature (0 °C) closed complex,⁸⁷ also known to be I₁ by extrapolation of thermodynamic data from 7 °C and higher temperatures, have an extended downstream footprint (to +20–25). Based on the holoenzyme structures, we proposed that protection of both strands from –11 to +15 results from a sharp ~90° bend at the upstream end of the –10 element that inserts downstream duplex DNA into the active-site cleft.⁵⁹ Additional protection to +20 and +25 likely arises from mobile elements at the downstream end of the cleft that block access to the DNA backbone (e.g., β′ SI3; see Fig. 1b). Conversion of RP_c to a more I₁-like complex appears to be driven by increasing temperature, by favoring the bend at –11/–12 and/or the interactions that stabilize the bend. At λP_R, I₁ is the most advanced closed complex because it opens in the next kinetic step.

In closed complexes with downstream boundaries between –5 and +15,^33,84,86 downstream DNA is presumably only partially inserted into the cleft. Is this because of the difficulty in bending the DNA or the difficulty in inserting the duplex in the cleft? Possible examples of both scenarios are available. Davis et al. found that the I₁ intermediate formed by RNAP at an upstream-truncated (UT-47) λP_R promoter exhibits a downstream boundary of +2(NT)/ +7(T).¹⁶ Since the interactions of the –10 element are presumably identical in UT-47 and full-length λP_R, the difference in the insertion of the downstream duplex in the cleft may therefore indicate an obstacle in the cleft that is removed as a result of interactions with far upstream DNA (above –47).

Upstream boundary and its implications for closed complexes

Interactions with DNA upstream of the –35 hexamer are established in the steps of promoter recognition.^16,77,78 They influence the stability of the intermediate I₁ and the rate of converting it to the next intermediate I₂. The αCTDs (see the text above) mediate upstream interactions by binding DNA either specifically^21,88 or nonspecifically^32–35 and by interacting with activator proteins.^1,89 Intriguingly, all proteins (including the αCTDs, based on DNase I enhancements) that bind upstream of the –35 hexamer and modulate transcription bend DNA [e.g., CRP (cyclic AMP receptor protein), IHF (integration host factor), and FIS (factor for inversion stimulation)]. The function of these DNA bends may be simply to provide better interactions with the activator and/or the αCTDs. However, many transcription factors bind to sites upstream of –90, presumably out of “reach” of the flexibly tethered αCTDs. In addition, the phasing of upstream binding sites for these factors (e.g., see Dethiollaz et al.⁹⁰ and Giladi et al.⁹¹) affects their action. Shifting the UP element in the upstream direction relative to the –35 hexamer of a given promoter abolishes UP element activation of transcription, regardless of phasing; lengthening the αCTD–αN-terminal domain linker does not restore full-length transcripts to their nondisplaced UP element levels.⁹² Moving the UP element upstream prevents the formation of a complementary interface between σ₄ and the adjacent (proximal) αCTD.^93,94 Likewise, shifting sites for transcription factors that naturally abut the –35 hexamer upstream destroys the interface that they form with σ₄.^95,96 In the wild-type context, formation of these complementary interfaces is a key event in transcription initiation.

The above data have engendered several hypotheses about the role of upstream interactions. One hypothesis is that protein–protein interfaces communicate “allosterically” with the active-site channel to affect steps in initiation. Alternatively, or in addition to possible allosteric effects, we have proposed that the network of interactions between σ₄ and the αCTDs (and transcription factors, when present) and DNA bends the DNA from ~–30 to –55 and thereby sets the trajectory of far upstream DNA in the early steps of RP_o formation¹⁶ (see Supplementary Fig. 1). The importance of the upstream DNA trajectory is based on several observations. First, the presence of DNA upstream of the –35 element at the lac UV5 and λP_R promoters profoundly accelerates the bottleneck isomerization step (see Fig. 3), now established as the DNA opening step at λP_R.³⁶ The isomerization rate constant k₂ (conversion of I₁ into I₂) for full-length λP_R is ~50-fold larger than that for truncated λP_R, with the upstream DNA deleted beyond –47 (UT-47), an effect as large as or larger than that exerted by activator binding. Surprisingly, deletion of this upstream DNA has little effect to no effect on the stability of I₁. ^33,34 Second, deletion of upstream DNA leads to a “less advanced” closed complex that only protects downstream DNA to +2 (T)/+7 (NT)¹⁶ relative to ~+20 observed for the full-length λP_R promoter.

How might upstream interactions facilitate the loading of downstream DNA in the cleft? In the final stable open complex at the λP_R promoter, the upstream boundary defined by DNase I or ·OH cleavage ends at ~–65. However, ·OH footprints of I₁ on full-length λP_R reveal modest protection of the DNA backbone on both strands to at least –85.¹⁶ DNase I hypersensitive sites in I₁^16,72 indicate that a bend occurs just upstream of the –35 element. Mapping the I₁ protection pattern and the inferred bend onto available X-ray structures indicates most simply that RNAP wraps DNA around the “back” of the β′ subunit (Supplementary Fig. 1). Alternatively, the upstream interactions responsible for the large effect on the kinetics of the DNA opening step discussed above and for the far upstream ·OH footprint could involve the mobile αCTDs (see the text below).

In the upstream wrapping model,¹⁶ the bend induced by σ₄ and the two αCTDs places far upstream DNA near the downstream end of the cleft (see Supplementary Fig. 1). Because the pattern of protection from –65 to –85 is not periodic (as typically observed for the αCTDs⁹⁷), we proposed that this region of DNA is directed into a surface groove formed by β′ and the N-terminal domain of the associated α subunit. If so positioned, the upstream DNA lies near a conserved mobile element in β′, termed the upstream clamp. The upstream clamp is directly connected to other dynamic elements at the downstream end of the cleft: conserved jaw, trigger loop, and β′SI3. Based on our kinetic and footprinting data,^16,33 we hypothesize that interactions between upstream DNA and upstream clamp restrain the movements of the jaw domain, trigger loop, and β′SI3. Without this constraint, it appears that these elements sterically interfere with the loading of the downstream DNA in the cleft (see Fig. 1b; Supplementary Fig. 1). Predictions of this model are currently being tested.

Alternative hypotheses for the role of upstream DNA invoke a direct role for the αCTDs in mediating the acceleration of the DNA melting step beyond bending upstream DNA. In the absence of the αCTDs, the presence of upstream DNA only increases k₂ by ~2.5-fold.³⁴ Cross-linking of RP_o indicates that the αCTDs can occupy multiple sites on the upstream DNA.³⁵ Discerning whether the role of upstream DNA is to simply provide additional nonspecific αCTD binding sites or whether the αCTDs and σ₄ together set a trajectory required for wrapping interactions between upstream DNA and other elements of RNAP in I₁ or other early complexes awaits further experiments.

I₁→I₂: DNA opening is the bottleneck step in RP_o formation at the λP_R promoter

How is DNA opened by RNAP? Two conflicting hypotheses describing this critical step exist in the field. One hypothesis, based on structural data, posits that opening is nucleated by DNA breathing above the active-site cleft, after which the T strand enters the cleft and diffuses to the active site.^50,60 Evidence in support of this proposal has been obtained from molecular dynamics simulations on modeled structures formed by the bacterial RNAP⁶¹ and from a comparison of the time evolution of downstream ·OH and MnO₄^- footprints in association experiments at the T7A1 promoter.⁷⁷ A second hypothesis proposes that the DNA duplex is first loaded in the cleft, where it is then actively opened by elements on RNAP.^15,36 This proposal is supported by extensive kinetic and footprinting (equilibrium and real time) experiments on the λP_R promoter (cf. Fig. 3), equilibrium footprinting experiments at other promoters (see the text above), and equilibrium footprinting experiments performed in the presence of antibiotics that block DNA melting (discussed above).

Closed promoter DNA–RNAP complexes that protect the DNA backbone to at least +15 demonstrate that duplex DNA can bind in the active-site cleft. However, in most cases, evidence that complexes populated at equilibrium were on-pathway kinetic intermediates was not obtained. However, extensive kinetic studies (filter binding) of RP_o formation at λP_R, combined with real-time footprinting experiments, provide strong evidence that duplex DNA (–11 to +20) occupies the active-site cleft in the final closed on-pathway intermediate I₁. Once bound in the cleft, the next step (I₁→I₂) opens DNA (–11 to +2), as detailed below.

At λP_R, DNA in I₁ is not MnO₄^- reactive and is continuously protected from ·OH and DNase I on both strands from –11 to positions +20–25.^16,72 Thymines in the subsequent kinetically significant intermediate I₂ are MnO₄^- reactive at all positions reactive in RP_o.³⁶ In addition, real-time footprinting experiments reveal that the extended downstream footprint (protected from ·OH cleavage to ~+20) of I₁ develops in 100 ms. In contrast, the MnO₄^- reactivities of thymines detected in RP_o develop much more slowly (tens of seconds; Heitkamp, Drennan, et al., in preparation). Therefore, we conclude that duplex DNA binds in the cleft in I₁, and that the entire bubble opens concertedly in the cleft at λP_R in one kinetic step.^15,16,36,59

In addition to evidence cited directly above, the following data also indicate that DNA opens in the cleft. The rate constant for I₁→ I₂ is strongly temperature dependent.⁵⁹ The corresponding activation energy (34 kcal) is consistent with the cooperative opening of at least 6–7 bp in the I₁–I₂ transition state. While salt and other solutes exhibit large effects on DNA opening in solution, they only exert small effects on the DNA opening (k₂) and closing (k_–2) steps for the λP_R promoter.^{13–15,36,59} Most simply, these data indicate that DNA opening occurs in the sequestered environment of the cleft and not outside it. Alternatively, compensating for the stabilizing and destabilizing effects of these salts and other solutes may accompany opening.

I₂→RP_o: Evidence for the assembly and DNA binding of a downstream clamp/jaw to stabilize the open complex

Evidence to date indicates that the final steps of isomerization involve the interconversion of multiple different open complexes, including on-pathway intermediates I₂ and I₃, as well as the final open complex RP_o. This striking discovery has significant implications for the regulation of transcription initiation.^14,36 Major changes in DNA and RNAP in the conversion of I₂ into RP_o were revealed for the first time by burst footprinting of the dissociation intermediate I₂ and by analysis of dissociation data as a function of both stabilizing and destabilizing solutes and salts.^13–15,36 These include downstream folding and assembly of >100 residues of mobile elements of RNAP (Figs. 1 and 3; Supplementary Fig. 1) to form a clamp/jaw on downstream DNA, as well as establishment of in-cleft interactions. Evidence for the latter includes a 2-fold increase in the MnO₄^- reactivity of thymine bases in the downstream region of the NT strand (–4, –3, and +2) in the conversion of I₂ into RP_o. Thymine bases in the upstream half of the NT strand remain protected from MnO₄^- oxidation by being bound to RNAP (σ₂) in both I₂ and RP_o. The start site thymine (+1, T strand) is equally MnO₄^- reactive in I₂ and RP_o, suggesting that it is correctly positioned in I₂.

Downstream interactions in the open complexes

Numerous lines of evidence demonstrate that RNAP undergoes a large-scale conformational change in the steps following DNA melting. The extreme effects of solutes, temperature, and salts on the steps converting I₂ into RP_o suggest that the late steps of RP_o formation create a new protein–DNA interface in a process that involves coupled folding.⁹⁸ Quantitative analysis of the effects of multiple solute probes on the dissociation rate constant K_d indicates that 75–100 residues fold in the conversion of I₂ to RP_o; the effects of salt are consistent with a burial of 10 or more DNA phosphates.¹⁵ The conformational changes that occur in these steps appear to be comparable in scale to those that occur in the conversion of the initiation complex to the elongation complex of the T7 phage RNAP (~300 amino acids refold⁹⁹).

To interpret these results, we proposed^13–15 that several large (50–70 residue) mobile regions of the β′ and β subunits fold and assemble into a jaw/ clamp superstructure that binds to duplex DNA (see Fig. 1b; Supplementary Fig. 1) after DNA opening. Given the tight binding interactions established upstream in the early steps, assembly of the downstream clamp is likely delayed to allow unimpeded rotation of the downstream DNA on its helical axis by 1.3 turns (470°).^13–15 Regions in β′ include: (i) the jaw; (ii) a highly positively charged helix hairpin helix; (iii) β′SI3; and (iv) a C-terminal region adjacent to (i). Individual deletions of (i)–(iii) all destabilize RP_o;^13,84,100 deletions/mutations in (iv) have not been studied. In addition, βSI1 also likely forms part of the clamp.¹⁴ In the recent model of RP_o¹⁰ (as well as in the complete model of an E. coli transcription elongation complex¹¹), all of the regions above are positioned near the downstream DNA, but do not necessarily directly interact with it. This may be a consequence of the low resolution of DNA in electron microscopy images and of basing the model on the transcription elongation complex, which only protects downstream DNA to ~+10–15¹⁰¹ (~1 turn of DNA shorter than RP_o). Nonetheless, these models clearly show that β′SI3, the jaw, and βSI1 are positioned to clamp downstream DNA from +10 to +20.^10,11

In both the RP_o and the transcription elongation complex models, the trigger loop, which connects to β′SI3 through flexible linkers, is unfolded. Intriguingly, using the folded form of the trigger loop¹⁷ creates steric clash, leading to the proposal that the jaw/β′SI3 domain likely toggles between two positions as the trigger loop folds and unfolds with each cycle of NTP addition.^10,11 In addition, the RP_o model shows density for βSI1 near +15–20. The volume of this density increases relative to the partially disordered state in free E. coli RNAP,⁶⁸ supporting our hypothesis that βSI1 folds on binding downstream DNA.¹³

In-cleft interactions in the open complexes

Are these large conformational changes in the RNAP downstream machinery during the conversion of I₂ to RP_o connected/correlated with smaller-scale but very significant conformational changes in the active site and surrounding regions of the cleft? Differences in the MnO₄^- reactivities of bases on the downstream portion of the NT strand in I₂ and RP_o suggest that rearrangements in the NT strand are coupled to the formation of the downstream DNA clamp in the conversion of I₂ to RP_o.^{13–15,59,102} Details of the communication between the cleft and the clamp/jaw remain to be established. This communication, if established in the isomerization steps, would likely persist in regulating the subsequent steps of the transcription cycle. Does the sequence/length of the discriminator region (–6 to +1) affect the “repositioning” of the NT strand and, thus, the stability of RP_o at different promoters? Are I₂ and I₃ functional in transcription? Do the different open complexes (one unstable and one stable) play distinct functional roles of open complex formation (e.g., Which open complex is the target of the stress sensor protein DksA?⁸⁵)? How do the contacts established in the late steps of RP_o formation affect promoter escape? Addressing these questions will likely bring new challenges and surprises, and will advance our understanding of the regulation of these late steps as a function of promoter sequence and solution conditions.

The role of σ⁷⁰ in promoter interactions, open complex formation, and early RNA synthesis

Promoter recognition

Structures of RNAP holoenzyme from Thermus aquaticus and Thermus thermophilus reveal that σ⁷⁰ consists of several independently folded domains (σ_1.2, σ₂, σ₄, and likely the N-terminal ~60 residues of σ_1.1 as well¹⁰³) connected by flexible linkers (σ_3.2 and the highly negatively charged C-terminal residues of σ_1.1). Recent evidence reveals that the structure of the free sigma factors is compact, and that σ_1.1 and σ₄ interact.¹⁰³ This interaction may lead to the observed autoinhibition of promoter DNA binding by free primary sigma factors.¹⁰⁴ Autoinhibition has not been observed for interaction with the double-stranded promoter DNA of free sigma factors lacking region 1.1 by deletion¹⁰⁵ or naturally,^106,107 or for free σ⁷⁰ interacting with the NT strand of promoter DNA.¹⁰⁸

In the holoenzyme, the interactions between sigma and the core cover an extended surface area of both proteins⁵⁸ (see Fig. 1a): various regions of σ⁷⁰, including the linkers, interact with core RNAP, thus affording the bound σ⁷⁰ considerable structural rigidity compared to the free σ⁷⁰. For example, σ_2.3/2.4 (–10 recognition) and σ_4.2 (–35 recognition) are now at a fixed orientation with respect to each other, imposing a rather strict limitation on the length of the spacer DNA separating the –10 and –35 elements (17±1 bp). Indeed, there are conditions where only σ₂ of sigma is bound to RNAP in a paused transcription complex where the RNA is 16–17 nt—long enough to have displaced σ₄ from the β flap. At this point, –10-like and –35-like elements that occur just downstream and upstream of the start site of transcription, respectively, can be jointly contacted by the now flexibly tethered σ₂ and σ₄ even if they are separated by only 1 bp.¹⁰⁹

Redundancy in σ⁷⁰ promoter elements

The –10 and –35 elements together constitute the classical prokaryotic promoter. However, in addition to the –10 and –35 elements mentioned above, there are several other regions of promoter DNA contacted by σ⁷⁰ or the α subunit (see the text above). This raises the question of whether other pairs of promoter elements can also constitute an active promoter. Promoter DNA melting initiates within the –10 element, rendering this the most important and the least dispensable of the promoter elements. Can other regions substitute for the –35 element? This has indeed been found to be the case (reviewed by Hook-Barnard and Hinton⁷). Notable among these are the extended –10 (TG+–10) and UP+–10, which has so far been only characterized as an artificial construct. Under in vitro conditions, promoter DNA strand separation has been observed with DNA containing just the –10 element.¹¹⁰ Even the combination of –35+TG has been found to be active, if provided with an A+T-rich region that has the all important –11A and –7T in the NT strand.¹¹⁰

Stringent promoter requirements for holoenzymes containing “minor” sigma factors

In addition to the primary (“housekeeping”) sigma factor, most bacteria have one or more minor sigma factors that can impart to RNAP the ability to transcribe genes whose products allow cells to deal with various types of stress. An example is the heat shock sigma factor σ^H (usually called σ³²),^111,112 which helps to mitigate the cytoplasmic consequences of transient exposure to higher-than-optimal temperatures. Promoter recognition occurs through a –35 element similar to that of σ⁷⁰ promoters and an extended –10 element of which the sequence deviates considerably from that recognized by Eσ⁷⁰.¹¹³ Importantly, while Eσ⁷⁰ is relaxed in its ability to recognize promoter DNA sequences, Eσ³² is found to be stringent in requiring promoters with consensus or near-consensus sequences. The reason for the difference is the DNA melting region (2.3) of σ⁷⁰: a broader promoter-recognition spectrum for Eσ³² is generated by replacing just two residues of region 2.3 with homologous aromatic σ⁷⁰ residues.¹¹⁴ These experiments indicate not only that σ³² is intrinsically melting deficient but also that this deficiency can be overcome by the use of consensus promoters. Indeed, similar behavior has also been observed for σ⁷⁰ mutants rendered melting deficient by substitutions in region 2.3: if provided with a very good promoter, RNAP containing a defective σ⁷⁰ could still form an open complex.⁷⁹ Observations similar to those described above for σ³² have also been made for RNAP containing σ²⁸, another minor E. coli sigma factor.¹¹⁴

Nucleation of DNA melting: Role of conserved residues in σ⁷⁰

The –11A element in the NT strand of E. coli σ⁷⁰ promoters plays an important role in the formation of an open complex, as detailed in a number of studies.^76,115–119 It is the most conserved base pair in the –10 element.¹²⁰ Substitution of the –11A element by 2-AP¹¹⁷ or loss of the base at this position^121,122 has much larger negative effects than at other positions. At the upstream end of the –10 region, the A-T base pair at –12 remains double-stranded in RP_o. A clear indication that strand separation is initiated at the –11 position is derived from studies demonstrating a correlation between the reduced stability of base pairing at –11 bp and the ability of promoter DNA to be melted by RNAP.¹¹⁶

Residues Y421, Y425, F427, T429, Y430, Y425, W433, and W434 (E. coli σ⁷⁰ numbering; see Fig. 4) are nearly invariant among 53 sigma factors analyzed¹²⁴ and are found within a short distance of each other and of the –11A base⁴⁹ (at the upstream single-stranded–double-stranded DNA boundary over σ₂ in Fig. 1a). T429, Y430, and W433 are near the double-stranded–single-stranded junction of the model DNA cocrystallized with the RNAP, consistent with their involvement in the initiation of DNA melting. Evidence for the vital roles of Y430 and W433 includes deleterious effects of substitutions on open complex formation^125–127 and their high extent of conservation. Compelling evidence has been obtained for an interaction of Y430 with –11A.⁷⁹ However, it is likely that Y430 (and W433) additionally also recognizes other bases, and that other amino acid residues recognize –11A. Support for such a network of interactions, with multiple roles for the participating groups, is derived from two sets of observations. First, the effects of various substitutions for Y430 and W433 on the ability of RNAP to form stable promoter complexes are evident even in the absence of the base at –11 of the NT strand.¹²² Second, a variant RNAP containing a multiply substituted σ⁷⁰ (alanine substitutions for F427, Y430, W433, and W434) has retained the capability for sequence recognition at –11.⁴⁰

Fig. 4.

Structure of region 2.3 of σ⁷⁰.¹²³ In the N→C direction are helix 13 (lower helix), a loop, and helix 14 (upper helix). The side chains of K414, K418, Y425, T429, Y430, W433, W434, and Q437 (Lys, green; Tyr, red; Thr, blue; Trp, purple; Gln, pink) stick out towards the viewer from approximately the same face of the protein, where they can interact with promoter DNA. Y430 has been shown to stack with –11A of the –10 region.⁷⁹ Y421 sticks out in another direction but may be able to interact with DNA. The structures of the T. aquaticus and T. thermophilus σ_2.3 are very similar.^50,57

It is envisioned that interactions with basic amino acid residues of σ₂ (including K414 and K418) anchor the promoter DNA to the surface of the RNAP in the closed complex.¹²⁸ Promoter DNA melting is likely initiated by the rotation (or “flipping”) of –11A out of the DNA helix^76,129 so that it now can stack onto Y430. Aromatic amino acid residues T429¹²² and W433^125,128 of σ_2.3 are likely closely involved in the actual process of flipping –11A out of the DNA helix.⁷⁹ From the flipped –11A, DNA strand opening would proceed in downstream direction to +2.

Melting 12–14 bp of duplex DNA at 25–37 °C in the absence of RNAP has a large enthalpic cost (~70–84 kcal), thought to arise primarily from base unstacking and not from breaking of hydrogen bonds.¹³⁰ However, the activation enthalpy for opening the transcription bubble at λP_R is approximately half as large (~34 kcal⁵⁹). Preservation of intrastrand stacking and favorable interactions between bases on the NT strand and aromatic residues (see the text above) likely reduce the enthalpic cost of opening the initiation bubble. Evidence for this hypothesis comes from the lack of permanganate reactivity of thymines at –7 and –10, indicating that these bases, once opened, either remain stacked with their neighbors or interact with residues in σ₂. In addition, NT strand bases (–4, –3, and +2) may be partially stacked in I₂, since they are only half as MnO₄^- reactive as in RP_o.³⁶ Thus, the model presented above is perhaps best described as “bind–bend/flip–melt,” followed by clamping. In this model, RNAP is an active participant in achieving DNA strand separation: both RNAP-induced DNA bending and the side chains of amino acids T429, Y430, and W433^{79,122,125,126,128} facilitate the DNA strand separation reaction. In addition, various elements in the cleft, such as the “fork loop 2” of β and multiple “tracks” of positively charged residues, appear positioned to capture and stabilize the open state via interactions with the DNA phosphate backbone.¹⁵

Sigma release versus retention

Since the discovery of sigma factor over 40 years ago,^18,19 it has been thought that an obligate late step in transcription initiation was the release of sigma factor from RNAP. However, two studies^131,132 clearly demonstrated the retention of σ⁷⁰ in transcription complexes beyond the early phases of transcription. Compelling evidence for the presence of σ⁷⁰ in elongation complexes was obtained from both FRET experiments and analysis of the proteins in the complexes. Global analysis techniques¹³³ provided support for σ⁷⁰ retention in vivo. While these studies did not demonstrate a function for the retained sigma factor, other work demonstrated that the retained sigma factor was instrumental in generating a promoter-proximal pause of transcription during the synthesis of bacteriophage (i.e., bacterial virus) λ mRNAs. Such a pause is vital for endowing the transcription complexes with the ability to read through termination signals.¹³⁴ Subsequently, similar sigma-dependent pausing was demonstrated for the transcription of various bacterial genes as well^{53–55,135,136} (see also a recent review by Artsimovitch¹³⁵).

The role of the retained σ⁷⁰ is to recognize –10-like sequences on the NT strand of the transcribed DNA.^53–55,134 The interaction of σ⁷⁰ with such regions was found to be similar to its interaction with bases of the –10 element on the NT DNA. The interaction may be further strengthened by contacts to G-C base pairs⁵² positioned similarly to the G-C base pairs at –5 and –6, which are contacted by σ_1.2 in RNAP–promoter complexes.^27,28 Interestingly, the –10-like element may not be absolutely necessary, although it greatly stabilizes the interaction of sigma with the transcription complex.⁵⁴ Recently described was an atypical example of σ⁷⁰-dependent pausing where the –10 element was lacking but pausing was shown to be dependent on the TG sequence of the extended –10 element of the actual promoter and contacts to a C-rich region at +2–6 of the NT strand.¹³⁶ The σ⁷⁰–NT DNA contacts serve to lock the elongating RNAP in position, thus impeding further movement of the elongation complex. The duration of the sigma-facilitated pause is reduced by GreA and GreB proteins in vitro, and evidence is consistent with this also being the case in vivo.^53–55,137 This behavior is indicative of backtracking of transcription complexes during the pause.

It remains to be established whether sigma retention is characteristic of most, or all, σ⁷⁰ promoters. It may be that retained σ⁷⁰ is not detected unless the NT strand has the proper –10-like sequence for σ⁷⁰-dependent pausing. Then the interaction of σ⁷⁰ with the transcription complex would be stabilized,^53,54 further delaying sigma release. Alternatively, delayed release of sigma may be a promoter-specific event for which the signals have not yet been discerned. In support of the former, experimental evidence has been obtained for a stochastic release of sigma factor from the elongating complex, which would manifest itself as a certain half-life (estimated to be on the order of 5 s) for σ⁷⁰ release from RNAP during transcription elongation.¹³⁸ A decreased stability of the RNAP– sigma complex is consistent with the suggestion that sigma's attachment to core during elongation differs from that in free holoenzyme.⁵³ Indeed, it is likely that the nascent RNA, by the time it has reached 10 nt in length, will have pried loose the contacts between σ_3.2 and the β′ subunit in the RNA channel of the core enzyme. At about 16 nucleotides, the contacts between σ₄ and the core enzyme β flap would be disrupted. Just the interaction of σ₂ and the clamp helices would then tie the sigma to the transcription complex. Interestingly, recent work indicates that release of σ₂ from the β′ clamp helices is required to load the elongation factors NusG¹³⁹ and rfaH,¹⁴⁰ which thus may play a role in the release of sigma from the elongation complex. Regardless of the nature of the contacts, sigma apparently is held in a position that allows it to scan the sequence of the NT strand.

Conclusion

Determination of high-resolution structures of free and promoter-bound holoenzymes, together with advances in our understanding of how salts and solutes interact with biopolymer surfaces and perturb biopolymer processes, has led to rapid progress in our understanding of the events of RNAP recruitment and promoter recognition to form the initial closed complex RP_c, and the massive conformational changes in RNAP and promoter DNA that occur to convert it to the most stable open complex RP_o. Challenges for the future include developing a molecular understanding of how the start site region is opened and how the T strand is placed in the active site; how conformational changes in the cleft involving σ_1.1 and the down stream NT strand in the conversion of the initial open complex (I₂) to RP_o are sensed by the assembling downstream clamp/jaw apparatus; how upstream DNA trajectory and interactions with the αCTDs and the upstream clamp allow the entry of downstream duplex DNA into the cleft; and how all these steps of isomerization are regulated by DNA sequence, factors, ligands, and environmental variables in the response of the cell to changing growth conditions or stress.

Abbreviations used

RNAP

RNA polymerase

NT

nontemplate strand

T

template strand

αCTD

α-subunit C-terminal domain

FRET

fluorescence resonance energy transfer

myxo

myxopyronin

lpm

lipiarmycin

SI

sequence insertion