Escherichia coli RNA Polymerase Recognition of a σ⁷⁰-Dependent Promoter Requiring a −35 DNA Element and an Extended −10 TGn Motif

Abstract

Escherichia coli σ⁷⁰-dependent promoters have typically been characterized as either −10/−35 promoters, which have good matches to both the canonical −10 and the −35 sequences or as extended −10 promoters (TGn/−10 promoters), which have the TGn motif and an excellent match to the −10 consensus sequence. We report here an investigation of a promoter, P_minor, that has a nearly perfect match to the −35 sequence and has the TGn motif. However, P_minor contains an extremely poor σ⁷⁰ −10 element. We demonstrate that P_minor is active both in vivo and in vitro and that mutations in either the −35 or the TGn motif eliminate its activity. Mutation of the TGn motif can be compensated for by mutations that make the −10 element more canonical, thus converting the −35/TGn promoter to a −35/−10 promoter. Potassium permanganate footprinting on the nontemplate and template strands indicates that when polymerase is in a stable (open) complex with P_minor, the DNA is single stranded from positions −11 to +4. We also demonstrate that transcription from P_minor incorporates nontemplated ribonucleoside triphosphates at the 5′ end of the P_minor transcript, which results in an anomalous assignment for the start site when primer extension analysis is used. P_minor represents one of the few −35/TGn promoters that have been characterized and serves as a model for investigating functional differences between these promoters and the better-characterized −10/−35 and extended −10 promoters used by E. coli RNA polymerase.

Recognition and binding of DNA promoter elements by RNA polymerase set the start site for transcription initiation. In Escherichia coli, these elements are recognized by a σ factor when σ is present in RNA polymerase holoenzyme, (σ plus core [β, β′, α, α, and ω]) (15, 36). E. coli encodes several σ factors that can be part of the holoenzyme, which are used under various conditions of growth and stress (39). Each σ factor interacts with different DNA sequences, and thus the recognition and usage of a given promoter is dependent upon the σ present in the holoenzyme. The presence of specific σ factors allows bacteria to coordinate the expression of gene sets and is one of the major ways bacteria regulate expression in response to changing growth conditions.

The primary σ factor of E. coli, σ⁷⁰, is used during exponential growth and belongs to a large family of prokaryotic primary σ factors related to each other by sequence, structure, and function (15, 39). Primary σ factors have four regions of similarity. It is known that residues in region 2 recognize a −10 element (TATAAT) (37), residues in region 3 recognize an extended TGn −10 motif (positions −15 to −13) (1), and residues in region 4 recognize a −35 element (TTGACA) (4). However, not all three of these promoter elements need to be present for promoter function. E. coli σ⁷⁰-dependent promoters have typically been characterized as either −10/−35 promoters, which have good matches to both the canonical −10 and −35 sequences and do not require the TGn motif (32), or as extended −10 promoters (TGn/-10 promoters), which have the TGn motif and an excellent match to the −10 consensus sequence and do not require a −35 element (2, 23, 26).

In addition to the sequence elements themselves, the distance between them is important for promoter recognition. Because σ⁷⁰ regions 4 and 2 simultaneously contact the −35 and the −10 elements, polymerase structure dictates the distance between the elements in a −10/−35 promoter (3, 10, 38). The −35 and −10 elements are ideally separated by a spacer length of 17 bp. Although this spacer length may vary, transcription is affected by a change of even one base pair (35). Likewise, the distance between the −10 element and the transcription start site is determined by the polymerase structure. The transcriptional start site is typically located seven nucleotides downstream of the −10 element (−12 TATAAT −7), with a preference for A as the incoming nucleotide (24, 28).

The consensus sequences for σ⁷⁰-dependent promoters have been studied extensively and are well defined (16, 27, 34). Promoter elements for σ⁷⁰-dependent promoters are initially assigned based on sequences that match −10/−35 or extended −10 consensus sequences at the appropriate distance from the +1 transcription start. However, the bacteriophage T4 P_minor promoter (50), which was identified by its activity in vitro, is an example of a promoter that does not readily fit into either the −10/−35 or the extended −10 promoter categories. Examination of the P_minor promoter region failed to locate good matches to any of the typical σ⁷⁰ DNA elements at proper positions relative to the transcriptional start site, which had been determined by primer extension. Nonetheless, recognition of P_minor is specific for polymerase containing σ⁷⁰, since it is not recognized by polymerase containing the closely related stationary phase σ factor, σ³⁸ (50). In addition, P_minor is of interest because the formation of stable polymerase/P_minor complexes increases when σ⁷⁰ lacks the N-terminal 99 residues (region 1.1); other tested promoters have been either unaffected or negatively affected by the lack of σ⁷⁰ region 1.1 (50, 53).

To understand how P_minor is recognized and used by polymerase containing σ⁷⁰, we have investigated how specific mutations within the P_minor promoter region affect transcription from this promoter. Here we define the minimal P_minor promoter and show that it functions in vivo as well as in vitro. We demonstrate that transcription from P_minor incorporates nontemplated ribonucleoside triphosphates (NTPs) at the 5′ end of the P_minor transcript, which results in an anomalous assignment of the start site when primer extension analysis is used. The correct assignment of the start site suggests that P_minor has both a good −35 element and a TGn motif but has an extremely poor −10 element. Our mutational analysis indicates that both the −35 element and the TGn motif are required for efficient transcription and that these elements compensate for the poor −10 element. P_minor represents one of only a few characterized −35/TGn promoters and is useful for comparing the properties of this class of promoter with the well-characterized −10/−35 and TGn/-10 classes.

MATERIALS AND METHODS

Constructs and DNA templates.

pFW11-P_null, pFW11-P₁, and pFW11-P₂ are pACYC184 derivatives that have a EcoRI/SalI fragment without a promoter sequence (P_null), with a good promoter (P₁), or with a very good promoter (P₂) upstream of the 5′ end of the lacZ gene (52). These plasmids also contain the 3′ end of the lacI gene and the kanamycin resistance gene upstream of the promoter insert. Thus, homologous recombination will transfer the antibiotic resistance and the promoter sequence into an F′ plasmid that contains the entire lac operon. To construct the P_minor-lacZ transcriptional fusion plasmid pXBJ203, a PCR product was first obtained using pDKT90 (29) as a template, Pfu polymerase (Stratagene), and primers designed to create restriction sites. The upstream primer annealed to pDKT90 starting at position −63 relative to the start of P_minor and included a 5′ EcoRI site, and the downstream primer annealed to pDKT90 starting at position +4 and included a 5′ SalI site. After isolation and digestion of this product with EcoRI and SalI, the fragment was ligated with pFW11-null that had been previously digested with EcoRI and SalI. The resulting plasmid is pXBJ203. Constructs pXBJ302, pXBJ402, and pXBJ503, which include the P_minor region from positions −44 to +4, −35 to +4, or −29 to +4, respectively, were created by inserting synthesized oligomers that contained the desired region of P_minor with upstream EcoRI and downstream SalI sites into pFW11-P_null that had been previously digested with EcoRI and SalI. pIH4021(P_−33A), pIH4022 (P_−31C), pIH4023 (P_{−14A−13A−12T}), pIH4024 (P_{−14A−13A−12T−31C}), pIH4025 (P_−14A), pIH4026 (P_−16G), pIH4027 (P_{−12T−10T−7T}), pIH4028 (P_+1C), and pIH4029 (P_−11G), which include the P_minor region from the +4 to −35 positions with specific promoter mutations, were constructed in the same manner using the appropriate oligomers to clone the promoter region. DNA sequence analyses (performed by the CBR-DNA Sequencing Facility of the University of Maryland or the Facility for Biotechnology Resources of the Food and Drug Administration) confirmed the promoter sequences in these various constructs.

Using the procedure of Whipple (52), pFW11-P_null, pFW11-P₁, pFW11-P₂, pXBJ203, pXBJ302, pXBJ402, pXBJ503, and pIH4028 (P_+1C) promoter constructs were transferred to single copy F′ plasmids by homologous recombination. The recombinant F′ plasmids were then transferred to the streptomycin-resistant E. coli strain FW102 by conjugation.

In some cases, templates for in vitro transcriptions were prepared by digesting plasmids, which had been previously isolated and purified, with BglI. This digestion linearized the plasmids at position +209, relative to the +1 position of P₂. Linear templates, which were used in transcriptions to identify the P_minor +1 (see Fig. 2A), were 120-bp PCR products containing only the P_minor promoter. PCR was carried out with pIH4022, Pfu polymerase (Stratagene), and primers chosen to produce a fragment from position −99 to position +21 relative to the P_minor +1 start site.

FIG. 2.

The P_minor transcriptional start is identified. In vitro transcription reactions were assembled by adding 1.95 μl of a solution containing reconstituted polymerase (0.2 pmol core plus 0.5 pmol of σ⁷⁰) in protein buffer I to 0.02 pmol of linear DNA in 2.05 μl of DNA buffer I. (A) Transcription was initiated by adding 1 μl of NTP mix to the protein-DNA mix containing P_minor DNA. As indicated, each NTP was added at the following concentrations: 1 mM UTP and 0.25 mM each ATP and GTP. The specific activity of [γ-³²P]GTP or [γ-³²P]ATP (where indicated by the asterisk) was 7 × 10⁵ dpm/pmol. Assignments of labeled RNA products are shown. The black arrow signifies 5-nt products, consistent with the migration of the pppACN₃ marker (not shown) kindly provided by N. Nossal. (B) Denaturing acrylamide gel showing the products of primer extension assays next to a P_minor DNA sequencing ladder. The unlabeled P_minor, P_+1C, and P₂ transcripts were generated by multiple-round transcription assays, which were initiated by the addition of 1 μl of NTP mix II. (C) Single-round transcription was initiated by adding 1 μl of NTP mix I with heparin. Transcripts arising from the P_minor, P_+1C, and P₂ promoters are indicated.

The labeled templates used in KMnO₄ footprinting were 156-bp PCR products containing only the P_minor promoter. PCR was carried out with either pXBJ402 (P_minor) or pIH4023 (P_{−14A−13A−12A}) template, Pfu polymerase (Stratagene), and primers chosen to produce a fragment from position −99 to position +57, relative to the P_minor +1 start. Primers were 5′ end labeled with [γ-^32P]ATP using T4 polynucleotide kinase (New England Biolabs) prior to 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.

Buffers and proteins.

σ⁷⁰ with an N-terminal His₆ tag was purified from E. coli BL21(DE3)/pLysE (46) cultures containing the rpoD plasmid (pETσ^fl [20, 53]), as previously described (50), by denaturation of inclusion bodies containing the protein, Ni²⁺ resin affinity chromatography under denaturing conditions, followed by a slow renaturation of the protein. E. coli RNA polymerase core was purchased from Epicenter Technologies. Protein buffer I contained 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 dithiothreitol, 154 mM potassium glutamate, 4.1 mM magnesium acetate, and 102.6 μg of bovine serum albumin/ml. DNA buffer I contained 21.9 mM Tris-Cl (pH 7.9), 43.4 mM Tris-acetate (pH 7.9), 71 mM NaCl, 3.4% (vol/vol) glycerol, 0.5 mM EDTA, 0.15 mM dithiothreitol, 219 mM potassium glutamate, 5.8 mM magnesium acetate, 146 μg of bovine serum albumin/ml, and 0.34 mM 2-mercaptoethanol. NTP mix I contained 1 mM each of ATP, GTP, and CTP and 25 μM [α-³²P]UTP (7 × 10⁵ dpm/pmol). NTP mix II contained 1 mM each ATP, GTP, CTP, and UTP. A set of NTP mixes was used for determining the P_minor +1. The NTPs present in each mix are indicated in Fig. 2.

β-Galactosidase assays.

The level of β-galactosidase activity in Miller units in FW102 lysates containing the single-copy F′ plasmids with the various promoter-lacZ constructs was determined as described previously (41) except that cells were grown in the presence of 30 μg of kanamycin and 100 μg of streptomycin/ml and no IPTG (isopropyl-β-d-thiogalactopyranoside) was added.

In vitro transcription assays.

Transcription reactions were assembled as indicated in the figure legends. Polymerase and DNA were incubated at 37°C for 10 min, NTPs were added, and reactions were incubated at 37°C for an additional 8 min. When indicated, single-round reactions were performed by including heparin (0.5 μl of 1 mg/ml) with the NTPs. Gel load solution (1× Tris-borate-EDTA, 7 M urea, 0.1% bromophenol blue, 0.1% xylene cyanol FF) was added at a volume three times that of the reaction aliquots, and the reactions were collected on ice. Each reaction solution was heated at 95°C for 2 min before electrophoresis on 6% polyacrylamide-7 M urea denaturing gels run in 1× Tris-borate-EDTA. After autoradiography, the films were scanned by using a Powerlook 2100XL densitometer and QuantityOne software from Bio-Rad, Inc.

Primer extension analyses.

Primer extension reactions were carried out as described previously (50). Primer IGH107 is complementary to P_minor, P_+1C, and P₂ transcripts from 155 to 171 nucleotides (nt) downstream of the SalI cut site, which is located from position +159 to position +175 relative to the P_minor +1A.

Potassium permanganate footprinting.

Reactions were assembled as indicated in the legend of Fig. 5. DNA-protein complexes were treated with potassium permanganate as described previously (29, 44). Potassium permanganate (0.5 μl of a 50 mM solution) was added to each 4-μl reaction mixture, and the mixture was incubated for 2 min at 37°C. After the addition of 5 μl of stop solution (0.69 M sodium acetate, 1 M 2-mercaptoethanol, 200 μg of salmon sperm DNA/ml), the DNA was precipitated with ethanol and dried. The pellets were resuspended in 100 μl of 1 M piperidine and incubated at 90°C for 30 min. The DNA was then precipitated with butanol, dried, and resuspended in gel loading solution. The DNA products were separated on denaturing gels as described above. As controls, G+A ladders were obtained using PCR product labeled on either the top or the bottom strand (31).

FIG. 5.

Potassium permanganate footprints of P_minor (A) and P_{−14A−13A−12T} (B). Reactions were assembled by adding 1.95 μl of a solution containing reconstituted polymerase (Eσ⁷⁰; 0.4 pmol core plus 1.0 pmol of σ⁷⁰) or core alone (0.4 pmol) in protein buffer I to 0.2 pmol of the indicated DNA in 2.05 μl of DNA buffer I. Eσ⁷⁰-dependent bands are marked by arrows and are numbered relative to the P_minor transcriptional +1 site. Nontemplate strand results for P_minor are consistent with the findings of Vuthoori et al. (50), but bands −6 and −4 are referred to as −2 and +1 in that study. Traces for the lanes are shown.

RESULTS

The 5′ end of P_minor RNA has nontemplated nucleotides.

Previously, P_minor was identified as a σ⁷⁰-dependent promoter active in vitro (50). Although this promoter is present within T4 DNA, located about 70 bp downstream of the T4 middle promoter P_uvsX, P_minor RNA has not been observed after T4 infection (17), suggesting that it is not active for T4 under typical growth conditions. In a previous study (50), the +1 transcription start for P_minor was identified as G (now position −3 in Fig. 1A). This assignment was based on primer extension experiments and the migration of the P_minor RNA on denaturing gels. However, typical σ⁷⁰ elements are not observed at the correct positions upstream of this assignment. Although an excellent −35 element (TTGAAA) is seen from 27 to 32 bp upstream of this start site, such a position would be highly unusual. Thus, we considered the possibility that the actual P_minor transcriptional start is 3 bp downstream of where the primer extension analyses indicated (+1A, Fig. 1). Primer extension acts as a measuring tape to determine the distance from a fixed point to the 5′ end of a transcript, which typically corresponds to the transcriptional +1 position. However, any extra nucleotides added to the transcript would result in a false measure of the transcriptional +1 position. This seemed plausible for the P_minor RNA since an A-rich sequence surrounds the start of P_minor transcription (GGAAAAT, positions −3 to +4 in Fig. 1). Such an A-rich sequence could facilitate slippage of the polymerase during initiation, resulting in the incorporation of extra A's at the start of the P_minor transcript. To determine whether the P_minor RNA starting nucleotide is a G, as assigned by primer extension, or an A, consistent with promoter element spacing, we carried out in vitro transcription with either [γ-³²P]GTP or [γ-³²P]ATP (Fig. 2A). During transcription only the initiating nucleotide retains the γ-phosphate; therefore, the labeled nucleotide must be incorporated at position +1 to result in detectable product. Reactions in which [γ-³²P]GTP was added yielded no observable product (Fig. 2A, lanes 1 and 2). In contrast, reactions that included [γ-³²P]ATP produced bands (Fig. 2A, lanes 3 to 5). This indicates that in vitro P_minor transcription begins with an A rather than a G.

FIG. 1.

Promoter constructs. Sequences between EcoRI (GAATTC) and SalI (GTCGAC) cloning sites (enclosed in boxes) in pFW11 are shown. (A) P_null (pFW11-null), P₁ (pFW11-P1), and P₂ (pFW11-P2) are as described elsewhere (52). A 67-bp P_minor fragment is from pDKT90 (29). Promoter elements (−35, TGn, and −10) and the +1 start site are noted in red above the sequence. Note that the +1A starts of P₁ and P₂ are located within the SalI site, whereas the P_minor +1 is 8 nt upstream. Arrows indicate 5′ end of the P_minor fragment before the EcoRI site in the P_-63, P_-44, P_-35, and P_-29 clones. (B) The P_-35 construct, used as wild-type P_minor promoter in the present study, is shown. Derivative promoters with the indicated changes are listed below P_minor. The dotted line indicates that the sequence is identical to P_minor.

The pattern of products seen in Fig. 2A suggests that stuttering occurs at P_minor. The addition of [γ-³²P]ATP alone (lane 3) resulted in a ladder of products, which is consistent with reiterative incorporation of NTPs (42). The addition of [γ-³²P]ATP and the next templated base, UTP, produced three distinct bands (lane 4). These bands migrate as four-, five-, and six-nucleotide (nt) products, which is consistent with a templated product (AAAU) plus products with one or two additional nucleotides. We assign these products as AAAAU and AAAAAU. The addition of [γ-³²P]ATP, UTP, and GTP in the reaction (lane 5) resulted again in three major bands consistent with the 6-nt templated product (AAAUGU) plus products with one or two additional nucleotides. We conclude that the added transcript length of P_minor is due to the incorporation of nontemplated A nucleotides. The sizes of the transcription products (≥4) also suggest that the first A (in the sequence 5′-GGAAAAUGU-3′) is not the +1 nt. Rather, transcription appears to begin at one of the other A's.

To further examine the P_minor transcriptional start, we changed the putative +1 nucleotide from an A to a C. C is the least favored nucleotide for beginning transcription, and changing the +1 nt to a C will often result in selection of the +2 nt as the transcriptional start (24, 28). We reasoned that, if the second A is the transcriptional start, then the +1 C mutation will disrupt the run of A's and prevent stuttering. To investigate the effect of this mutation, we used a primer that annealed 159 to 175 nt downstream of the assigned +1 start site of the P_minor transcript. As seen previously (50), after primer extension the product from P_minor RNA comigrated with the −3 position of the P_minor DNA sequence. This is because of the additional nontemplated nucleotides at the 5′ end of the P_minor RNA. In contrast, the primer extension product using the (P_+1C) RNA was slightly shorter. Similar results were observed in a transcription gel (Fig. 2C), in which the P_+1C transcript migrated slightly faster than P_minor RNA. We conclude that interrupting the run of A's at the P_minor start site results in a loss of stuttering, and thus we assign the P_minor transcription start as the +1A depicted in Fig. 1. As a control for these analyses, we used the P₂ transcript (Fig. 1 and see below). As expected, in the primer extension assay (Fig. 2B) the P₂ product migrated correctly, with the +9 position of the P_minor DNA. (Note in Fig. 1 that the P_minor DNA construct has an extra 8 bp relative to the P₂ DNA.) In addition, in the transcription gel (Fig. 2C) the P₂ RNA migrated as a smaller product, which is consistent with its expected size of 209 nt.

Defining the minimal P_minor promoter.

To precisely define the salient sequence features needed for the recognition of P_minor, a set of lacZ transcriptional fusions was designed based on the system of Whipple (52). A 67-bp fragment known to contain the P_minor promoter (−63 to +4) was first cloned into pFW11 and designated P₋₆₃ (Fig. 1). Similar constructs were then made (P₋₄₄, P₋₃₅, and P₋₂₉) with smaller promoter fragments, resulting in 5′ nested deletion constructs. pFW11-P₂ and pFW11-P₁, which contain the P_lacUV5-derived −10/−35 promoters P₂ and P₁, respectively, represented the positive controls. In vivo, P₂ and P₁ produce high and moderate levels of transcription, respectively (52) (see also Fig. 3A). pFW11-P_null, which lacks a promoter, was used as a negative control. The lacZ transcriptional fusions were moved into F′ plasmids, so promoter activity could be measured in vivo by β-galactosidase activity. The positive and negative controls behaved as expected (Fig. 3A), yielding significant or negligible activity, respectively. Strains carrying either P₋₆₃ or P₋₄₄ were less active than P₁, whereas P₋₃₅ and its derivative P_+1C yielded even more activity than P₁. However, the removal of the next six base pairs, containing the sequence TTGAAA, significantly reduced P_minor activity (P₋₂₉). We conclude that P_minor is active in vivo and that sequences downstream of position −36 are sufficient for this activity. In addition, these results suggest that the sequence from −35 to −30 is required for significant P_minor activity in vivo.

FIG. 3.

Minimal P_minor promoter defined. (A) Graph showing the β-galactosidase assay activity (in Miller units) determined in vivo for each promoter. (B) A denaturing acrylamide gel showing the products of multiple-round in vitro transcription reactions is overlaid with a graph of the quantitation data. Transcription reactions were assembled as described in Fig. 2C. Multiple-round transcription was initiated by adding 1 μl of NTP mix I (without heparin). The amounts of RNA were determined by densitometry and are shown relative to P₂, which is set at 100. The values represent the average of three or more transcriptions.

In multiple-round in vitro transcription assays, we found that the relative activities of the promoters differed somewhat from that observed in vivo. However, deletion of the TTGAAA sequence (P₋₂₉, Fig. 3B) was again deleterious, in this case eliminating promoter activity. Thus, because P₋₃₅ contained the minimum functional P_minor promoter both in vivo and in vitro, this construct was designated and used as wild-type P_minor for the rest of the present study.

Positions −35 to −30 of P_minor, TTGAAA, define the P_minor −35 element.

To determine whether the TTGAAA sequence of P_minor (positions −35 to −30) functions as the −35 element, we investigated how specific mutations in this sequence affected transcription. Altering the −33G:C base pair within a σ⁷⁰ −35 element (TTGACA) has been shown to significantly reduce transcription from promoters because the C moiety is directly contacted by σ⁷⁰ region 4 residues (22). The −33A mutation eliminated the production of P_minor RNA (Fig. 4), whereas the −31C mutation, which makes the −35 element a perfect match to a consensus, dramatically increased transcription (Fig. 4). These results indicate that the TTGAAA sequence of P_minor functions as the −35 promoter element and confirm that this sequence is required for P_minor activity. In addition, other work has shown that the anti-sigma factor, AsiA (19), which blocks σ⁷⁰ region 4 recognition of −35 elements, inhibits transcription from P_minor (20), further confirming the requirement for a −35 recognition element at P_minor.

FIG. 4.

TTGAAA functions as −35 element and is required for P_minor transcription. A denaturing acrylamide gel showing the products of single-round in vitro transcription reactions is overlaid with a graph of the quantitation data. Transcription reactions were assembled and carried out as described in Fig. 2C. The amounts of RNA were determined by densitometry and are shown relative to P_minor, which is set at 100. The values represent the average of three or more transcriptions.

The transcription bubble at P_minor ranges from position −11 to position +4.

Identification of the P_minor −35 element and +1A suggested that GAAAAC should function as the −10 element based on location. However, such a sequence deviates substantially from the canonical σ⁷⁰ −10 sequence, TATAAT (positions −12 to −7) (26). Previous work has shown that the nontemplate strand −11A is crucial for DNA melting and that transcription bubbles typically extend from this −11A to position +3 (11, 25, 30). Thus, KMnO₄ footprinting, which reveals single-stranded thymines, is a reliable way to identify the −10 element because it can identify the single-stranded T on the template strand opposite the nontemplate −11A.

We performed KMnO₄ experiments with DNA containing P_minor and either Eσ⁷⁰ or core alone (Fig. 5). When the nontemplate strand was 5′ end labeled (Fig. 5, top left), reactive bands occurred at positions +4, −4, and −6. A band was also visible at position +6 but was σ⁷⁰ independent. Footprinting using labeled template strand produced eight bands representing every thymine from positions −11 to +3 (Fig. 5, top right). Therefore, in P_minor, the transcription bubble extends from −11A to about position +4, as is typical (5, 7, 44), and the polymerase-DNA interaction within the −10 element appears to be normal as judged by this assay. We conclude that GAAAAC (positions −12 to −7) is the functional −10 element, a finding consistent with data discussed above assigning the +1A as the transcriptional start site.

The TGn motif of P_minor compensates for its noncanonical −10 element.

Although the P_minor −10 element (GAAAAC) deviates substantially from the consensus −10 sequence (TATAAT), P_minor has a TGn motif located just upstream at positions −15 to −13. To address whether the TGn was important for P_minor activity, we made a set of constructs to examine the roles of the TGn motif and the −10 element (Fig. 1 and 6). When the −14G was changed to −14A, transcription was significantly reduced, a finding consistent with the idea that the TGn motif is required for P_minor function. In contrast, the multiple mutation, −14A−13A−12T, had a positive effect on transcription. In this construct, the TGn motif was changed to TAn, but the −12G was also changed to T. Improving the −10 element allowed transcription in the absence of the TGn motif. These results suggest that the TGn motif of P_minor compensates for the poor −10 consensus.

FIG. 6.

P_minor requires TGn motif to compensate for a weak −10 element. A denaturing acrylamide gel showing the products of single-round in vitro transcription reactions is overlaid with a graph of the quantitation data. Transcription reactions were assembled and carried out as described in Fig. 2C. The amounts of RNA transcript were determined by densitometry and are shown relative to P_minor, which is set at 100. The values represent the average of three or more transcriptions.

Another consequence of the −14A−13A−12T mutation is that it creates a perfect −10 consensus sequence (TATAAT) at positions −17 to −13 relative to the P_minor start. The presence of this sequence could create a situation in which this perfect −10 sequence would be favored over the weaker −10 element of TAAAAC, present in this mutant. However, this is not the case since migration of the transcriptional product is unchanged (Fig. 6), and KMnO₄ footprints with this mutant show a transcription bubble in the same location as that seen with wild-type P_minor (Fig. 5, bottom). Thus, the functional −10 element is not changed in this mutant. Combining the −14A−13A−12T mutations with a −31C mutation created a promoter with a perfect −35 element and the improved −10 element. As expected, transcription from this promoter was significantly higher than from wild-type P_minor (data not shown) and roughly the same as that seen with the −31C mutation alone. Finally, when the −12, −10, and −7 nucleotides were all changed to T, creating a −10 element that perfectly matches consensus (P_{−12T−10T−7T}), transcription again was greatly improved (Fig. 6). As a control, we found that changing the −16A of P_minor, which is upstream of the TGn motif, to a G had no effect.

As discussed above, the nontemplate A at position −11, which is also highly conserved among σ⁷⁰-dependent promoters, is thought to lie at the transition between double-stranded DNA (upstream and including position −12) and the open (single-stranded) DNA within the transcription bubble (from position −11 to position +3) (11, 25, 30). To investigate the importance of the −11A in P_minor, we changed the −11A to a G. This change was particularly deleterious, resulting in a significant decrease in transcription (Fig. 6). We conclude that the −11A is a crucial determinant for P_minor activity.

DISCUSSION

Within promoter DNA, specific sequences provide the recognition elements needed to define the start of transcription. E. coli σ⁷⁰-dependent promoters have three well-characterized recognition elements: −35 sequences, TGn, and −10 sequences. Previous work has shown that base determinants within these elements are recognized by specific contact with σ⁷⁰ residues. The σ⁷⁰ region 4.2 residues R584 and E585 contact the −31G and −33C, respectively, on the template (bottom) strand of the double-stranded −35 element (TTGACA) (4, 14, 45), whereas recognition of the TGn element is thought to arise by the interaction of region 3 residues E458 and H455 (1, 43) with the −14G:C base pair. For the −10 element, σ⁷⁰ residues recognize and interact both with double-stranded DNA and with the single-stranded base determinants formed at the transcription bubble. Region 2.4 residues T440 (45) and Q437 (51) are required for recognition of the −12 T:A base pair in double-stranded DNA, facilitating closed complex formation (8, 9, 12, 37, 54). Aromatic residues Y425, Y430, W433, and W434 in region 2.3 promote DNA strand separation beginning at −11 and interact with the nontemplate strand of the −10 element in the open transcription bubble (12, 21, 37, 40, 48, 49).

Despite the multiple contacts between σ⁷⁰ and various base determinants in the promoter DNA, recognition of all of these elements is not biologically optimal because maximum promoter binding works against the need to leave the promoter once transcription has been initiated (reference 13 and references therein). In addition, deviation from consensus allows transcription to be conditionally regulated by transacting factors, allowing the best level of expression for the environment rather than the highest level of expression. Recognition of just the −10 and −35 elements (−10/−35 promoters) or recognition of just the extended −10 motif TGnTATAAT (TGn/-10 promoters) is sufficient for excellent promoter activity (2, 26, 34). In fact, the absence of σ⁷⁰ region 4 is nonessential for transcription from an extended promoter, indicating that the total loss of the σ⁷⁰/−35 contacts can be tolerated if both the TGn and canonical −10 sequences are present (23).

Although −35/-10 and TGn/-10 promoters have been established as σ⁷⁰ promoter classes, previous work has suggested that perhaps any combination of the −35, TGn, and −10 modules might be acceptable for recognition (13, 33). A study of more than 500 promoters (34) revealed that promoters with poorer matches to the canonical −10 sequences were more likely to have the TGn motif than those with consensus −10 sequences. This finding led to the speculation that the TGn motif might be able to compensate for a less canonical −10 sequence. Recent work with the gapA P1 promoter of E. coli (47) has demonstrated that the −35/TGn promoter is indeed a viable promoter architecture. Thus, the P_minor promoter, described here, and the gapA P1 promoter represent the first well-characterized members of a −35/TGn promoter class. Both of these promoters have excellent −35 sequences and the TGn motif. However, both of these promoters lack the conserved T at position −12 within the −10 element. It is this contact in particular for which −14G seems to compensate at P_minor, as seen by the activity of P_−14A versus P_{−14A−13A−12T}. In addition, P_minor also has a C rather than a T at the highly conserved position −7. With both promoters, mutation of either the −35 or the TGn motif away from consensus essentially eliminates transcription, suggesting that at these promoters, the −35 and TGn elements are the primary elements for promoter recognition. Thus, it seems that having a minimal number of strong contacts is required, but these contacts can occur in any combination. Despite their deviations from the −10 consensus sequence, both gapA P1 and P_minor retain the A nucleotide at position −11, and this base determinant is indeed crucial for P_minor activity. Thus, the presence of the −11A, which lies at the upstream edge of the single-stranded transcription bubble in the stable polymerase-promoter complex may be important for promoter melting despite whatever promoter modules are used for recognition.

Promoter elements are frequently assigned by an inspection of sequences upstream of the identified +1 sequence. Because the −35/TGn class has not been previously appreciated, other members of this class may have escaped notice and instead have been assigned as −35/-10 promoters with less than optimal spacer distances. Indeed, gapA P1 was not originally identified as a −35/TGn promoter and was not included in the study by Mitchell et al. (34).

Previous work in our lab failed to identify the correct promoter elements of P_minor in part because it is a member of the −35/TGn class but also because P_minor RNA contains nontemplated NTPs at the 5′ end. Consequently, primer extension analyses were misleading, and the +1 of P_minor was incorrectly assigned (50). Although several promoters have been identified that generate transcripts with nontemplated NTPs at the 5′ end due to polymerase slippage (6, 18, 42), in most of these cases, there is not a well-defined number of nontemplated bases incorporated. Rather, primer extension reveals a “stutter” stop as the primer is extended along RNAs of various lengths. However, with P_minor, primer extension gives a well-defined stop because of the addition of (mostly) three nontemplated bases. If this phenomenon occurs at other promoters, then the incorrect assignment of the transcriptional start may be more common than is currently realized. Indeed, several promoters align with a distance between the −7 and +1 that is less than the standard 6 bp (see the study by Mitchell et al. [34]). However, studies examining start site selection found that the 6-bp distance between the −10 element and the start of transcription was strongly preferred, such that decreasing the distance by 1 bp moved the start site downstream (24, 28). This makes sense because start site selection is dictated by polymerase structure, with the −10 nontemplate strand DNA held by region 2.3, whereas the template +1 is in the active site. Shorter distances between the transcriptional start site and −10 element might not be tolerated. However, polymerase slippage that specifically incorporates two to three nontemplated NTPs could be interpreted as a short distance between the −7 and +1 positions. Our study suggests that a reconsideration of the DNA elements of promoters with short distances between the −10 element and the transcriptional start is warranted.