The promoter spacer influences transcription initiation via σ⁷⁰ region 1.1 of Escherichia coli RNA polymerase

Abstract

Transcription initiation is a dynamic process in which RNA polymerase (RNAP) and promoter DNA act as partners, changing in response to one another, to produce a polymerase/promoter open complex (RPo) competent for transcription. In Escherichia coli RNAP, region 1.1, the N-terminal 100 residues of σ⁷⁰, is thought to occupy the channel that will hold the DNA downstream of the transcription start site; thus, region 1.1 must move from this channel as RPo is formed. Previous work has also shown that region 1.1 can modulate RPo formation depending on the promoter. For some promoters region 1.1 stimulates the formation of open complexes; at the P_minor promoter, region 1.1 inhibits this formation. We demonstrate here that the AT-rich P_minor spacer sequence, rather than promoter recognition elements or downstream DNA, determines the effect of region 1.1 on promoter activity. Using a P_minor derivative that contains good σ⁷⁰-dependent DNA elements, we find that the presence of a more GC-rich spacer or a spacer with the complement of the P_minor sequence results in a promoter that is no longer inhibited by region 1.1. Furthermore, the presence of the P_minor spacer, the GC-rich spacer, or the complement spacer results in different mobilities of promoter DNA during gel electrophoresis, suggesting that the spacer regions impart differing conformations or curvatures to the DNA. We speculate that the spacer can influence the trajectory or flexibility of DNA as it enters the RNAP channel and that region 1.1 acts as a “gatekeeper” to monitor channel entry.

Transcription initiation is a multistep process that requires both recognition of promoter DNA and structural isomerization of the RNA polymerase (RNAP)/promoter complex to form a machine competent for transcription (reviewed in refs. 1–4). This process must be flexible enough to initiate transcription at a variety of promoter sequences yet rigid enough to provide specificity. In bacteria, the σ subunit of RNAP holoenzyme is the primary factor that sets this specificity. Although bacteria can have multiple σ factors, the primary σ, such as Escherichia coli σ⁷⁰, is responsible for the expression of housekeeping genes during exponential growth (5, 6). All σ factors share related regions 2, 3, and 4, but only primary σ proteins have a related, negatively-charged N-terminal portion, region 1.1 (6).

Transcription initiation begins with the initial binding of RNAP to dsDNA elements to form the polymerase/promoter closed complex (RPc) (7–9) (reviewed in ref. 1) (Fig. 1A). In RPc, polymerase interacts with a fully ds promoter (P). Promoter recognition can arise from interactions between the C-terminal domains (CTDs) of the α-subunits (α-CTDs) and ds promoter sequences between −40 and −60 (UP elements), between σ⁷⁰ region 4 and a −35 element, between σ⁷⁰ region 3 and sequences at −15, −14 (the extended −10 motif), and between σ⁷⁰ region 2.4 (a portion of region 2) and sequences at −12/−11 (the 5′ end of the −10 element) (reviewed in ref. 1). The RPc, which is usually unstable and competitor sensitive, gives an abbreviated protection footprint that does not include DNA downstream of the transcription start site (7, 9, 10). Creation of the stable polymerase/promoter open complex (Rpo) requires bending and unwinding of the DNA (11) and major conformational changes (isomerization) of the polymerase (Fig. 1A) (12–14). The result of these changes generates a complex in which the promoter is unwound from −11 to around +3, and the protection footprint extends to around +25 (9, 11, 15–21). In addition, RPo is normally competitor resistant, although RPo at the very strong ribosomal promoters does not follow this rule (22, 23).

Fig. 1.

Process of transcription initiation, promoter sequences, and transcription with P_min, P_min7, and P_min11 with Eσ^fl and Eσ^Δ1.1 (A) Diagram depicting polymerase promoter contacts in RPc and RPo with core polymerase (β, β′, α₂, and ω) in purple, σ⁷⁰ regions 2–4 in white, σ⁷⁰ region 1.1 in green, and DNA in red. R is RNA polymerase; P is the promoter DNA. The transcriptional start site is designated +1. Interactions between the α-CTDs and the UP element(s), σ⁷⁰ region 4 and the −35 element, σ⁷⁰ region 3 and the ⁻¹⁵TGn⁻¹³ element, and σ⁷⁰ region 2 and the −10 element are indicated. In RPc, the dsDNA has not yet entered the primary channel; full entry of DNA into the channel is blocked by σ⁷⁰ region 1.1. In RPo, σ⁷⁰ region 1.1 has moved, the DNA is bent and is unwound from −11 to +3, the template strand has descended into the active site of polymerase, and a portion of β,β′, called jaws, has secured the downstream DNA. (B) Sequences of P_min, P_min derivatives, and P_lacUV5-Mut. Consensus sequences for the σ⁷⁰-dependent −35, TGn, and −10 promoter elements are shown at the top. The EcoRI and SalI restriction sites used for plasmid constructions are boxed. σ⁷⁰ elements are shaded in gray, and base-pair substitutions in the P_min derivatives are in red. (C) Effect of promoter mutations on activity with Eσ^fl or Eσ^Δ1.1. Single-round transcription reactions were performed as described in Materials and Methods using Eσ^fl (blue) or Eσ^Δ1.1 (green). The amount of RNA from the indicated promoter (relative to the amount of RNA obtained at the 10-min time point with Eσ^{fl ×} 100) is plotted versus the length of the incubation of polymerase with the DNA (in min) before the addition of rNTPs and heparin.

Much work has been done to understand the conformational changes that occur as RPc transitions to RPo (reviewed in ref. 1). In addition, structures of σ, core polymerase, and holoenzyme from thermophilic bacteria (4, 24–29) or portions of E. coli σ⁷⁰ (30, 31) have provided 3D scaffolds on which to model these steps. Kinetic analyses using the λ promoter P_R have revealed transcriptional intermediates in the pathway from RPc to RPo (refs. 12–14, 20, and 32 and references therein). Initially, the ds promoter DNA is thought to lie across the polymerase, making sequence-specific contacts with σ⁷⁰ and the α-CTDs. The interaction of the DNA with the downstream DNA channel (portions of β and β′) generates an early intermediate (I₁), which, like RPc, is unstable and competitor sensitive. The DNA then moves deeper into the DNA channel through extensive interactions with portions of β and β′, forming a competitor-resistant intermediate, I₂. Finally, the DNA around the +1 site begins to melt, and σ⁷⁰ region 2.3 contacts single-stranded (ss)DNA bases at positions −10 through −7 on the nontemplate strand. For some promoters, contract(s) between residues in σ⁷⁰ region 1.2 and ss bases at −5 and −6 occurs also (22, 33). The protein/ssDNA interactions stabilize the polymerase/promoter complex, allowing the template strand to descend into the active site of core and the dsDNA downstream to fully enter the downstream channel. RPo is achieved when portions of β and β′, designated the polymerase “jaws,” close onto the downstream DNA, securing the DNA within polymerase.

σ⁷⁰ region 1.1 does not contact DNA, but is thought to play a crucial role in the transition from RPc to RPo (34, 35). At some promoters (λP_R, P_tac, P_RNAI) region 1.1 is needed for efficient formation of the open complex (34, 35). However, at the P_minor promoter the rate of RPo formation is actually inhibited by region 1.1 (34). Although the structure of region 1.1 of Thermophilus maritima has been reported (59), the structure of σ⁷⁰ region 1.1 has yet to be determined, presumably because its flexibility has made crystallization difficult. However, FRET data modeled with structural analyses indicate that in holoenzyme, region 1.1 lies within the channel that will be occupied by the downstream DNA when RPo is formed (Fig. 1A) (36). Consequently, region 1.1 must move for the DNA to occupy the channel as it does in RPo. Kinetic data with λP_R suggests that a conformational rearrangement, occurring at the I₁ ↔ I₂ transition, is consistent with the movement of 1.1 out of the channel (12), and FRET analyses have suggested that region 1.1 moves to a portion of core called the β pincer tip (36). It has been proposed that region 1.1 could facilitate DNA entry into the channel by holding the jaws open for free incoming DNA and/or by supplying energy needed for closure of the jaws by 1.1 movement (12).

In this article, we have investigated which features of P_minor are responsible for inhibition by region 1.1. We find that changing the promoter recognition elements of P_minor do not change this effect. However, substituting the P_minor spacer with a more GC-rich spacer or the complement sequence of this spacer generates promoters that are equally active with polymerase lacking region 1.1, Eσ^Δ1.1 or polymerase with full-length σ⁷⁰ (Eσ^fl). The P_minor spacer is an AT-rich sequence with features that can affect DNA conformation, and we demonstrate that the presence of this spacer affects the mobility of P_minor DNA during gel electrophoresis. We speculate that it is the DNA conformation, perhaps the trajectory or bend, as set by the spacer region, that influences the effect of region 1.1 on the initiation process. Furthermore, our DNase I footprinting indicates that the stable, open complex at P_minor is similarly and fully protected to +27 whether region 1.1 is present or absent. Our data suggest that movement of region 1.1 is not an obligatory step for the closing of the β/β′ jaws onto the downstream promoter DNA.

Results

Effect of σ⁷⁰ Region 1.1 on Promoter Activity Is Not Determined by Promoter Class or Promoter Recognition Elements.

Although early work led to the idea that σ⁷⁰ region 1.1 is required for efficient open complex formation (35), this work was performed using only 2 promoters, λP_R and P_tac. Subsequent work showed that the effect of σ⁷⁰ region 1.1 varies depending on the specific promoter tested (34). For P_uvsX-sigma, region 1.1 has little effect on the rate of open complex formation. For another promoter, P_minor, region 1.1 significantly inhibits formation of RPo.

Deletion analyses indicate that a minimal P_minor promoter, P_min, which contains P_minor sequences from only −35 to +4 (Fig. 1B), is also inhibited by region 1.1 (Fig. 1C). Our previous work demonstrated that P_minor and P_min belong to a newly identified promoter class, − 35/TGn, that is characterized by a requirement for both an excellent −35 element and an extended −10 TGn sequence to compensate for a poor −10 element (1, 37, 38). To investigate whether the effect of region 1.1 with P_minor reflects a general characteristic of −35/TGn promoters, we compared the formation of active transcription complexes at P_min with that at P_min16, a derivative with perfect −35 and −10 elements, and at P_min7, a derivative with a good −35 and a perfect TGn and −10 element (Fig. 1B). In addition, we tested P_min derivatives with more subtle modifications: P_min2, which has a perfect −35 element, and P_min3, which has a poor extended −10, but an improved −10 element (Fig. 1B). We incubated Eσ^Δ1.1 or Eσ^fl with the DNA for various times. Ribonucleoside triphosphates (rNTPs) and the polyanion heparin were then added together, allowing a single round of transcription and providing a read-out of the amount of transcriptionally-competent complexes present at any given time.

The promoter derivatives P_min2, P_min7, and P_min16 have improved polymerase recognition sequences (Fig. 1B) and, as expected, yield greater promoter activity than P_min (ref. 37 and data not shown). However, each of these promoter variants is still inhibited by region 1.1 (Fig. 1C and Fig. S1A). Our transcription analyses also indicate that P_min3, which lacks the TGn element, is inhibited by region 1.1 (Fig. S1A). Thus, the effect of region 1.1 on forming a heparin-resistant complex at P_minor appears to be determined by promoter context, i.e., something other than the promoter recognition elements (−35 element, extended-10 sequence, and −10 element) themselves.

Effect of σ⁷⁰ Region 1.1 on Promoter Activity Is Not Determined by a Short Polymerase/Promoter Half-Life, a Transcription “Stutter” Start, or Sequences Downstream of the −10 Element.

Besides its unusual promoter class, P_minor has other peculiar properties. First, the open complex at P_minor has a half-life of only a few minutes whether it is made with Eσ^Δ1.1 or Eσ^fl (34). However, the P_min derivatives, P_min2, P_min7, and P_min16, which have improved binding elements, form heparin-resistant polymerase/promoter complexes that remain stable for at least 30 min in contrast to the polymerase/P_min complexes that decrease after a few minutes (Fig. 1C and Fig. S1A). Thus, the increased activity of Eσ^Δ1.1 with these promoters is not associated with a short complex half-life.

P_minor is also unusual in that it has a stutter start, in which ≈3 nontemplated nucleotides are incorporated at the 5′ end of its transcript (37). To examine whether this start was responsible for the inhibition of P_minor by region 1.1, we tested a P_minor derivative, P_min8, that contains a +1 A to C mutation (Fig. 1B). Although this change eliminates the stutter start at P_minor when using Eσ^fl (37) or Eσ^Δ1.1 (Fig. S1B), the P_min8 promoter is still inhibited by region 1.1 (Fig. S1A). Therefore, the greater activity of P_minor with Eσ^Δ1.1 is not determined by its unusual start.

Because region 1.1 is thought to lie within the DNA downstream channel of holoenzyme and must move out of this channel during the formation of open complex, we tested two other P_min derivatives, besides P_min8, with mutations downstream of the −10 element. First, we examined P_min/P_lacUV5-Mut (Fig. 1B). This is a promoter in which P_min sequences downstream of position −7 have been replaced with the corresponding sequences of P_lacUV5-Mut, a promoter that is more active in the presence of region 1.1 (Fig. S1A). Although the P_min/P_lacUV5-Mut mutation changes nearly every base pair downstream from position −6, it is still inhibited by region 1.1 (Fig. S1A). We also tested P_min11, which contains a −5 C to G mutation (Fig. 1B). A comparable mutation in the ribosomal promoter rrnB P1 has been shown to increase RPo half-life via interaction with σ⁷⁰ region 1.2 residue M102 and, in addition, the absence of region 1.1 increases the detection of this contact (22, 33). P_min11 is also inhibited by region 1.1 (Fig. 1C). We conclude that the downstream region is not responsible for the increased activity of P_min with Eσ^Δ1.1.

The RPo complex is characterized by a transcription bubble surrounding the start of transcription, stability upon heparin challenge, and a DNase I protection footprint extending to around +25 (11, 19, 32). Previous kinetic analyses using the strong lambda promoter P_R have suggested that the movement of region 1.1 is coupled to the late folding of the β/β′ jaws of core onto the downstream DNA and that this movement could provide the energy needed for jaw closure and stable RPo formation (12). Thus, as another way to assess the effect of region 1.1 on the downstream DNA and on RPo formation and jaw closure, we determined the DNase I protection patterns of heparin-resistant complexes with Eσ^fl and Eσ^Δ1.1 at P_min and at the more consensus P_min7 promoter. Previous KMnO⁴ footprinting analyses have shown that these complexes contain the expected transcription bubble (ref. 37 and data not shown). We find that DNase I footprints at P_min7 with either polymerase are quite similar and in both cases extend to +27, as expected for RPo (Fig. 2A). Likewise, P_min is also protected to +27 with either polymerase although the footprint at P_min is not as strong as at P_min7 (Fig. 2B). This finding is reasonable because P_min is a weaker promoter than P_min7 and open complexes at P_minor decay with time (ref. 34 and Fig. 1C). This analysis suggests that for these promoters, the polymerase jaws close sufficiently to protect the downstream DNA from DNase I cleavage, with or without the movement of region 1.1. Thus, we conclude that the movement of region 1.1 is not a prerequisite for jaw closure at all promoters.

Fig. 2.

Absence of σ⁷⁰ region 1.1 does not affect DNase I protection of P_min or P_min7 downstream DNA. 5′-³²P fragments, labeled on the nontemplate strand, containing P_min7 (A) or P_min (B) were incubated with polymerase and treated with DNase I, and the DNA bound with Eσ^fl or Eσ^Δ1.1 (as indicated) or the unbound DNA was obtained and run on a 7 M urea, 6% polyacrylamide, denaturing gel. Lane 1 of A and lanes 1 and 5 of B are G+A marker ladders; lane 3 of A and lane 4 of B represent alternate exposures of lane 5 A and lane 6 B, respectively, and are shown for better comparison of the protection patterns. Positions −49, +1, and +27 within the promoters and the −10 and −35 elements are indicated on the gels. To the left of each gel is the trace for unbound DNA (red), and bound DNA with Eσ^fl (blue) or Eσ^Δ1.1 (green).

The Spacer Sequence Affects Promoter Activity with Eσ^Δ1.1.

A comparison of various promoter sequences indicated that the AT richness of the spacer sequence correlates with the effect of region 1.1 on promoter activity (Fig. S2). Although the length of a promoter spacer dictates the distance between the promoter elements and is an important factor in promoter usage, the spacer is not known to make sequence-specific contact with RNAP (reviewed in refs. 1 and 4) and exactly how the spacer affects the interaction of RNAP with a promoter is not fully understood (39–41). The spacer in P_minor is the preferred length for a σ⁷⁰-dependent promoter, eliminating the possibility that simply its length makes P_minor unusual. To test whether the P_minor spacer sequence (⁻²⁹AGATTAAAGAAATA⁻¹⁶) affects inhibition by region 1.1, we replaced this sequence with either the corresponding GC-rich 14 bp of the P_lacUV5-Mut spacer to generate P_min/GC, or with the sequence complementary to the P_minor spacer to generate P_min/comp. Transcription assays indicated that both the GC and the comp spacer substitution essentially eliminate transcription from P_min with either Eσ^fl or Eσ^Δ1.1 after incubations of polymerase and the DNA for 1 min (Fig. 3), 10 min (Fig. S3A), or even 30 min (data not shown).

Fig. 3.

The P_minor spacer region affects the modulation of stable complex formation by Eσ^Δ1.1. Except for P_min/GC and P_min/comp, in which no RNA was detected with either Eσ^fl or Eσ^Δ1.1, the amount of RNA seen with Eσ^fl and a particular promoter was set to 1. Values and standard deviations for Eσ^Δ1.1 were determined from 3 or more independent single-round in vitro transcription reactions, performed as in Fig. 1C, except P_min, P_min/GC, P_min/comp (or P_min7 set) templates were prepared to produce transcripts of 3 different lengths, allowing the promoters to be assayed in the same reaction. The total template DNA (0.02 pmol) is the same as that used in Fig. 1C, with the polymerase/DNA ratio maintained at 10:1. P_RNAI is present on each template DNA and serves as an internal control; this promoter is more transcriptionally active with Eσ^fl than with Eσ^Δ1.1 (34). Polymerase was incubated with the template DNA for 1 min before the addition of rNTPs and heparin.

Previously, it has been shown that a spacer can influence promoter activity (38, 40–45). In particular, a GC-rich spacer can result in reduced promoter activity compared with that seen with an AT-rich spacer (39, 40). Spacers with runs of Ts (46) or As (47) have been shown to dramatically increase the overall activity of a promoter. Thus, the decrease in transcription observed with P_min/GC is not surprising. However, P_min/comp has the same base-pair composition as P_min, yet yields no detectable transcription, indicating that something other than the overall AT richness of the P_min spacer is determining P_min activity.

Although the GC and comp spacers had a dramatic effect on P_min activity, the transcriptional levels were too weak to assess the effect of region 1.1; therefore, we tested the spacer exchanges within P_min7 because this promoter is recognized well by both Eσ^fl and Eσ^Δ1.1. In this experiment, the DNA templates were prepared to produce transcripts of three different lengths, allowing the promoters to be assayed in the same reaction. As shown in Fig. 3 and Fig. S3, the relative amounts of P_min7 RNA with Eσ^fl and Eσ^Δ1.1 seen under these conditions are similar to those observed when P_min7 is the only template (Fig. 1C), i.e., the presence of region 1.1 lowers the amount of RNA from P_min7 ≈2-fold. However, the presence of region 1.1 has no significant effect on the amount of RNA from the P_min/GC or P_min/comp promoters (Fig. 3 and Fig. S3). As expected (34), the level of P_RNAI RNA is greater when using Eσ^fl than when using Eσ^Δ1.1 (Fig. 3). Taken together, these results indicate that the P_min spacer region determines the inhibition by region 1.1 and that this property is not merely a result of the spacer's AT richness.

Previous work has indicated that AT-rich sequences, such as that present in the P_minor spacer, are apt to unwind more easily or be more flexible and easily distorted (48, 49). In particular, a T:A step within an A-tract has been shown to generate a flex point within the helix (49). KMnO₄ footprinting using P_minor did not reveal unpaired Ts within the spacer region when using either Eσ^fl or Eσ^Δ1.1 (ref. 37 and data not shown), suggesting that the spacer region itself is not grossly distorted. However, this analysis would not detect subtle or transient distortions in the spacer DNA. A-tracts can also result in an intrinsic bend in the DNA (50). However, a run of at least four contiguous A nucleotides is typically needed (50), and a T:A or G:C interruption, as here, has been shown to interrupt an A-tract bend (50, 51). To investigate the possibility that the P_minor sequence imparts a particular conformation or bend to the DNA, the mobilities of 51-bp fragments containing P_min, P_min/GC, P_min/comp, the P_min7 derivatives, or P_lacUV5-Mut (sequences shown in Fig. 1B) were compared on native, polyacrylamide gels. This analysis revealed that these promoter DNAs migrate differently despite their identical length (Fig. 4A). In particular, P_min and P_min/comp have the same base-pair composition, as do P_min7 and P_min7/comp, yet have the most divergent migration rates, i.e., P_min and P_min7 migrate more slowly than the GC derivatives, whereas P_min/comp and P_min7/comp migrate faster. These results suggest that these spacer regions can impart differing conformations or curvatures to the DNA fragments. The sequences examined by electrophoresis were also analyzed in silico. The structures for P_min, P_min/GC, P_min/comp, and P_lacUV5-Mut DNAs were predicted by the “model it” (52) program and then aligned based on their identical sequence of GAATTC at the 5′ end (Fig. 4B). These models predict that the putative P_min and P_min/comp structures are significantly curved in the spacer region relative to P_lacUV5-Mut and P_min/GC, a result that is consistent with the native gel analysis. We conclude that the presence of the P_minor spacer results in a DNA promoter conformation that differs from that formed with either the GC or comp spacers.

Fig. 4.

P_minor spacer affects the conformation of P_minor DNA. (A) Mobilities of P_min derivatives with different spacer sequences and P_lacUV5-Mut (5′-³²P end-labeled 51-bp fragments) upon electrophoresis in a native, 12% polyacrylamide gel. (B) Predicted structures of P_min, P_min/GC, P_min/comp, and P_lacUV5-Mut DNA. All structures were aligned using the common sequence GAATTC at the 5′ end. In addition, P_min, P_min/GC, and P_min/comp have identical sequences except for their 14-bp spacer sequences (Fig. 1B).

Discussion

Region 1.1, the negatively-charged domain found at the N terminus of primary σ factors, such as σ⁷⁰, is known to serve several important roles. In free σ⁷⁰, the presence of this region prevents recognition of promoter DNA (53, 54). In holoenzyme, region 1.1 lies within a channel of core polymerase that will interact with dsDNA downstream of the transcription start site upon formation of the stable promoter/polymerase complex (36). Consequently, region 1.1 functions as a negatively-charged space keeper, which is replaced by the negatively-charged DNA. In addition, the interaction of region 1.1 with this channel increases the overall stability of the σ⁷⁰/core interaction (55). Finally, in holoenzyme, the presence of σ⁷⁰ region 1.1 modulates the rate of open complex formation, depending on the particular promoter (34, 35). At some promoters, σ⁷⁰ region 1.1 stimulates the open complex formation, but at P_minor RPo formation is inhibited by region 1.1.

The important finding here is that the sequence of the P_minor spacer can decrease the activity of a promoter in the presence of region 1.1. How can a spacer influence the role of region 1.1? In the transition to RPo, the DNA around the start of transcription must melt, region 1.1 moves, the DNA is bent sharply into the channel, and the polymerase jaws close, stabilizing the complex (Fig. 1A) (reviewed in refs. 1–4). An AT-rich spacer, such as that in P_minor, may provide extra flexibility needed as the downstream region bends and unwinds (56). In particular, a T:A step within an A-tract, like position −17 of P_minor, can generate a flex point within the helix (49). In addition or alternatively, the spacer may provide a trajectory for the DNA that is conducive for channel entry. This possibility is consistent with the native gel and predicted promoter structure, which support the idea that the P_minor spacer imparts a particular curvature to the DNA. Therefore, the effect of region 1.1 may be caused by flexibility of the spacer in the context of RNA polymerase and/or promoter curvature via an intrinsic bend; in either case, region 1.1 would be inhibitory if the DNA is ready to enter the channel, but must wait for region 1.1 to exit. We speculate that P_minor exemplifies this type of promoter. Because the inhibition by region 1.1 is overcome when the P_minor spacer is replaced with either the GC-rich spacer or the AT-rich P_minor spacer complement (in the context of the P_min7 promoter), we presume that the GC and complement spacers do not yield the same conformation as that of P_minor. It should be noted, though, that depending on the particular promoter, the presence of region 1.1 can change the rate of forming heparin-resistant complexes, the maximum amount of stable complexes that can be made, or both (Fig. 2, Fig. S1A, and Fig. S3B). Thus, further work is needed to determine what specific steps in initiation are affected by region 1.1.

It is interesting that the presence of the P_lacUV5-Mut spacer within P_min does not switch the sigma preference from σ^Δ1.1 to σ^fl even though the P_lacUV5-Mut promoter is preferred by Eσ^fl. Thus, at P_lacUV5-Mut, like λP_R, P_tac, and P_RNAI (34, 35), region 1.1 has a positive effect on the formation of stable polymerase/promoter complexes. This finding suggests that other properties of these promoters, not shared by P_min, are needed for region 1.1 to stimulate promoter activity. We conclude that region 1.1 may have a pronounced inhibitory or “gatekeeper” function at promoters like P_minor, which have weak recognition elements coupled with a trajectory/conformation favorable for channel entry. In this way, region 1.1 may help polymerase discriminate against nonpromoter DNA and disfavor promoter sequences that are specific for alternate sigma factors, which lack region 1.1 (57). It is also worth noting that, unlike the promoter sequence elements, the spacer structure may be responsive to conditions (temperature) (41), DNA modifications (methylation), or transcription factors (CRP, MerR) (40, 58), and thereby provide an opportunity to regulate gene expression by modulating the formation of RPo.

Materials and Methods

DNA.

According to convention, the transcriptional start site is designated +1; DNA downstream of +1 is positively numbered and corresponds to the RNA transcript; the DNA upstream of +1 is numbered in the negative, beginning with −1 (there is no 0 nucleotide). The P_lacUV5-Mut plasmid, pFW11-P₂, contains a P_lacUV5 derivative promoter (called P₂) inserted between the EcoRI–SalI site of pFW11-null as described (37). Construction of the P_min plasmid, pXBJ402, which contains P_minor sequences from −35 to +4 inserted between the EcoRI–SalI site of pFW11-null and the plasmids containing the P_min derivatives P_min2 (pIH4022), P_min3 (pIH4023), P_min7 (pIH4027), and P_min8 (pIH4028) have been described (37). The P_min11, P_min16, P_min/GC, P_min/comp, P_min7/GC, P_min7/comp, and P_{min/lacUV5-Mut} plasmids were constructed similarly. Linear templates for in vitro transcriptions, which were digested using the indicated restriction enzymes, and the 5′-³²P end-labeled, 156-bp fragments used for the DNase I footprinting were prepared as described (37). The 51-bp oligomers, containing P_min or P_min derivatives (positions −41 to +10) or P_lacUV5-Mut (positions −44 to +7) (Fig. 1B) were synthesized by Operon Biotechnologies, labeled at the 5′ end of the nontemplate strand using [γ-³²P]ATP and T4 polynucleotide kinase, and then annealed to obtain the ds fragments used for native gel analysis.

In Vitro Transcription and DNase I Footprinting.

For single-round in vitro transcriptions, RNA polymerase was first reconstituted by incubating 0.2 pmol of core (Epicentre) and 0.5 pmol of either σ^fl or σ^Δ1.1, purified as described (34), in a 1.95 μL solution containing 27 mM Tris·Cl (pH 7.9), 54 mM Tris·acetate (pH 7.9), 52 mM NaCl, 40% (vol/vol) glycerol, 0.9 mM EDTA, 0.007% Triton X-100, 0.24 mM DTT, 154 mM potassium glutamate, 4.1 mM magnesium acetate, and 103 μg of BSA/μL for 15 min at 37 °C. A 2.05 μL solution containing 0.02 pmol of linearized DNA template and 22 mM Tris·Cl (pH 7.9), 43 mM Tris·acetate (pH 7.9), 71 mM NaCl, 3.4% (vol/vol) glycerol, 0.5 mM EDTA, 0.15 mM DTT, 220 mM potassium glutamate, 5.8 mM magnesium acetate, 146 μg of BSA/μl, and 0.34 mM 2-mercaptoethanol was then added, and the solution was incubated for the indicated times at 37 °C. Transcription was initiated by adding 1 μL of a solution containing 1 mM each of GTP, CTP, ATP and 50 μM [α-³²P]UTP (6 × 10⁴ dpm/pmol) and 0.5 mg/mL heparin. After 8 min at 37 °C, reactions were stopped by the addition of 15 μL of gel loading solution (94% deionized formamide, 9.4 mM EDTA, 0.1% bromophenol blue, 0.1% xylene cyanol FF) and heating at 95 °C for 2 min before electrophoresis on 6% polyacrylamide, 7 M urea, denaturing gels run in 1 × Tris-borate- EDTA. For the reactions in Fig. 1C and Fig. S1A, templates were digested with BglI. For the reactions in Fig. 3, the P_min and P_min7 templates were digested with BglI, resulting in 220-nt runoff transcripts, whereas the P_min/GC and P_min7/GC DNAs were digested with Bsu36I, and P_min/comp and P_min7/comp were digested with BaeI, yielding transcripts of 290 and 282 nt, respectively. The 3 transcript lengths allowed promoters with variant spacers to be assayed in the same reaction.

DNase I footprinting reactions were assembled as described for in vitro transcription assays except that the reaction also contained 2.25 mM CaCl₂. The labeled templates used in DNase I footprinting were 156-bp PCR products containing either the P_min or P_min7 promoter. PCR was performed with either pXBJ402 (P_min) or pIH4027 (P_min7) template, Pfu polymerase (Stratagene), and primers chosen to produce a fragment from position −99 to position +57, relative to the transcriptional +1 start. Primers were 5′ end-labeled with [γ-³²P]ATP using T4 polynucleotide kinase (New England Biolabs) before PCR. Each reaction contained labeled primer that annealed to one strand, and unlabeled primer that annealed to the other strand. The [γ-³²P]-labeled PCR product was purified by gel electrophoresis. For the DNase I reactions, DNA and reconstituted RNAP were incubated at 37 °C for 10 min. Protein–DNA complexes were challenged with heparin for 30 s and then treated with DNase I (0.5 unit) for 30 s at 37 °C. Loading-stop solution [28 μL of 192 mM ammonium acetate, 32 mM EDTA, 0.14% (wt/vol) SDS, and 0.036 mg/mL calf thymus DNA] was added, and the mixture was immediately loaded onto 4% native-polyacrylamide gels. Retarded protein–DNA complexes, identified after autoradiography, were cut out of the gel, embedded in a 1% (wt/vol) agarose gel, electro-eluted onto NA45 membranes (Schleicher & Schuell), and eluted off the membranes by incubating in a solution of 10 mM Tris·HCl (pH 8), 1 mM EDTA, and 1 M NaCl for 30 min at 65 °C. The DNA, obtained after extraction with phenol and precipitation in ethanol, was then run on the denaturing gels.

After autoradiography, films were scanned by using a Powerlook 100XL densitometer and various species were quantified by using Quantity One software from Bio-Rad.

Native PAGE.

Annealed promoter fragments were diluted in 1 × T4 ligase buffer (NEBL) to 0.2 mM. 5 × loading dye (40% glycerol, 0.1% bromophenol blue, 0.1% xylene cyanol FF) was added, and the samples were loaded onto a 12% native gel (acrylamide/bis-acrylamide of 29:1). Electrophoresis was carried out in 1 × TAE, at 4 °C at 200 V for 26 h.

Modeling of DNA Fragments in Silico.

The 51-bp sequence of each DNA promoter fragment assessed by native-PAGE was copied into the “model.it” web server (http://hydra.icgeb.trieste.it/dna/model_it.html) (52). Parameters were set for “Electrophoresis (dinucleotide).” The resulting structure predictions were downloaded in pdb format and aligned in MacPyMOL. Each DNA sequence shares the common sequence GAATTC at the 5′ end, such that all structures are identical and perfectly align in this region. P_min, P_min/GC, and P_min/comp have identical sequences except for their 14-bp spacer sequences.