Identification and analysis of ‘extended –10’ promoters in Escherichia coli

Abstract

We have compiled and aligned the DNA sequences of 554 promoter regions from Escherichia coli and analysed the alignment for sequence similarities. We have focused on the similarities and differences between promoters that either do or do not contain an extended –10 element. The distribution of –10 and –35 hexamer element sequences, the range of spacer lengths between these elements and the frequencies of occurrence of different nucleotides, dinucleotides and trinucleotides were investigated. Extended –10 promoters, which contain a 5′-TG-3′ element, tend to have longer spacer lengths than promoters that do not. They also tend to show fewer matches to the consensus –35 hexamer element and contain short runs of T residues in the spacer region. We have shown experimentally that the extended –10 5′-TG-3′ motif contributes to promoter activity at seven different promoters. The importance of the motif at different promoters is dependent on the sequence of other promoter elements.

INTRODUCTION

In bacteria, gene expression is dependent on transcription by a single RNA polymerase. For transcription, the bacterial RNA polymerase must be recruited to promoters and transcription initiation orchestrated. A large number of promoter contacts are made by RNA polymerase during initiation, and most sequence-specific contacts are made by the σ subunit (1–3). The C-terminal domain of the RNA polymerase α subunit can also make sequence-specific contacts with UP elements, which are usually located 40–60 bp upstream from the transcription start point (4,5). Transcription initiation of most genes in Escherichia coli is driven by RNA polymerase containing σ⁷⁰ rather than one of the six alternative σ factors. The most highly conserved promoter elements are the –10 and –35 hexamers, located 10 and 35 bp upstream from the transcription start point. For RNA polymerase containing σ⁷⁰, consensus sequences for these elements are 5′-TATAAT-3′ and 5′-TTGACA-3′, respectively, and the spacing between these elements is crucial for promoter function. The optimum spacer region length is 17 nucleotides. In general, deviation from either the consensus sequences or the optimum spacer length leads to a reduction in promoter activity (6–9).

Escherichia coli σ⁷⁰-dependent promoters also contain a less well conserved promoter element, the ‘extended –10’ element. This was originally identified in Bacillus subtilis where, as with other Gram-positive bacteria, it is more conserved than in E.coli (10–14). The extended –10 element is located one base upstream of the –10 hexamer, with the major determinant, 5′-TG-3′, positioned at –15/–14 with respect to the transcription start point. We reported previously that the extended –10 promoter element is present in ∼20% of all E.coli promoters (15). In addition to the 5′-TG-3′ determinant, in Gram-positive bacteria there is extended sequence conservation to give the consensus 5′-TRTG-3′, known as the –16 region (16,17).

Previous work from this laboratory has focused on transcription initiation at promoters containing the extended –10 element, using the galP1 promoter and several synthetic derivatives as models (15). In this work, we first analysed the sequences of 554 E.coli promoters and identified those promoters that contain an extended –10 5′-TG-3′ motif. We examined other conserved sequence elements, and investigated whether extended –10 promoters have different patterns of sequence conservation compared with non-extended –10 promoters. We also checked whether 5′-TRTG-3′ sequences were conserved. We then showed experimentally that, for several naturally occurring extended –10 promoters, the 5′-TRTG-3′ motif is an important determinant for promoter activity.

MATERIALS AND METHODS

Promoter compilation and analysis

The starting point was the compilation of 300 E.coli promoter sequences assembled by Lisser and Margalit (18), with the addition of 54 promoters from original papers cited by Ozoline et al. (19). The 1983 Hawley and McClure (6) and 1987 Harley and Reynolds (7) compilations were also consulted. The final major source of DNA sequences came from the inspection of references cited in the E.coli K-12 Linkage Map (20). We originally identified over 800 promoters, but only 554 of these had a mapped transcription start and an identified –10 region, thereby qualifying them for inclusion in our analysis (the compilations that we used are available at http://www.biosciences.bham.ac.uk/labs/minchin/mitchell2003/). The position and sequences of the –10 and –35 hexamer elements for each promoter were taken from the original literature, which may have assumed that the promoter is recognised by RNA polymerase containing σ⁷⁰, therefore some promoters in our list may be recognised by alternative σ factors. Our compilation focuses on the extended –10 promoter element and, therefore, promoters were aligned starting with the last base of the –10 hexamer, which is denoted as position –7. Since this study focuses on the extended –10 region, no attempt was made to align –10 and –35 element sequences by the insertion of gaps within the spacer, and analysis has been restricted to the bases downstream from position –28. The exception is the linked consensus analysis of both the –10 and –35 elements shown in Table 2 that uses only the 12 nucleotides of the two hexamer elements. Analysis of the –35 hexamer is based on 514 promoters (96 TG extended –10 promoters and 418 without the TG motif) for which the –35 element was identified in the original literature.

Table 2. Conservation of nucleotides within promoters at –10 and –35 elements.

The table shows an analysis of the degree of correspondence of promoter –10 and –35 hexamer sequences to the consensus sequences in the total collection of E.coli promoters (n = 514), or in the TG (n = 96) or non-TG (n = 418) subsets. The numbers are quoted as percentages and, due to rounding, may not exactly total 100.

Methods of analysis

The sequences and names of the promoters were initially aligned using Microsoft Excel. Using simple and basic functions within this program, the conservation of the –10 hexamer was calculated base by base. The same program was used to divide the promoters into subgroups, with and without the extended –10 5′-TG-3′ element, and then to probe the relationship and conservation of –35 and –10 regions and the distribution of different spacer lengths. For nucleotide and dinucleotide frequencies within the promoters, sequences were investigated from positions –7 to –28. Microsoft Access was used for the trinucleotide analysis.

Strains, promoters and plasmids

The E.coli K12 strain used throughout this work was DH5α (21). Promoter fragments were amplified from genomic DNA using PCR. Promoter fragments were between 100 and 175 bp in length and included ∼30 bp of DNA downstream of the transcription start point. The sequences, between positions –50 and +10, of all the promoters used are listed in Table 5. PCR fragments were cloned into the plasmid pGEM-T Easy. For promoter activity assays, the promoter fragments were subcloned using EcoRI and HindIII restriction sites into the low copy number, broad host range lac expression vector, pRW50, to give a promoter::lac operon fusion (22).

Table 5. Promoter region sequences and activity relative to the aroF promoter.

The table lists the sequence and activities of the 11 different TG promoters selected for further study. Activities were deduced from β-galactosidase measurements and are expressed as a fraction of the measured activity of the aroF promoter (1050 Miller units).

Mutagenesis and activity measurements for mutant promoters

Promoter mutants were constructed using the QuikChange™ site-directed mutagenesis kit (Stratagene). The 5′-TG-3′ motif was changed to TT, CG and CT in the envA, purEK, purMN and purFP1 promoters. In addition, all possible combinations of bases at positions –15 and –14 of the aroF, ompF and gyrA promoters were constructed. The activities of these different promoters were compared with the activities of derivatives of the synthetic KAB promoter carrying all possible base combinations at positions –15 and –14, previously constructed by Burr et al. (15). The bases at positions –17/–16 in the aroF promoter were also replaced by all possible combinations, and activities were compared with the corresponding set of KAB derivatives (also previously constructed by Burr et al.). All mutant promoters were checked by automated sequencing, provided by the University of Birmingham Genomics Laboratory. To measure promoter activities, DNA fragments carrying different promoters were subcloned into pRW50 and the activity of promoter::lac operon fusions was determined using β-galactosidase assays, as described by Miller (23). To do this, pRW50 plasmid derivatives were transformed into DH5α cells, and pre-cultures and assay cultures were grown in Lennox Broth (LB) plus tetracycline (35 µg/ml). Each assay was performed independently three times.

RESULTS AND DISCUSSION

Analysis of 554 promoter regions: –10 elements

Previous authors have searched for sequence patterns in different collections of bacterial promoters, e.g. O’Neill (24) and Galas et al. (25). In this study, we have focused on the analysis of promoters that contain the extended –10 motif, i.e. the sequence 5′-TG-3′ at positions –15/–14, located 1 bp upstream from the –10 hexamer (promoters containing this element are referred to here as TG promoters). We identified 106 such promoters. We have compared the sequence conservation of these 106 TG promoters with the sequence conservation of the 448 non-TG promoters (i.e. those lacking an extended –10 motif).

First, we considered promoter –10 hexamer sequences (Table 1). Overall, the conservation of the –10 hexamer according to our alignment is in good agreement with previously published compilations. The base T at position –7 is most highly conserved (90%), followed closely by the base A at position –11 (87%) and the base T at position –12 (79%). Bases at the other three positions are less well conserved. The TG promoters show less conservation within the –10 element compared with their non-TG counterparts. At positions –12 and –10, T occurs 7 and 8.5% less frequently in TG promoters compared with the non-TG counterparts. Thus, the 5′-TG-3′ motif may be compensating for a poor –10 hexamer, or TG promoters may have specific sequence requirements within their –10 element.

Table 1. Conservation of the –10 hexamer in E.coli promoters.

T	A	T	A	A	T	Promoter set or subset
79	87	50	50	54	90	All 554 promoters
74	87	42.5	51	52	88	106 TG promoters
81	87	51	50	55	90	448 non-TG promoters

The table shows the percentage of total promoters, or the percentage of promoters in the TG or non-TG subsets of promoters, that carry bases that correspond to the –10 hexamer consensus, 5′-TATAAT-3′, at different positions on the non-template strand.

Relationship between –10 and –35 elements

Analysis of the promoter set reveals that 37% of promoters have a –10 hexamer that has a 4/6 match to the consensus and the majority (95%) of promoters have a match of at least 3/6 (Table 2). Examination of the –35 hexamer in the set shows that most promoters have either a 4/6 or 3/6 match to the consensus (both occur at a frequency of 29%). The –35 hexamer is not as well conserved as the –10 hexamer, with 17% of promoters having a –35 hexamer with a match to the consensus of less than 3/6. Due to the flexibility in the length of the spacer between the –35 and –10 hexamers, and the fact that we are looking for conservation within a window of six residues, a match of at least two nucleotides would be expected by chance. Many promoters that have a poor match to the –35 hexamer appear to compensate by having a better match to the –10 hexamer consensus. A perfect match to the consensus within both the –10 and the –35 hexamer was not found. Thus, a typical E.coli promoter contains a 4/6 match to the consensus within both the –10 and –35 hexamers.

Sequence conservation within the –10 and –35 hexamers of the TG promoters differs from the non-TG promoters. TG promoters appear to be more tolerant of sequence variation within the –10 and –35 hexamers. For example, 26% of TG promoters have a –35 hexamer with a match of less than 3/6 to the consensus compared with 15% for non-TG promoters. Similarly, 38% of TG promoters have a –10 hexamer with three or less bases corresponding to the consensus (24% for non-TG promoters).

Data presented in Table 3 show that the most frequent spacer distance between promoter –10 and –35 elements is 17 bp, with 44% of all promoters having this spacer length. These data are in agreement with those obtained by Lisser and Margalit (18). Comparison of TG with non-TG promoters shows that a higher proportion of TG promoters have a longer spacer region. Thirty-nine percent of TG promoters have a spacer length of ≥18 bp (26% for non-TG promoters). This is in agreement with the study by O’Neill (24), which analysed sequence conservation as a function of spacer length. Thus, many extended –10 promoters have atypical spacing between their –10 and –35 elements, as well as greater degeneracy in their sequences.

Table 3. The distribution of spacer length at E.coli promoters.

Length of spacer region	% of promoters from Lisser and Margalit (18) compilation	% of promoters from this study
		All promoters	TG promoters	Non-TG promoters
<15	N/A	3.6	1.4	4
15	11.7	5.3	8.3	4.8
16	17.4	18.8	12.5	19.9
17	43.3	44.0	38.9	44.9
18	17.1	16.0	23.6	14.6
19	10.4	6.4	13.9	5.1
>19	N/A	5.8	1.4	6.6

The table shows an analysis of the spacer length between promoter –10 and –35 hexamer sequences in the total collection of E.coli promoters, or in the TG or non-TG subsets. The numbers are quoted as percentages. For comparison, data from the previous analysis made by Lisser and Margalit (18) are included.

Dinucleotide and trinucleotide analysis of E.coli promoters

Sequence conservation of selected dinucleotides from positions –17 to –7 in our compilation of promoters is shown in Table 4. The dinucleotide 5′-TG-3′ at positions –15/–14 is present in 19% of the promoters and occurs three times more often than predicted by chance (6.25%). It has been suggested that the E.coli extended –10 promoter element is analogous to the –16 promoter region found in Gram-positive bacteria, which has the consensus 5′-TRTG-3′ (16,17). Analysis of the TG promoters shows that 18% have the sequence 5′-TRTG-3′ (10% 5′-TATG-3′; 8% 5′-TGTG-3′), which is marginally more than for non-TG promoters (15%).

Table 4. Dinucleotide occurrence around the extended –10 element of E.coli promoters.

The table shows an analysis of the percentage frequency of occurrence of certain dinucleotides between positions –17 and –7 in the total collection of E.coli promoters, or in the TG or non-TG subsets. R = A or G; n = G, A, T or C.

Previous analyses have revealed the presence of short poly(A) or poly(T) tracts in the spacer region of E.coli promoters (26,27). To investigate whether these tracts are preferentially distributed at TG or non-TG promoters, we calculated the percentage of promoters carrying AAA or TTT tracts centred at positions from –8 to –27 at the two sets of promoters. The data presented in Figure 1 show that there is a clear preference, in the TG promoter set, for TTT tracts centered at positions –18, –20 and –25. We did not observe conservation of any other dinucleotide or trinucleotide sequences within positions –8 to –27.

Figure 1.

Trinucleotide occurrence in E.coli promoter regions. (a) The frequency of occurrence of the trinucleotide AAA on the non-template strand, centred at different locations from positions –7 to –28 in TG promoters (triangles) or non-TG promoters (squares). Frequencies are expressed on the ordinate as a percentage of total promoters with AAA at particular locations. (b) The frequency of occurrence of the trinucleotide TTT on the non- template strand, centred at different locations from positions –7 to –28 in TG promoters (triangles) or non-TG promoters (squares). Frequencies are expressed on the ordinate as a percentage of total promoters with TTT at particular locations.

The extended –10 motif is required for transcription from several E.coli promoters

To date, the 5′-TG-3′ motif has been studied in E.coli at a small number of promoters, namely galP1, cysG and phage λ P_RE (28–31), and is essential for the maximal activity of these promoters. Our sequence analysis shows that the extended –10 motif is found at many other E.coli promoters. Thus, for further study, we selected 11 different promoters, all with a 5′-TG-3′ motif, but with different –10, –35 and UP elements. These promoters showed a range of activities when assayed (Table 5). The contribution of the extended –10 motif to the activity of the seven most active promoters was investigated experimentally by changing the 5′-TG-3′ motif to 5′-TT-3′, 5′-CG-3′ or 5′-CT-3′. Table 6 lists the activities of the 21 mutant promoters. The data show that the 5′-TG-3′ motif is essential for optimal activity of all seven promoters. The purEK, purMN, purFP1, aroF and ompF promoters are highly dependent on the motif, whereas envA and gyrA are less dependent. It appears that promoters with poorer matches to the –10 and –35 consensus hexamers are more dependent on the 5′-TG-3′ motif.

Table 6. Activity of mutants of different TG promoters.

	Relative β-galactosidase activity				Core sequence
	Wild-type	TT	CG	CT	–10 hexamer	–35 hexamer
aroF	1.0	0.06	0.21	0.13	TATCGT	TTGAAA
ompF	1.0	0.07	0.09	0.07	AAAGAT	TAGCGA
gyrA	1.0	0.34	0.34	0.29	TATAAT	TTTACC
envA	1.0	0.19	0.30	0.15	TAAACT	TTGGGT
purEK	1.0	0.03	0.06	0.03	TATTCT	TTTTCC
purMN	1.0	0.07	0.17	0.10	TAGAAT	TCGCAA
purFP1	1.0	0.04	0.12	0.07	TAGAAT	TACGCA

The table lists the activities of derivatives of different TG promoters carrying TT, CG or CT at positions –15 and –14 instead of TG. Activities were deduced from β-galactosidase measurements and are expressed as a fraction of the measured activities of the starting TG promoter. These are 1050 Miller units for the aroF promoter, 750 units for the ompF promoter, 160 units for the gyrA promoter, 510 units for the envA promoter, 380 units for the purEK promoter, 310 units for the purMN promoter and 430 units for the purF P1 promoter. The corresponding –10 and –35 sequences on the non-template strand of each promoter are also listed.

The aroF, ompF and gyrA promoters have been studied in more detail. We constructed sets of 15 derivatives of each promoter, which contained all possible combinations of bases at positions –15 and –14. Figure 2 illustrates the activity of each set of mutant promoters. The activities are compared with the activities of a similar set of mutant derivatives of the semi-synthetic KAB promoter studied by Burr et al. (15). The data show that the different bases at positions –15 and –14 result in similar patterns of activity. This argues strongly that the 5′-TG-3′ motif plays a similar role at different promoters.

Figure 2.

Activities of promoters with mutations at positions –15 and –14. Each panel illustrates the activity of families of different promoters with different base combinations at positions –15 and –14, as indicated on the abcissae. The activities were deduced from β-galactosidase assays in cells containing plasmid pRW50 carrying fusions of the different promoters to the lacZ gene. Activities are expressed as fractions of the activity of the starting promoters with 5′-TG-3′ on the non-coding strain at positions –15 and –14. (a) Activities of the aroF promoter mutants. (b) Activities of the gyrA promoter mutants. (c) Activities of the ompF promoter mutants. (d) Activities of the synthetic KAB promoter (15) and different mutants.

The role of the dinucleotide at positions –17 and –16

In the final part of this study, we investigated the importance of the weakly conserved base sequence at positions –17 and –16. To do this, we used the aroF promoter, which was the strongest of the 11 selected TG promoters (Table 5). It contains –10, –35 and UP elements that correspond well to the consensus, two 5′-TTT-3′ tracts in the spacer region, and an extended –10 element, 5′-TGTG-3′. We constructed a set of 15 derivatives of the aroF promoter, which contained all possible combinations of bases at positions –17 and –16. Figure 3 illustrates the activity of each set of mutant promoters. The activities are compared with the activities of a similar set of mutant derivatives of the semi-synthetic KAB promoter studied by Burr et al. (15). The data show that the different bases at positions –17 and –16 result in similar patterns of activity, suggesting that these bases play a similar role at different promoters.

Figure 3.

Activities of promoters with mutations at positions –17 and –16. The panels illustrate the activity of families of different promoters with different base combinations at positions –17 and –16, as indicated on the abcissae. The activities were deduced from β-galactosidase assays in cells containing plasmid pRW50 carrying fusions of the promoters to the lacZ gene. Activities are expressed as fractions of the activity of the starting promoters. (a) Activities of the aroF promoter mutants (starting sequence: 5′-TGTG-3′ from positions –17 to –14 on non-template strand). (b) Activities of KAB promoter mutants (starting sequence: 5′-TTTG-3′ from positions –17 to –14 on non-template strand).

There is no direct evidence for contact between the bases at positions –17 and –16 of a promoter and σ. It is likely that a 5′-TR-3′ dinucleotide at this position affects the DNA structure and/or flexibility such that it aids the interaction of σ with the downstream promoter DNA. In Gram-positive bacteria, many promoters contain both a conserved –35 hexamer and an extended –10 motif; in such cases, the 5′-TG-3′ motif contributes to optimal activity (32,33). This also appears to be true in E.coli. For example, the aroF promoter contains both a 5′-TG-3′ motif and a –35 hexamer close to consensus, but is still dependent upon the extended –10 motif for optimal activity. Thus, the extended –10 motif may be important at all promoters where it is present, irrespective of the presence of other core promoter elements, but the level of dependence must be a function of these sequences. We conclude that the extended region immediately upstream of the –10 hexamer at many E.coli promoters works together with other promoter elements to drive optimal promoter activity.