Utilization of variably spaced promoter-like elements by the bacterial RNA polymerase holoenzyme during early elongation

SUMMARY

The bacterial RNA polymerase (RNAP) holoenzyme consists of a catalytic core enzyme in complex with a σ factor that is required for promoter-specific transcription initiation. During initiation, members of the σ⁷⁰ family of σ factors contact two conserved promoter elements, the −10 and −35 elements, that are separated by ~17 base pairs (bp). σ⁷⁰ family members contain four flexibly linked domains. Two of these domains, σ₂ and σ₄, contain determinants for interactions with the promoter −10 and −35 elements, respectively. σ₂ and σ₄ also contain core-binding determinants. When bound to core the inter-domain distance between σ₂ and σ₄ matches the distance between promoter elements separated by ~17 bp. Prior work indicates that during early elongation the nascent RNA-assisted displacement of σ₄ from core can enable the holoenzyme to adopt a configuration in which σ₂ and σ₄ are bound to “promoter-like” DNA elements separated by a single base pair. Here we demonstrate that holoenzyme can also adopt configurations in which σ₂ and σ₄ are bound to “promoter-like” DNA elements separated by 0, 2 or 3 bp. Thus, our findings suggest that displacement of σ₄ from core enables the RNAP holoenzyme to adopt a broad range of “elongation-specific” configurations.

INTRODUCTION

The bacterial RNA polymerase (RNAP) holoenzyme consists of a catalytic core enzyme (α₂ββ′ω) complexed with a σ factor. σ factors confer on the holoenzyme the ability to initiate promoter-specific transcription. Bacteria typically contain a number of σ factors, each specifying recognition of a distinct class of promoters (Gross et al., 1998, Gruber & Gross, 2003). The primary σ factor in Escherichia coli is σ⁷⁰, and a typical σ⁷⁰-dependent promoter bears two conserved sequence elements, the −10 and the −35 elements (consensus sequences: TATAAT and TTGACA, respectively), which are separated by a spacer of ~17 base pairs (bp). Members of the σ⁷⁰ family of σ factors share four regions of conserved sequence, regions 1–4, which have been further subdivided (Lonetto et al., 1992, Gross et al., 1998, Murakami & Darst, 2003, Paget & Helmann, 2003). Structural work indicates that σ⁷⁰ family members contain four flexibly linked domains σ_1.1 (containing region 1.1), σ₂ (containing regions 1.2–2.4), σ₃ (containing regions 3.0 and 3.1) and σ₄ (containing regions 4.1 and 4.2) (Malhotra et al., 1996, Gross et al., 1998, Severinova et al., 1996, Campbell et al., 2002, Murakami & Darst, 2003, Paget & Helmann, 2003).

σ₂ and σ₄ contain core-binding determinants as well as DNA-binding domains responsible for sequence-specific interactions with the promoter −10 element and −35 element, respectively (Gross et al., 1998, Murakami & Darst, 2003). Interaction between σ₂ and a domain of the β′ subunit known as the clamp helices (also referred to as the β′ coiled coil) is required for both the stable association of σ with core and sequence-specific interaction between σ₂ and the promoter −10 element (Arthur & Burgess, 1998, Young et al., 2001, Young et al., 2004). Interaction between σ₄ and the flap domain of the β subunit (the β flap), while not essential for the stable association of σ with core, is required for sequence-specific interaction between σ₄ and a promoter −35 element separated by ~17 bp from a promoter −10 element (Kuznedelov et al., 2002). In particular, when σ₄ is bound to the β flap and σ₂ is bound to the β′ clamp helices, the distance between σ₂ and σ₄ matches the distance between promoter elements separated by ~17 bp (Murakami et al., 2002a, Murakami et al., 2002b, Vassylyev et al., 2002, Murakami & Darst, 2003). Moreover, when holoenzyme is reconstituted in vitro using a mutant core lacking a portion of the β flap containing determinants for interaction with σ₄, the interdomain distance between σ₂ and σ₄ decreases, resembling the distance between σ₂ and σ₄ observed in the context of free σ (Callaci et al., 1999, Kuznedelov et al., 2002).

During the transition from transcription initiation to transcription elongation, the interaction between σ and the remainder of the transcription complex is weakened (reviewed in Mooney et al., 2005). Weakening of the σ-core interaction is due (at least in part) to a nascent RNA-mediated sequential displacement of discrete σ-core contacts, leading to a partial release of σ from the transcription complex (reviewed in Mooney et al., 2005). Structures of RNAP indicate that two regions of σ lie along the path of the nascent RNA (Murakami et al., 2002a, Murakami et al., 2002b, Vassylyev et al., 2002, Vassylyev et al., 2007): (1.) σ region 3.2, an extended linker that connects σ₃ to σ₄, which lies buried within the RNA exit channel (the channel through which the nascent RNA is extruded during transcription elongation) and (2.) σ₄, which by virtue of its interaction with the β flap is positioned directly adjacent to the end of the RNA exit channel. Thus, the nascent RNA first displaces σ region 3.2 from the RNA exit channel when the nascent RNA enters the channel (Marr et al., 2001), at a length of ~10–11 nucleotides (nt), and second, destabilizes the σ₄-β flap interaction when the nascent RNA emerges from the RNA exit channel, at a length of ~16 nt (Nickels et al., 2005, Nickels et al., 2006). Contacts between σ₂ and the β′ clamp helices can be maintained during transcription elongation enabling σ to remain associated with (or rebind to) the elongation complex (reviewed in Mooney et al., 2005).

Studies of the bacteriophage λ Q antiterminator protein (λQ) have illustrated that displacement of σ region 3.2 and σ₄ from core during early elongation confers unique properties upon the transcription complex (Roberts et al., 1998, Marr et al., 2001, Nickels et al., 2002, Nickels et al., 2006). λQ is an operon-specific elongation factor that is required for expression of the bacteriophage λ late genes, which are under the control of the late promoter λP_R′ (reviewed in Roberts et al., 1998) (Figure 1). λQ, which engages the RNAP holoenzyme during early elongation, requires three specialized cis-acting sequence elements embedded within the λP_R′ promoter region to efficiently engage RNAP (Figure 1): (1.) a Q binding element (QBE) located between the λP_R′ promoter −10 and −35 elements, (2.) a pause-inducing element located in the initial transcribed region of λP_R′ and (3.) a TTGACT motif located one base pair upstream of the pause-inducing element.

Figure 1. Sequence of events at λP_R′.

A. Presence of λQ (shown in blue) allows RNAP that has initiated transcription from λP_R′ to read through terminator t_R′. Blow-up shows the functionally important elements at λP_R′ including: the promoter −10 and −35 elements, the λQ-binding element (QBE), the pause-inducing −10-like element, and the −35-like element. Note that the −35-like element is separated by a single base pair from the pause-inducing −10-like element.

B. Sequence of events during initiation and early elongation at λP_R′. Top panel: the initiation complex; σ⁷⁰ region 4 (σ₄) is shown bound to the β flap (blue triangle).σ⁷⁰ region 3.2 (σ_3.2) is dashed to indicate that it is located within the RNA exit channel (Murakami et al., 2002b, Vassylyev et al., 2002). Middle panel: the paused early elongation complex at λP_R′ with σ⁷⁰ region 2 (σ₂) bound to the pause-inducing −10-like element (shown in red). The nascent RNA (pink and red circles) has displaced σ_3.2 from the RNA exit channel (Marr et al., 2001) and has destabilized the σ₄-β flap interaction (Nickels et al., 2005). Bottom panel: λQ-engaged complex; λQ (shown as a dimer) binds to the QBE and engages the paused elongation complex making protein-protein interactions with both the β flap and σ₄. λQ, via protein-protein interaction with σ₄, stabilizes interactions between σ₄ and the −35-like element (grey rectangle) causing the holoenzyme to adopt a conformation in which σ₂ and σ₄ are simultaneously bound to promoter-like DNA elements that are separated by a single base pair (Nickels et al., 2002).

Interactions between RNAP and the pause-inducing element, which resembles a promoter −10 element, cause the RNAP holoenzyme to pause during early elongation after addition of the 16th or 17th nucleotide (nt) to the nascent RNA transcript (reviewed in Roberts et al., 1998). Pausing of holoenzyme is mediated by protein-DNA interaction between σ₂ and the pause-inducing −10-like element (Figure 1) (Ring et al., 1996). λQ binds the QBE in the context of this paused elongation complex and stabilizes the binding of σ₄ to the TTGACT motif. Interactions that occur between QBE-bound λQ and the paused transcription complex facilitate λQ’s stable association with the elongation complex once RNAP has escaped from the pause and λQ has relinquished contacts with the QBE (Yarnell & Roberts, 1999, Deighan & Hochschild, 2007). λQ’s association with RNAP alters the elongation properties of RNAP, making RNAP resistant to transcription pausing (i.e., RNAP exhibits increased processivity) and insensitive to downstream termination signals.

λQ’s initial engagement with RNAP requires λQ to make direct contact with a surface of the RNAP core enzyme, the β flap, which is occluded by σ₄ during transcription initiation (Nickels et al., 2006, Deighan et al., 2008). At least two factors facilitate productive interaction between QBE-bound λQ and the β flap during the engagement process. First, pausing occurs at a nascent RNA length of 16 nt or 17 nt. Thus, in the context of the paused transcription complex at λP_R′, the nascent RNA is of a length at which σ region 3.2 has been displaced from the RNA exit channel and the σ₄-β flap interaction is destabilized (Marr et al., 2001, Nickels et al., 2005, Nickels et al., 2006) (Figure 1). Second, protein-protein interaction between λQ and σ₄ stabilizes a conformation of the holoenzyme with σ₄ fully displaced from the β flap and bound to the TTGACT motif (Nickels et al., 2002) (Figure 1).

The TTGACT motif, which resembles a promoter −35 element, is located one base pair upstream of the pause-inducing −10-like element. Thus, as a consequence of both the nascent RNA-mediated destabilization of the σ₄-β flap interaction and the λQ-σ₄ interaction, the RNAP holoenzyme adopts a conformation in which σ₂ and σ₄ are simultaneously bound to “promoter-like” DNA elements separated by a single base pair (Nickels et al., 2002) (Figure 1B). The geometry of the λQ-engaged complex stands in striking contrast to the geometry of a typical initiation complex with σ₂ and σ₄ bound to promoter elements separated by ~17 bp. Thus, formation of an apparent “elongation-specific” conformation of the RNAP holoenzyme facilitates productive interaction between λQ and RNAP.

Here we show that, in the context of the λQ-engaged complex, the RNAP holoenzyme can access several distinct configurations in which σ₂ and σ₄ are bound to closely spaced DNA elements. In particular, we provide evidence that displacement of σ₄ from the β flap enables the σ⁷⁰–containing holoenzyme to adopt configurations in which σ₂ and σ₄ are simultaneously bound to promoter-like DNA elements separated by 0, 1, 2 or 3 bp. Our findings suggest that the structural flexibility provided by the partial release of σ from the core enzyme enables the σ⁷⁰-containing holoenzyme (and holoenzyme species containing other σ⁷⁰ family members) to access a broad range of “elongation-specific” configurations.

RESULTS

Preliminary considerations

Displacement of σ region 3.2 from the RNA exit channel and σ₄ from the β flap (while σ₂ remains bound to the β′ clamp helices) is predicted to result in σ₄ being tethered to the remainder of the transcription complex by a flexible ~30 amino acid linker composed primarily of σ region 3.2. Previous work has shown that in the context of the λQ-engaged complex at λP_R′, displacement of both σ region 3.2 and σ₄ from core enables the RNAP holoenzyme to adopt an conformation in which σ₂ and σ₄ bind “promoter-like” DNA elements separated by a single base pair (Nickels et al., 2002). Based on the presumed flexibility afforded to σ₄ during early elongation we hypothesized that additional configurations in which σ₂ and σ₄ are bound to closely spaced promoter-like DNA elements would be accessible to the RNAP holoenzyme. To investigate this hypothesis we wished to test whether the RNAP holoenzyme could access a range of such “elongation-specific” configurations during the λQ-engagement process. To do this, we tested the effects of altering the position of the −35-like element relative to the pause-inducing −10-like element in the context of λP_R′.

Design and construction of λP_R′-based templates with varied spacing between the −35-like element and the −10-like element

We constructed a panel of λP_R′-based templates in which we varied the position of the pause-inducing −10-like element relative to the position of the −35-like element such that these hexamers were separated by 0, 1, 2 or 3 bp (Figure 2). To facilitate this analysis we designed templates bearing pause-inducing sequence elements that matched (or nearly matched) a consensus promoter −10 element (TATAAT) and −35-like elements that fully matched a consensus promoter −35-element (TTGACA). For each of these templates, the sequences upstream of the −35-like element and the sequences downstream of the pause-inducing sequence element were identical to those present in the context of λP_R′. We refer to this collection of templates as the “−10/−35 spacing derivatives”.

Figure 2. λP_R′ promoter derivatives analyzed.

Shown is the sequence of the wild-type λP_R′ promoter and the sequence of each of the five “−10/−35 spacing derivatives”. Indicated are the promoter −10 and −35 elements, the QBE, the pause-inducing −10-like element, the −35-like element, the transcription start-site (+1), and the primary position where RNAP pauses on each template (see Figure 3).

σ⁷⁰-dependent pausing in the context of the −10/−35 spacing derivatives in vitro

To establish whether the pause-inducing sequence elements associated with each of the −10/−35 spacing derivatives facilitate σ⁷⁰-dependent pausing during early elongation we performed single round in vitro transcription experiments (Figure 3). We performed these assays as a time course, monitoring the RNA content of each reaction at discrete times after the initiation of transcription. RNA transcripts generated in these reactions were labeled specifically at their 5′ end by inclusion of γ-³²P-ATP in the transcription reactions.

Figure 3. σ-dependent pausing on the −10/−35 spacing derivatives.

Shown are the results of in vitro transcription assays using the linear templates depicted in Figure 2. Aliquots of each transcription reaction were removed and stopped at the indicated times after the initiation of transcription. The transcripts were end-labeled with γ-³²P-ATP. The 16 nt and 17 nt RNA species that appear in reactions performed with a template carrying the wild-type λP_R′ promoter are indicated (+16, +17) along with the ~200 nt terminated transcript (T).

In the context of λP_R′, the pause-inducing sequence element is located between positions +1 and +6 (relative to the position of the transcription start-site, +1) and pausing is manifest in transcription complexes carrying a 16 or 17 nt nascent RNA transcript (Ring et al., 1996, Roberts et al., 1998). Thus, in reactions performed using a wild-type λP_R′ template, a prominent 17 nt RNA species and a less prominent 16 nt RNA RNA species were observed (Figure 3, lanes 1–5).

When reactions were performed using either the “0 bp” or “1 bp” templates, which carry pause-inducing elements located between positions −1 and +5, a prominent 16 nt RNA species was observed (Figure 3, lanes 6–15). When reactions were performed using the “2 bp” template, which carries a pause-inducing element located between positions +1 and +6, a prominent 17 nt RNA species was observed (Figure 3, lanes 16–20). Finally, when reactions were performed using the “3 bp” template, which carries a pause-inducing element located between positions +2 and +7, a prominent 18 nt RNA species was observed (Figure 3, lanes 21–25). We infer that the 16 nt RNA species observed in reactions performed with the “0 bp” and “1 bp” templates, the 17 nt RNA species observed in reactions performed with the “2 bp” template and the 18 nt RNA species observed in reactions performed with the “3 bp” template are products of σ⁷⁰-dependent pausing on each of these templates. Furthermore, there is a direct correspondence between the position of the pause-inducing element relative to the transcription start site and the position where σ⁷⁰-dependent pausing is observed. In particular, when the pause element is located between positions −1 and +5, pausing is manifest primarily in complexes carrying a 16 nt RNA transcript, when the pause element is located between positions +1 and +6, pausing is manifest primarily in complexes carrying a 17 nt RNA transcript, and when the pause element is located between positions +2 and +7, pausing is manifest primarily in complexes carrying a 18 nt RNA transcript. Thus, a single base pair shift in the position of the pause-inducing element relative to the transcription start-site results in a single nucleotide difference in the length of the RNA associated with the paused complexes.

In contrast to the RNA transcripts observed in reactions performed using the wild-type λP_R′ template, the RNA transcripts observed in reactions performed using each of the −10/−35 spacing derivatives did not disappear over the 8 minute time-course of the assays (Figure 3). Thus, the pause half-life of complexes formed on each of the −10/−35 spacing derivatives is greater than the pause half-life of complexes formed on the wild-type λP_R′ template. The increased stability of the σ⁷⁰-dependent paused complexes formed on the −10/−35 spacing derivatives likely is a result of the stability of protein-DNA interactions between σ₂ and the pause-inducing elements present on these templates. In particular, the wild-type pause-inducing element is only a 3/6 match (AACGAT) to a consensus promoter −10 element (TATAAT). In contrast, the pause-inducing elements associated with the −10/−35 spacing derivatives match (or nearly match) a consensus promoter −10 element. Thus, interactions between σ₂ and the pause-inducing element present on each of −10/−35 spacing derivatives are likely more stable than those between σ₂ and the wild-type pause-inducing element, leading to a decreased efficiency of escape from the pause.

λQ-dependent antitermination in the context of the −10/−35 spacing derivatives in vitro

To assess whether the −10/−35 spacing derivatives could be used as tools to probe configurations of the RNAP holoenzyme during early elongation we sought to determine whether λQ could functionally engage RNAP on each of the −10/−35 spacing derivatives in vitro. To do this we performed single round in vitro transcription experiments in the presence of increasing concentrations of λQ using the linear templates depicted in Figure 2 that contained the t_R′ terminator ~200 bp downstream of the transcription start-site. Thus, the percentage of full-length transcripts (% read through) emanating from each promoter in the presence of λQ provides a measure of the ability of λQ to functionally engage RNAP. Assays were performed in the presence of NusA, a highly conserved transcription elongation factor that enhances λQ antitermination activity at λP_R′ in vitro, at least in part by stabilizing λQ’s interaction with the paused transcription complex (Yarnell & Roberts, 1992).

λQ-dependent antitermination activity could be observed in reactions performed with each of the −10/−35 spacing derivatives (Figure 4A). The maximal λQ activity observed under the reaction conditions used for the assays shown in Figure 4A ranged from a low of ~15% (on the “0 bp” template) to a high of ~60% (on the “2 bp” and “3 bp” templates). We conclude that λQ can functionally engage RNAP on each of the −10/−35 spacing derivatives in vitro.

Figure 4. λQ-dependent antitermination on the −10/−35 spacing derivatives in vitro: effects of disrupting interactions between σ₂ and the pause-inducing element.

A. λQ-dependent antitermination in vitro. Results of single-round in vitro transcription assays performed in the presence of increasing concentrations of λQ (5 nM, 20 nM, 100 nM and 500 nM) using the indicated template. Graphs show the percentage of transcripts derived from terminator readthrough (readthrough/[readthrough + terminated]). In the absence of λQ less than 1% readthrough was observed on each template (data not shown).

B. Effects of mutations in the pause-inducing element on λQ-dependent antitermination in vitro. Results of single-round in vitro transcription assays performed in the presence of 500nM λQ using either a linear template carrying the indicated wild-type −10/−35 spacing derivative (grey bars) or a linear template carrying base pair mutations in the pause-inducing element (black bars). Graphs show the percentage of transcripts derived from terminator readthrough [readthrough/(readthrough + terminated)]. Plotted on the graphs are the mean and SEM of two independent measurements.

We note that λQ antitermination activity was severely reduced on templates carrying more than 3 bp between the −35-like element and the pause-inducing −10-like element (data not shown). These findings are consistent with prior observations (Ring et al., 1996) and likely reflect geometrical constraints on the location of the QBE relative to the pause-inducing element.

Effects of mutating the pause-inducing −10-like element on Q-dependent antitermination in vitro

In the context of wild-type λP_R′, base-pair substitutions that destabilize protein-DNA interactions between σ₂ and the pause-inducing element compromise the ability of λQ to functionally engage RNAP (Yarnell & Roberts, 1992, Ring & Roberts, 1994, Ring et al., 1996). We wished to establish whether the functional integrity of the pause-inducing element was similarly required for the λQ-dependent antitermination observed on each of the −10/−35 spacing derivatives. To do this, we determined the effects of base pair substitutions in the pause-inducing elements associated with each of the −10/−35 spacing derivatives. In particular, we constructed mutant templates bearing both an A to G mutation at position 2 of the pause-inducing element and a T to G mutation at position 6 of the pause-inducing element. Introduction of these base pair substitutions into the pause-inducing elements associated with each of the −10/−35 spacing derivatives severely reduced λQ-dependent antitermination (Figure 4B). Thus, the functional integrity of the pause-inducing element is required for the λQ-dependent antitermination observed in reactions performed with each of the −10/−35 spacing derivatives (Figure 4A).

[We note that for reactions done using the “0 bp” template, the concentration of NusA used in the assays shown in Figure 4B was 3-fold higher than the concentration of NusA used in assays shown in Figure 4A (450 nM versus 150 nM). This 3-fold increase in the concentration of NusA enhanced the λQ-dependent antitermination activity observed in reactions using the “0 bp” template ~2-fold (Figure 4). In contrast, the same 3-fold increase in the concentration of NusA moderately enhanced the λQ-dependent antitermination activity observed in reactions using the “1 bp” template and did not affect the λQ-dependent antitermination activity observed in reactions using the “2 bp” template or “3 bp” template (data not shown).]

Effects of disrupting interactions between σ₄ and the −35-like element on Q-dependent antitermination in vitro

In the context of wild-type λP_R′, mutations that disrupt the protein-DNA interaction between σ₄ and the −35-like element or the protein-protein interaction between λQ and σ₄, reduce λQ-dependent antitermination (Nickels et al., 2002, Nickels et al., 2006). We therefore wished to establish whether disrupting the interaction between σ₄ and the −35-like element or the protein-protein interaction between λQ and σ₄ similarly reduced the efficiency with which λQ functionally engages RNAP on each of the −10/−35 spacing derivatives.

First, to determine the effects of disrupting the interaction between σ₄ and the −35-like element, we tested the effects of base pair substitutions in the −35-like element that destabilize interactions with σ₄. In particular, we introduced a G to A mutation at position 3 of the −35-like element (mutant sequence TTAACA) or a C to A mutation at position 5 of the −35-like element (mutant sequence TTGAAA). Introduction of either of these mutations into the −35-like element associated with each of the −10/−35 spacing derivatives significantly reduced the efficiency of λQ-dependent antitermination (Figure 5). Next, we directly established that the effects of the G to A mutations or C to A mutations were due to the disruption of the binding of σ₄ to the −35-like element. To do this we took advantage of two amino acid substitutions in σ₄: (1.) R588H, which suppresses a G to A mutation at position 3 in the context of a promoter −35 element (Gardella et al., 1989) and (2.) R584A, which suppresses a C to A mutation at position 5 in the context of a promoter −35 element (Gregory et al., 2005). We found that λQ-dependent antitermination activity on templates carrying the G to A mutation (TTAACA) was significantly increased when reactions were performed with holoenzyme reconstituted with σ⁷⁰ R588H compared to reactions performed with wild-type holoenzyme. Similarly we found that λQ-antitermination activity on templates carrying the C to A mutation (TTGAAA) was significantly increased when reactions were performed with holoenzyme reconstituted with σ⁷⁰ R584A compared to reactions performed with wild-type holoenzyme. The suppressive effects of the R588H and R584A substitutions were template specific; i.e. the R588H substitution did not exhibit a suppressive effect in reactions performed with templates carrying the C to A mutation and the R584A substitution did not exhibit a suppressive effect in reactions performed with templates carrying the G to A mutation. We therefore conclude that the G to A and C to A mutations affect λQ function in the context of the −10/−35 spacing derivatives by disrupting the binding of region σ₄ to the −35-like element.

Figure 5. λQ-dependent antitermination on the −10/−35 spacing derivatives in vitro: effects of disrupting interactions between σ₄ and the −35-like element.

Results of single-round in vitro transcription assays performed using the indicated template in the presence of 500 nM λQ (for the “0 bp”, “1 bp”, and “3 bp” templates) or 5 nM λQ (for the “2 bp” templates). Reactions were performed using holoenzyme reconstituted with wild-type σ⁷⁰, σ⁷⁰ A553D, σ⁷⁰ R588H, or σ⁷⁰ R584A, as indicated. Graphs show the percentage of transcripts derived from terminator readthrough (readthrough/[readthrough + terminated]). Plotted on the graphs are the mean and SEM of three independent measurements.

Second, to determine the effects of disrupting the protein-protein interaction between λQ and σ₄ in the context of the −10/−35 spacing derivatives we tested the effect of an amino acid substitution in σ₄ (A553D) that disrupts the λQ-σ₄ interaction (Nickels et al., 2002). We observed a reduction in λQ-dependent antitermination when reactions were performed with holoenzyme reconstituted with σ⁷⁰ A553D compared to reactions performed with wild-type holoenzyme on each of the −10/−35 spacing derivatives (Figure 5).

Taken together the results presented in Figure 5 establish that disrupting the protein-DNA interaction between σ₄ and the −35-like element or the protein-protein interaction between λQ and σ₄, reduces λQ-dependent antitermination on each of the −10/−35 spacing derivatives in vitro.

Effects of disrupting interactions between σ₄ and the −35-like element on Q’s engagement with the paused early elongation complex in vitro

Prior work has established that disrupting the interaction between σ₄ and the −35-like element reduces the stability of the complex that forms when λQ initially engages RNAP at λP_R′ (Nickels et al., 2002). We therefore wished to determine whether disrupting the interaction between σ₄ and the −35-like element similarly reduces the stability of complexes that form when λQ initially engages RNAP in the context of the −10/−35 spacing derivatives. The stability of λQ-engaged paused complexes can be assessed using an exonuclease challenge assay (Yarnell & Roberts, 1992). In this assay, transcription complexes are artificially stalled at the pause site by withholding CTP. After the stalled complexes are formed they are incubated with λQ. The λQ-bound complexes are challenged with exonuclease III, which digests the DNA in a 3′ to 5′ direction until its progress is blocked by the presence of a DNA-bound protein (Figure 6A). Several barriers arise; the barrier at position −32 is due to λQ, the barriers at positions −21 and −11 are due to σ⁷⁰ in the paused transcription complex, while the barrier at position +4 is due to RNAP core (Yarnell & Roberts, 1992).

Figure 6. Effects of disrupting interactions between σ₄ and the −35-like element on the stability of λQ-engaged paused complexes in vitro.

A. Schematic of templates used for exonuclease III challenge assay. Template is end labeled at the 5′ end of the bottom strand as indicated (black star); digestion of DNA from the downstream end is blocked by presence of bound EcoRIGln111 (grey oval). Positions at which the progress of exonuclease III digestion is blocked by bound protein are indicated: −32 (λQ-dependent barrier), −21 and −11 (σ⁷⁰-dependent barriers), −4 (RNAP core-dependent barrier).

B. Exonuclease challenge assays. Assays were done using wild-type versions of each template (WT) or mutant versions carrying a base pair substitution in the −35-like element that disrupts the interaction with σ₄ (MUT −35) or a base pair substitution in the QBE that weakens λQ’s binding to the QBE (MUT QBE). λQ was added to 200 nM for reactions performed using the λP_R′ template or the 1 bp template, 50 nM for reactions performed using the 2 bp template and 100 nM for reactions performed using the 3 bp template. Assignment of the barriers was made by comparison to a DNA sequencing ladder run on the same gel (not shown). (We do not know the origin of the additional band, indicated by asterisk, which appears in reactions performed using the 1 bp template).

A λQ-dependent barrier to exonuclease digestion does not form when λQ is added to template DNA alone or when reactions are performed using a template carrying a base pair substitution that disrupts the interaction between λQ and the QBE (Yarnell & Roberts, 1992) (Figure 6B). Thus, the binding of λQ to the QBE is stabilized by the λQ-RNAP interaction and, furthermore, the half-life of the λQ-dependent barrier to exonuclease III digestion provides a direct measure of the stability of the complex that forms when QBE-bound λQ initially engages RNAP.

As shown in Figure 6B, a base pair substitution that disrupts the interaction between σ₄ and the −35-like element significantly reduces the stability of the λQ-dependent barrier to exonuclease III digestion in the context of wild-type λP_R′, consistent with prior observations (Nickels et al., 2002). Under the conditions of these exonuclease challenge assays, σ₂ is bound to the pause-inducing −10-like element, as indicated by the appearance of the −21 and −11 barriers (Yarnell & Roberts, 1992). Thus, the finding that disrupting the interaction between σ₄ and the −35-like element reduces the stability of the complex that forms when λQ initially engages RNAP at λP_R′ indicates that λQ can stabilize a conformation of the holoenzyme in which σ₂ and σ₄ are simultaneously bound to promoter-like DNA elements separated by a single base pair (Nickels et al., 2002).

To establish whether λQ can stabilize additional conformations of the holoenzyme in which σ₂ and σ₄ are simultaneously bound to closely spaced promoter-like DNA elements we performed exonuclease challenge assays using each of the −10/−35 spacing derivatives (Figure 6B). In the absence of λQ, barriers corresponding in position to the σ⁷⁰-dependent barriers seen with the wild-type λP_R′ template (−21 and −11) were observed (Figure 6B and data not shown), indicating that σ₂ is bound to the pause-inducing −10-like element in assays performed with each of the −10/−35 spacing derivatives.

(We note that the “−11” barrier exhibits a longer half-life in reactions performed with the −10/−35 spacing derivatives than in reactions performed using the wild-type λP_R′ template. Therefore we do not observe a “−4” barrier in these reactions. We suspect that the increased half-life of the “−11” barrier reflects the increased stability of protein-DNA interactions between σ₂ and the consensus pause-inducing elements present on the −10/−35 spacing derivative templates.)

Addition of λQ to stalled complexes formed using the “1 bp”, “2 bp”, or “3 bp” template resulted in the appearance of a barrier to exonuclease III digestion that corresponds in position to the λQ-dependent barrier observed with the wild-type λP_R′ template (Figure 6B). The half-life of the λQ-dependent barrier was longer in reactions performed with the “2 bp” and “3 bp” templates compared with the “1 bp” template (Figure 6B). We did not detect a λQ-dependent barrier to exonuclease III digestion in reactions performed with the “0 bp” template (data not shown), suggesting that the λQ-engaged complex formed on this template is not stable enough to be detected by this method.

Next, we determined the effects of disrupting the interaction between σ₄ and the −35-like element on the “1 bp”, “2 bp” and “3 bp” templates. To do this, we performed assays using templates that carried a G to A mutation at position 3 of the −35-like element (mutant sequence TTAACA). Disrupting the interaction between σ₄ and the −35-like element strongly reduced the half-life of the λQ-dependent barrier to exonuclease III digestion (Figure 6B). Therefore, these results establish that disrupting the interaction between σ₄ and the −35-like element reduces the stability of the complex that forms when λQ initially engages RNAP in the context of the “1 bp”, “2 bp” or “3 bp” template. Furthermore, these data provide strong evidence that λQ can stabilize conformations of the holoenzyme in which σ₂ and σ₄ are simultaneously bound to closely spaced promoter-like DNA elements in the context of the “1 bp”, “2 bp” and “3 bp” templates.

As mentioned above, we were unable to detect a λQ-dependent barrier to exonuclease digestion in reactions performed using the “0 bp” template. Therefore we could not use the exonuclease challenge assay to obtain evidence that λQ can stabilize a conformation of the holoenzyme in which σ₂ and σ₄ are simultaneously bound to promoter-like DNA elements separated by 0 bp. Nevertheless, given that λQ’s ability to functionally engage active transcription complexes depends upon the functional integrity of the pause-inducing −10-like element and the −35-like element in the context of each −10/−35 spacing derivative (Figures 4 and 5), the mechanism of engagement is likely the same. Therefore, our finding that σ₂ and σ₄ can simultaneously bind promoter-like elements separated by 1, 2, or 3 bp in the context of the “1 bp”, “2 bp”, and “3 bp” templates (Figure 6) suggests that σ₂ and σ₄ can simultaneously bind the promoter-like elements separated by 0 bp in the context of the “0 bp” template as well.

λQ-dependent modification of RNAP in the context of the −10/−35 spacing derivatives in vivo

We next examined the ability of λQ to functionally engage RNAP on each of the −10/−35 spacing derivatives in vivo. To do this, we fused sequences derived from each −10/−35 spacing derivative to a lacZ reporter gene; the sequences that were fused to lacZ extended from ~100 bp upstream of the transcription start-site to ~240 bp downstream of the transcription and include the terminator t_R′ (Figure 7A). The level of lacZ expression from these fusion constructs can therefore report on the ability of plasmid-encoded λQ to function as an antiterminator for transcripts initiating from each of the −10/−35 spacing derivatives.

Figure 7. λQ-dependent antitermination on the −10/−35 spacing derivatives in vivo.

A. Schematic of lacZ fusions used to examine λQ-dependent antitermination on the −10/−35 spacing derivatives in vivo.

B. Results of β-galactosidase assays performed with reporter strain cells harboring the indicated −10/−35 spacing derivative-lacZ reporter. Cells contained a plasmid that directed the synthesis of λQ, under the control of an IPTG-inducible promoter, or a plasmid that did not direct the synthesis of λQ. Plotted on the graph is the fold-stimulation (β-galactosidase activity observed in cells carrying λQ/β-galactosidase activity observed in cells not carrying λQ) determined when cells were grown in LB supplemented with no IPTG or with 2 μM, 20 μM, or 100 μM IPTG.

C. Results of β-galactosidase assays performed with reporter strain cells harboring the indicated −10/−35 spacing derivative-lacZ reporter. Cells contained a plasmid that directed the synthesis of λQ (grey bars), under the control of an IPTG-inducible promoter, or a plasmid that did not direct the synthesis of λQ (black bars), Plotted on the graphs are the mean and SEM of five independent measurements. For each −10/−35 spacing derivative, the level of λQ induction was chosen to reveal the maximal effect of disrupting the interaction between σ₄ and the −35-like element or the interaction between σ₄ and λQ. Cells containing the “0 bp” spacing derivatives were grown in the presence of 100 μM IPTG, cells containing the “3 bp” spacing derivatives were grown in the presence of 5 μM IPTG, whereas cells containing the “1 bp” and “2 bp” derivatives were grown in the absence of IPTG.

We introduced these lacZ reporter constructs in single copy into a lacZ⁻ strain of Escherichia coli and assayed the ability of plasmid-encoded λQ to induce the expression of the lacZ reporter gene. λQ increased lacZ reporter gene expression in each of these reporter strains, suggesting that λQ could modify transcription complexes initiating from each of the −10/−35 spacing derivatives in vivo (Figure 7B).

We then determined the effects of mutating the pause-inducing −10-like element associated with each −10/−35 spacing derivative (Figure 7C). Introduction of base pair substitutions into the pause-inducing element of each template essentially abolished the λQ-dependent increase in lacZ expression observed in strains containing unmutated templates (compare the first set of bars with the bars labeled “Δ pause”). [We note that base pair substitutions in the pause-inducing elements increased the basal level of lacZ expression observed in the absence of λQ (Figure 7C). In vitro transcription experiments indicated that these base pair substitutions did not increase the levels of terminator readthrough observed in the absence of λQ (data not shown). Therefore, we suspect that the increase in lacZ expression observed with each “Δ pause” derivative reflects a reduction in σ⁷⁰-dependent pausing that leads to a greater density of RNAP along the lacZ gene and not a reduction in the efficiency of the t_R′ terminator on the mutated templates (Goliger et al., 1989, Telesnitsky & Chamberlin, 1989).]

Next, we determined the effects of disrupting the interaction between σ₄ and the −35-like element or the interaction between σ₄ and λQ. To determine the effect of disrupting the interaction between σ₄ and the −35-like element we introduced either a G to A mutation at position 3 of the −35-like element (mutant sequence TTAACA) or a C to A mutation at position 5 of the −35-like element (mutant sequence TTGAAA). Introduction of these substitutions in the −35-like element associated with each −10/−35 spacing derivative significantly reduced the λQ-dependent increase in lacZ expression (Figure 7C; compare the first, third and fifth set of bars). To determine the effect of disrupting the protein-protein interaction between λQ and σ₄ we generated versions of the reporter strains that carried an A553D mutation in rpoD (the gene encoding σ⁷⁰). Compared to strains carrying a wild-type rpoD gene, the λQ-dependent increase in lacZ expression was reduced in reporter strains carrying rpoD-A553D (Figure 7C; compare the first and second set of bars). The rpoD-A553D mutation did not significantly reduce lacZ expression in strains already containing base pair substitutions in the −35-like element. This finding is consistent with the proposal that the rpoD-A553D mutation and substitutions in the −35-like element both affect the same step.

We note that the effects of disrupting the interaction between σ₄ and the −35-like element or the interaction between σ₄ and λQ were more pronounced in cells containing sub-saturating levels of λQ (Figure 7C and data not shown). Furthermore, for each −10/−35 spacing derivative, the maximal effect of disrupting the interaction between σ₄ and the −35-like element or the interaction between σ₄ and λQ was revealed at distinct intracellular levels of λQ. Thus, the data presented in Figure 7C show the results of assays performed where the levels of λQ are such that the maximal effect of disrupting the interaction between σ₄ and the −35-like element or the interaction between σ₄ and λQ are revealed.

Taken together, the results presented in Figure 7C establish that disrupting the protein-DNA interaction between σ₂ and the pause-inducing −10-like element, the protein-DNA interaction between σ₄ and the −35-like element, or the protein-protein interaction between λQ and σ₄ reduces λQ-dependent antitermination on each of the −10/−35 spacing derivatives in vivo.

DISCUSSION

Here we provide evidence that the holoenzyme can access a broad range of “elongation-specific” configurations in which σ₂ and σ₄ are bound to variably spaced promoter-like DNA elements. Prior work had demonstrated that, in the context of wild-type λP_R′, λQ can stabilize a conformation of the holoenzyme in which σ₂ and σ₄ are bound to promoter-like DNA elements separated by a single base pair (Nickels et al., 2002). This conclusion was reached based on the demonstration that λQ’s ability to functionally engage both active transcription complexes and artificially stalled elongation complexes (halted at the position of the pause-site) depends upon the functional integrity of the pause-inducing −10-like element (Yarnell & Roberts, 1992, Ring & Roberts, 1994, Ring et al., 1996) and the −35-like element (Nickels et al., 2002). The results presented in Figures 3–7 demonstrate that the functional integrity of the pause-inducing −10-like element and the −35-like element are also required for λQ function when these sequence elements are separated by 0, 2 or 3 bp. We conclude that displacement of σ₄ from the β flap enables the RNAP holoenzyme to adopt not only a conformation in which σ₄ and σ₂ are bound to promoter-like DNA elements separated by a single base pair, but also conformations in which σ₂ and σ₄ are bound to promoter-like DNA elements separated by 0, 2, or 3 bp.

Structural considerations

To gain insight into the nature of the configurations of the RNAP holoenzyme examined in this work, we modeled conformations of σ in which σ₂ and σ₄ were simultaneously bound to promoter elements separated by 0, 1, 2 or 3 bp (Figure 8). To do this, we superimposed the structure of σ₄ bound to a promoter −35 element (Campbell et al., 2002) onto the DNA upstream of the −10 element of a structural model of an RNAP holoenzyme-DNA initiation complex (Murakami et al., 2002a). Modeling σ₄ directly adjacent to the −10 element (spacing of 0 bp) results in severe steric clashes between σ₃ regions 3.0 and 3.1 and σ₄ regions 4.1 and helix 1 of 4.2. In contrast, modeling σ₄ 1 bp upstream of −10 element results in only minor clashes between σ₃ and σ₄, whereas modeling σ₄ 2 bp or 3 bp upstream of the −10 element causes no clashes with other domains of σ or with other parts of the RNAP core enzyme. The minor clashes between σ₃ and σ₄ observed in the model with σ₂ and σ₄ bound to DNA elements separated by 1 bp could be resolved if the upstream path of the DNA changes slightly (as it may when σ₄ is no longer tethered to the β flap). However, the path of the DNA would have to change significantly to accommodate the clashes between σ₃ and σ₄ observed in our model with σ₂ and σ₄ bound to DNA elements separated by 0 bp. Thus, our structural modeling suggests that the position of σ₃ relative to σ₂ observed during transcription initiation must be altered to accommodate formation of a conformation in which σ₂ and σ₄ are bound to DNA elements separated by 0 bp (and possibly 1 bp). Consistent with this idea, biochemical observations indicate that the position of σ₃ relative to DNA is altered in paused early elongation complexes compared with initiation complexes (Marr et al., 2001). Thus, displacement of the σ region 3.2 linker from the RNA exit channel during early elongation may allow σ₃ to adopt different spatial orientations with respect to σ₂ [perhaps resembling the relative positions of these domains observed in structures of free σ factors (Campbell et al., 2002), or structures of σ factors bound to anti-σ factors (Sorenson et al., 2004)].

Figure 8. Models of σ₂ and σ₄ bound to promoter-like elements separated by 0, 1, 2 or 3 bp.

Top panel shows surface rendition of an RNAP holoenzyme-DNA transcription initiation complex (Murakami et al., 2002a). The protein is shown as a molecular surface and the DNA (indigo blue) is in cartoon. Highlighted features are colored as follows: −10 promoter element, magenta; −35 promoter element, yellow; σ₂ (including region 1.2 and the non conserved region), blue; σ₃, cyan; region 3.2 linker, orange; σ₄, red; β flap tip, pink. Core RNAP is grey and rendered partially transparent. The four panels below the initiation complex show models of σ₄ (rendered in molecular surface on the right and worm on the left) bound to a −35-element positioned 0, 1, 2, or 3 bp upstream of the promoter −10 element (as indicated). For these models, the σ₄ associated with the initiation complex structure and all of core (except the β flap tip) has been removed for clarity. Images were prepared using PyMol [DeLano, W.L. The PyMOL Molecular Graphics System (2002) on World Wide Web http://www.pymol.org].

We note that the close proximity of σ₃ and σ₄ observed in our structural models indicate the potential for protein-protein interactions between these domains, which could, in principle, stabilize configurations of the holoenzyme with σ₂ and σ₄ bound to closely spaced promoter elements.

Utilization of closely spaced promoter-like elements during transcription

In the context of λP_R′, the growth of the nascent RNA facilitates displacement of σ region 3.2 from the RNA exit channel and σ₄ from the β flap. Displacement of both σ₄ and σ region 3.2 from core likely results in σ₄ being flexibly tethered to the remainder of the transcription complex by a ~30 amino acid linker. In contrast, if σ₄ were displaced from the β flap without the concomitant displacement of σ region 3.2 from the RNA exit channel, movements of σ₄ relative to the transcription complex would likely be limited. Furthermore, as discussed above, displacement of σ region 3.2 from the RNA exit channel may also facilitate the repositioning of σ₃, and thus alleviate the potential for steric clash between σ₄ and σ₃ in certain configurations of the holoenzyme in which σ₂ and σ₄ are bound to closely spaced DNA elements. Thus, conformations of the holoenzyme in which σ₂ and σ₄ are bound to closely spaced DNA elements may be readily accessible only under conditions where both σ₄ and σ region 3.2 are displaced from core.

In principle, configurations of the holoenzyme in which σ₂ and σ₄ are bound to closely spaced DNA elements could also occur during transcription initiation. In the context of λP_R′ and the −10/−35 spacing derivatives examined here, binding of σ₂ and σ₄ to closely spaced DNA elements is facilitated by the nascent-RNA mediated displacement of σ region 3.2 and σ₄ that occurs during early elongation. In contrast, binding of σ₂ and σ₄ to closely spaced DNA elements during initiation would presumably require σ region 3.2 and σ₄ to be displaced from core via a nascent RNA-independent mechanism. Although certain promoters (so-called extended −10 promoters) can be recognized by RNAP independently of the σ₄-β flap interaction (Kuznedelov et al., 2002, Gregory et al., 2004, Hinton et al., 2005), we suspect that transcription initiation events that require the displacement of σ region 3.2 from the RNA exit channel would be heavily disfavored because of the important role σ region 3.2 plays during transcription initiation. Specifically, σ region 3.2, by virtue of its position within the RNA-exit channel, forms a loop that approaches the active center of RNAP and promotes formation of the first phosphodiester bond of the nascent RNA (via interactions with the NTP in the i +1 site) (Murakami et al., 2002b, Kulbachinskiy & Mustaev, 2006). Therefore, displacement of σ region 3.2 from the RNA-exit channel prior to transcription initiation would likely inhibit de novo RNA synthesis. Thus, we refer to configurations of the holoenzyme accessible only when σ region 3.2 and σ₄ are displaced from core as “elongation-specific”, i.e. disfavored during transcription initiation but readily accessible during elongation when displacement of σ region 3.2 and σ₄ from core is facilitated by the growth of the nascent RNA. Our findings (coupled with prior work) suggest that the holoenzyme can adopt at least four such “elongation-specific” configurations during the λQ-engagement process. Current objectives are to establish whether “elongation-specific” configurations of the holoenzyme can be detected in the absence of a protein factor that stabilizes their formation and whether “elongation-specific” configurations of the holoenzyme can be utilized for gene regulation in contexts other than λP_R′.

EXPERIMENTAL PROCEDURES

Proteins

His-tagged versions of wild-type σ⁷⁰, σ⁷⁰ R588H, σ⁷⁰ R584A, and σ⁷⁰ A553D were purified as described after overproduction from plasmid pLHN12-His (Panaghie et al., 2000). Escherichia coli RNAP core enzyme was obtained from Epicentre, and holoenzyme was made by incubation with a five-fold excess of the appropriate σ. λQ protein was purified as described (Yarnell & Roberts, 1992). Josh Filter and Jeff Roberts provided EcoRIGln111.

A His-tagged version of NusA was purified from BL21-DE3 harboring plasmid pET15b-NusA. Cells were grown at 37 °C in LB (100 mL) containing carbenicillin (100 μg/ml) to an OD600 of 0.5. Production of NusA was induced by addition of 1mM IPTG followed by 3 hours of growth at 37°C. Cells were harvested by centrifugation (3500 g for 20 min at 4 °C). The cells were resuspended in 10 mL lysis buffer (50 mM NaH₂PO₄, 0.3 M NaCl, 10 mM Imidazole, pH 8.0) and sonicated. The cell debris was removed by centrifugation (8000 rpm for 30 min at 4 °C). The supernatant was mixed with a 1 mL suspension of Ni-Sepharose HP beads (GE HealthCare) in wash buffer (50 mM NaH₂PO₄, 0.3 M NaCl, 20 mM Imidazole, pH 8.0) by shaking for 1 hr at 4 °C. The slurry was transferred to a 20 mL chromatography column (Econo-pac chromatography column, Biorad). Buffer containing unbound proteins was removed by gravity flow and the beads were then washed with 5 mL of wash buffer. NusA was eluted from the Ni- Sepharose column with elution buffer (50 mM NaH₂PO₄, 100 mM Imidazole, 0.3 M NaCl, pH 8.0). Protein-containing fractions were dialyzed against storage buffer (20 mM Tris-HCl, pH 8.0, 0.2 M NaCl, 1 mM DTT, 50% glycerol).

Strains and plasmids

A list of plasmids is provided in Table 1 and a list of strains is provided in Table 2.

Table 1.

Plasmids used in this study.

Name	Description	Source
pLHN12His	pT7-rpoD-His6 (N-terminal His tag)	(Panaghie et al., 2000)
pLHN12His (R588H)	pT7-rpoD-R588H-His6 (N-terminal His tag)	(Nickels et al., 2002)
pLHN12His (R584A)	pT7-rpoD-R584A-His6 (N-terminal His tag)	(Gregory et al., 2005)
pLHN12His (A553D)	pT7-rpoD-A553D-His6 (N-terminal His tag)	(Nickels et al., 2002)
pET15b-NusA	pT7-nusA-His6 (N-terminal His tag); full-length NusA cloned on an NdeI-XhoI fragment into pET15b (Novagen)	P. Deighan, A. Hochschild
pBR λQ	placUV5-λQ	(Nickels et al., 2002)
pBR λQ Δ−35	placUV5 Δ−35-λQ	(Nickels et al., 2002)
pBR ΔQ	placUV5 but harbors no λQ gene	(Nickels et al., 2002)
pFW11 Tet λPR′	carries sequences from −109 to +232 of λPR′	(Nickels et al., 2002)
pFW11 Tet λPR′ G–5A	carries sequences from −109 to +232 of λPR′ with a mutation (G–5A) in the −35-like element	(Nickels et al., 2002)
pFW11 Tet λPR′ G–25C	carries sequences from −109 to +232 of λPR′ with a mutation (G–25C) in the QBE	(Nickels et al., 2002)
“0 bp” constructs
pBEN 13	pFW11 Tet λPR′ 0 bp; same as pBN1250 except for changes indicated in Figure 2	This work
pBEN 15	pFW11 Tet λPR′ 0 bp- TTAACA; G–5A mutation introduced into pBEN 13	This work
pPGD 1	pFW11 Tet λPR′ 0 bp- TTGAAA; C–3A mutation introduced into pBEN 13	This work
pBEN 129	pFW11 Tet λPR′ 0 bp- Δ pause; A+1G and T+4G mutations introduced into pBEN 13	This work
“1 bp” constructs
pBEN 17	pFW11 Tet λPR′ 1 bp; same as pBN1250 except for changes indicated in Figure 2	This work
pBEN 18	pFW11 Tet λPR′ 1 bp- TTAACA; G–6A mutation introduced into pBEN 17	This work
pPGD 3	pFW11 Tet λPR′ 1 bp- TTGAAA; C–4A mutation introduced into pBEN 17	This work
pBEN 131	pFW11 Tet λPR′ 1 bp- Δ pause; A+1G and T+4G mutations introduced into pBEN 17	This work
“2 bp” constructs
pBEN 19	pFW11 Tet λPR′ 2 bp; same as pBN1250 except for changes indicated in Figure 2	This work
pBEN 22	pFW11 Tet λPR′ 2 bp- TTAACA; G–6A mutation introduced into pBEN 19	This work
pPGD 5	pFW11 Tet λPR′ 2 bp- TTGAAA; C–4A mutation introduced into pBEN 19	This work
pBEN 133	pFW11 Tet λPR′ 2 bp- Δ pause; A+2G and T+6G mutations introduced into pBEN 19	This work
“3 bp” constructs
pBEN 23	pFW11 Tet λPR′ 3 bp; same as pBN1250 except for changes indicated in Figure 2	This work
pBEN 25	pFW11 Tet λPR′ 3 bp- TTAACA; G–6A mutation introduced into pBEN 23	This work
pPGD 7	pFW11 Tet λPR′ 3 bp- TTGAAA; C–4A mutation introduced into pBEN 23	This work
pBEN 136	pFW11 Tet λPR′ 3 bp- Δ pause; A+3G and T+7G mutations introduced into pBEN 23	This work

Table 2.

Strains used in this study.

Name	Description	Source
FW102	Escherichia coli host strain for promoter-lacZ fusions on single copy F′ episomes bearing either a tetracycline resistance gene (Tet)	(Whipple, 1998)
BN 162	FW102 rpoD-A553D-Kan	(Nickels et al., 2002)
BN 147	FW102 containing an F′ Tet bearing the wild-type λPR′-lacZ fusion	(Nickels et al., 2002)
“0 bp” strains
BN 1002	FW102 containing an F′ Tet bearing the λPR′ 0 bp-lacZ fusion; strain was generated using plasmid pBEN 13 [procedure as described in (Whipple, 1998)]	This work
BN 1003	FW102 containing an F′ Tet bearing the λPR′ 0 bp TTAACA-lacZ fusion; strain was generated using plasmid pBEN 15 [procedure as described in (Whipple, 1998)]	This work
PGD 1	FW102 containing an F′ Tet bearing the λPR′ 0 bp TTGAAA-lacZ fusion; strain was generated using plasmid p PGD 1 and [procedure as described in (Whipple, 1998)]	This work
BN 1047	FW102 containing an F′ Tet bearing the λPR′ 0 bp Δ pause-lacZ fusion; strain was generated using plasmid pBEN 129 [procedure as described in (Whipple, 1998)]	This work
AL 3	FW102 containing an F′ Tet bearing the λPR′ 0 bp-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1002	This work
AL 4	FW102 containing an F′ Tet bearing the λPR′ 0 bp TTAACA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1003	This work
PGD 1-1	FW102 containing an F′ Tet bearing the λPR′ 0 bp TTGAAA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with PGD 1	This work
“1 bp” strains
BN 1004	FW102 containing an F′ Tet bearing the λPR′ 1 bp-lacZ fusion; strain was generated using plasmid pBEN 17 [procedure as described in (Whipple, 1998)]	This work
BN 1005	FW102 containing an F′ Tet bearing the λPR′ 1 bp TTAACA-lacZ fusion; strain was generated using plasmid pBEN 18 [procedure as described in (Whipple, 1998)]	This work
PGD 3	FW102 containing an F′ Tet bearing the λPR′ 1 bp TTGAAA-lacZ fusion; strain was generated using plasmid pPGD 3 and [procedure as described in (Whipple, 1998)]	This work
BN 1048	FW102 containing an F′ Tet bearing the λPR′ 1 bp Δ pause-lacZ fusion; strain was generated using plasmid pBEN 131 [procedure as described in (Whipple, 1998)]	This work
AL 5	FW102 containing an F′ Tet bearing the λPR′ 1 bp-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1004	This work
AL 6	FW102 containing an F′ Tet bearing the λPR′ 1 bp TTAACA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1005	This work
PGD 3-3	FW102 containing an F′ Tet bearing the λPR′ 1 bp TTGAAA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with PGD 3	This work
“2 bp” strains
BN 1006	FW102 containing an F′ Tet bearing the λPR′ 2 bp-lacZ fusion; strain was generated using plasmid pBEN 19 [procedure as described in (Whipple, 1998)]	This work
BN 1007	FW102 containing an F′ Tet bearing the λPR′ 2 bp TTAACA-lacZ fusion; strain was generated using plasmid pBEN 22 [procedure as described in (Whipple, 1998)]	This work
PGD 5	FW102 containing an F′ Tet bearing the λPR′ 2 bp TTGAAA-lacZ fusion; strain was generated using plasmid pPGD 5 and [procedure as described in (Whipple, 1998)]	This work
BN 1049	FW102 containing an F′ Tet bearing the λPR′ 2 bp Δ pause-lacZ fusion; strain was generated using plasmid pBEN 133 [procedure as described in (Whipple, 1998)]	This work
AL 7	FW102 containing an F′ Tet bearing the λPR′ 2 bp-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1006	This work
AL 8	FW102 containing an F′ Tet bearing the λPR′ 2 bp TTAACA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1007	This work
PGD 5-5	FW102 containing an F′ Tet bearing the λPR′ 2 bp TTGAAA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with PGD 5	This work
“3 bp” strains
BN 1008	FW102 containing an F′ Tet bearing the λPR′ 3 bp-lacZ fusion; strain was generated using plasmid pBEN 23 [procedure as described in (Whipple, 1998)]	This work
BN 1009	FW102 containing an F′ Tet bearing the λPR′ 3 bp TTAACA-lacZ fusion; strain was generated using plasmid pBEN 25 [procedure as described in (Whipple, 1998)]	This work
PGD 7	FW102 containing an F′ Tet bearing the λPR′ 3 bp TTGAAA-lacZ fusion; strain was generated using plasmid pPGD 7 and [procedure as described in (Whipple, 1998)]	This work
BN 1050	FW102 containing an F′ Tet bearing the λPR′ 3 bp Δ pause-lacZ fusion; strain was generated using plasmid pBEN 136 [procedure as described in (Whipple, 1998)]	This work
AL 9	FW102 containing an F′ Tet bearing the λPR′ 3 bp-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1008	This work
AL 10	FW102 containing an F′ Tet bearing the λPR′ 3 bp TTAACA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with BN 1009	This work
PGD 7-7	FW102 containing an F′ Tet bearing the λPR′ 3 bp TTGAAA-lacZ fusion, rpoD-A553D-Kan strain was generated by mating BN 162 with PGD 7	This work

In vitro transcription

Linear templates were obtained by PCR from plasmids (see Table 1) bearing the indicated promoter. Each linear template carried sequence extending from ~100 bp upstream of the transcription start-site to ~240 bp downstream of the transcription start site; these sequences include the terminator t_R′.

To assay early elongation pausing (Figure 3) template DNA (5 nM) was incubated with RNAP (20 nM) in transcription buffer (20 mM Tris-HCl pH 8.0, 0.1 mM EDTA, 50 mM KCl, 100 mg/ml BSA, 10 mM DTT) plus an NTP mix containing 200 mM GTP, 200 mM CTP, 200 mM UTP, and 50 mM γ-³²P-ATP at 1.5 mCi/ml for 5 minutes at 37° to form open complexes. A single round of transcription was initiated by adding 4 mM MgCl₂ and 10mg/ml rifampicin. The final volume of each reaction was 120μl. At the indicated times after the addition of the MgCl₂/rifampicin mixture, 20μl of each reaction was removed and quenched by addition of 100μl of 1.2X stop solution (0.6M Tris-HCl [pH 8.0], 12 mM EDTA, 80 μg/ml tRNA). Samples were then extracted with phenol/chloroform (1:1), precipitated with ethanol, resuspended in 10 μl of loading buffer (95% [v/v] formamide, 20 mM EDTA, 0.05% [w/v] bromophenol blue, and 0.05% [w/v] xylene cyanol), and electrophoresed on 15% polyacrylamide sequencing gels. Bands were visualized by phosphorimager and the data analyzed by Imagequant.

To assay λQ-dependent antitermination (Figures 4 and 5) template DNA (5 nM) was incubated with RNAP (20 nM) in transcription buffer (20 mM Tris-HCl pH 8.0, 0.1 mM EDTA, 50 mM KCl, 100 mg/ml BSA) along with 150 nM NusA (or, in certain cases, 450 nM NusA, see below) plus an NTP mix for 5–10 minutes at 37° to form open complexes. For the reactions shown in Figures 4A and 5 the NTP mix contained 200 mM GTP, 200 mM CTP, 200 mM UTP, and 50 mM γ-³²P-ATP at 1.5 mCi/ml whereas for the reactions shown in Figure 4B the NTP mix contained 200 mM GTP, 200 mM CTP, 200 mM α-³²P-UTP at 1.5 mCi/ml and 50 mM ATP. (We used α-³²P-UTP for the reactions of Figure 4B because mutating the pause-inducing elements associated with the “0 bp” and “1 bp” templates changed the initiating NTP from an A to a G). One tenth volume λQ or λQ dilution buffer (10 mM Tris pH 7.5, 500 mg/ml BSA, 100 mM DTT, 10% glycerol, 50 mM potassium glutamate) was added, and after 30 seconds transcription was initiated by adding 4 mM MgCl and 10 mg/ml rifampicin. Reactions were allowed to proceed for 10–20 minutes then stopped by addition of 5 reaction volumes of 1.2X stop solution [0.6M Tris-HCl (pH 8.0), 12 mM EDTA, 80 μg/ml tRNA]. Samples were then extracted with phenol/chloroform (1:1), precipitated with ethanol, resuspended in 10 μl of loading buffer [95% (v/v) formamide, 20 mM EDTA, 0.05% (w/v) bromophenol blue, and 0.05% (w/v) xylene cyanol], and electrophoresed on 6% polyacrylamide sequencing gels. Bands were visualized by phosphorimager and the data analyzed by Imagequant.

The concentration of NusA in assays performed using the “0 bp” template derivatives shown in Figures 4B and 5 was 450 nM. Use of a 3-fold higher concentration of NusA enhanced λQ antitermination activity in reactions using the “0 bp” template ~2-fold (compare Figure 4A and 4B). In reactions performed using the other −10/−35 spacing derivative templates use of 150 nM NusA or 450 nM NusA did not significantly enhance λQ antitermination activity (data not shown).

Exonuclease III challenge assays

Assays were done essentially as described (Yarnell & Roberts, 1992). For reactions done using the −10/−35 spacing derivatives, open complexes were formed by incubating 20 nM RNAP and 2 nM DNA in transcription buffer (20 mM Tris-HCl pH 8.0, 0.1 mM EDTA, 50 mM KCl, 100 mg/ml BSA) containing 100 μM ATP, GTP and UTP for 10 minutes at 37°. Transcription was initiated by addition of 5 mM MgCl₂, and the reactions were incubated for 4 minutes to allow formation of stalled complexes. Next, NusA (150 nM), λQ (see legend to Figure 6 for concentrations) or λQ dilution buffer, EcoRIGln111 (25 nM) and salmon sperm DNA (50 μg/ml) were added. The total volume of each reaction was 150 μl. Finally, ExoIII (0.03 U/μl) was added, and after 1, 2, 4, 6, or 10 min, 25 μl of the reaction was removed, quenched as above, prepared as above, analyzed on a 7% sequencing gel, and bands were visualized by phosphorimager.

DNA templates for exonuclease challenge assays were generated by PCR using oligos 5′-ACACGTTAGCAGCATGATTGCCA-3′ (upstream) and 5′-GGAATTCATACGTCGAAGTGACCAACTAG-3′ (downstream) and the following plasmids as templates: pFW11 Tet λP_R′, pFW11 Tet λP_R′ G–5A, pFW11 Tet λP_R′ G–25C, pBEN 17 (1 bp), pBEN 18 (1 bp TTAACA), pBEN 19 (2 bp), pBEN 22 (2 bp TTAACA), pBEN 23 (3 bp) and pBEN 25 (3 bp TTAACA). The resultant PCR products contained promoter sequences extending from −86 to +89 (for λP_R′ templates), −86 to +88 (for 1 bp templates), −87 to +89 (for 2 bp templates) and −87 to +90 (for 3 bp templates); the downstream sequences also contained a binding site for the catalytically inactive EcoRIGln111, which blocked exonuclease III digestion from this end. The downstream oligo was ³²P-labelled with T4 polynucleotide kinase; thus, PCR products were 5′-end labeled only on the bottom strand (see Figure 6A).

For assays done using the λP_R′ templates, conditions were identical except for the concentrations of the NTPs (reactions contained 100 μM ApApC and 25 μM ATP, GTP and UTP) and NusA was not present.

β-galactosidase assays

For the experiment shown in Figure 7B, reporter strain cells carrying either a wild-type λP_R′ template or the −10/−35 spacing derivative templates were transformed with either plasmid pBR λQ, which directs the expression of high levels of λQ under the control of an Isopropyl-β-D-thiogalactoside (IPTG)-inducible promoter or plasmid pBR ΔQ, which encodes no functional λQ.

For the experiment shown in Figure 7C, reporter strain cells carrying the “0 bp”, “1 bp”, or “3 bp” template derivatives were transformed with either plasmid pBR λQ or plasmid pBR ΔQ while reporter strain cells carrying the “2 bp” were transformed with either plasmid pBR λQ Δ−35, which directs the expression of low levels of λQ under the control of an IPTG inducible promoter, or plasmid pBR ΔQ.

Individual transformants were selected and grown in LB supplemented with carbenicillin (100 mg/ml) and tetracycline (10 mg/ml). When present, IPTG was added to the indicated concentrations. β-galactosidase assays were performed as described (Thibodeau et al., 2004) using microtiter plates and a microtiter plate reader. Miller Units were calculated as described (Thibodeau et al., 2004).