Specific Recognition of the -10 Promoter Element by the Free RNA Polymerase σ Subunit*

Bacterial RNA polymerase holoenzyme relies on its σ subunit for promoter recognition and opening. In the holoenzyme, regions 2 and 4 of the σ subunit are positioned at an optimal distance to allow specific recognition of the -10 and -35 promoter elements, respectively. In free σ, the promoter binding regions are positioned closer to each other and are masked for interactions with the promoter, with σ region 1 playing a role in the masking. To analyze the DNA-binding properties of the free σ, we selected single-stranded DNA aptamers that are specific to primary σ subunits from several bacterial species, including Escherichia coli and Thermus aquaticus. The aptamers share a consensus motif, TGTAGAAT, that is similar to the extended -10 promoter. We demonstrate that recognition of this motif by σ region 2 occurs without major structural rearrangements of σ observed upon the holoenzyme formation and is not inhibited by σ regions 1 and 4. Thus, the complex process of the -10 element recognition by RNA polymerase holoenzyme can be reduced to a simple system consisting of an isolated σ subunit and a short aptamer oligonucleotide.

The subunit of bacterial RNA polymerase (RNAP) 4 holoenzyme plays a key role in promoter recognition and opening. Most promoters of housekeeping genes are recognized by a holoenzyme containing the primary subunit, called 70 in Escherichia coli or A in most other bacterial species. Primary subunits from different bacteria share four conserved regions each consisting of two or more subregions. Conserved regions 2.4, 3.0, and 4.2 of the subunit have been implicated in specific recognition of the Ϫ10 (consensus sequence TATAAT), the extended Ϫ10 (TG), and the Ϫ35 (TTGACA) promoter elements, respectively (1). Interactions of regions 2.3 and 2.4 with the nontemplate DNA strand at the Ϫ10 element have been shown to play a crucial role in RNAP-induced promoter melting and stabilization of the open promoter complex (1)(2)(3)(4)(5). Recognition of the Ϫ10 element can be modeled in a complex of holo-RNAP with short oligonucleotides corresponding to the nontemplate promoter strand and containing the Ϫ10 sequence (nontemplate oligonucleotides) (6 -9). However, despite playing a crucial role in promoter recognition by holo-RNAP, free is unable to specifically recognize either promoters or nontemplate oligonucleotides (1, 10 -12).
Crystal structures of holo-RNAPs from Thermus aquaticus and Thermus thermophilus revealed that contains three domains (2, 3, and 4, named after corresponding conserved regions) connected by flexible linkers (13,14). In holoenzyme, these domains are spread on the core enzyme surface, with 2 and 4 being positioned at an optimal distance relative to each other to allow interactions with the Ϫ10 and Ϫ35 promoter elements. The structure of the isolated individual domains of is almost identical to their structures in holo-RNAP (13)(14)(15)(16). However, the structure of the full-length subunit in a free state remains unknown.
Based on indirect biochemical and biophysical data, several mechanisms of inhibition of DNA-binding activity in free have been proposed. First, the N-terminal region of (region 1.1) was shown to inhibit DNA binding. Deletion of this region allows 70 to weakly bind double-stranded promoter DNA with some specificity (17,18) but does not allow the recognition of single-stranded nontemplate oligonucleotides by free (15,19). It was proposed that region 1.1 may inhibit DNA binding by (i) direct masking of region 4 (17,18), (ii) direct interactions with region 2 (20,21), or (iii) by an indirect allosteric mechanism (22).
Second, it was shown that free adopts a compact conformation in which DNA-recognition domains 2 and 4 are brought closer to each other than in the holoenzyme (23). The sub-optimal positioning of 2 and 4 is likely to interfere with simultaneous recognition of the Ϫ10 and Ϫ35 elements. Furthermore, the close proximity of 2 and 4 to each other and to other protein parts may directly mask DNA binding sites in free (24 -26). Finally, it was proposed that some local differences in the structure of the DNA binding domains in free and in holo-RNAP may also be associated with the differences in DNA binding properties of the protein (12,27). However, because the structure of free is unknown, the exact nature of the coreinduced rearrangements and the mechanism of inhibition of DNA binding by free are unclear.
Previously, we characterized ssDNA aptamers that contained the Ϫ10 promoter element and were recognized specifically by the free A subunit from T. aquaticus (sTaps, for sigma T. aquaticus aptamers) (26). It was shown that the binding of sTaps is accompanied by a conformational change in A and leads to an increase in the distance between domains 2 and 4. It was concluded that (i) the isolated A subunit has all determinants required for specific recognition of the Ϫ10 promoter element, and (ii) the increase in the 2-4 distance may be required for the unmasking of the DNA binding sites of A . However, it remained unknown whether free s from other bacterial species can also recognize promoter sequences and whether this requires similar conformational changes in the protein. In this work, we describe a novel class of aptamers that contain motifs similar to the Ϫ10 and TG promoter elements and are recognized efficiently by subunits from various bacterial species. We demonstrate that the aptamers interact with region 2 and that this interaction does not require major conformational changes of , which has been observed upon the holoenzyme formation (23) and sTap binding (26). The aptamer binding is also not inhibited by regions 1 and 4. Thus, specific DNA recognition by region 2 can occur without large structural changes of the subunit.

EXPERIMENTAL PROCEDURES
Proteins-Wild-type and mutant E. coli 70 , T. aquaticus, and Deinococcus radiodurans A were obtained as described (26,28,29). The fragment of the rpoD gene encoding for amino acids 1-565 of E. coli 70 was amplified from genomic DNA and cloned between NdeI and EcoRI sites of pET28. The protein was overexpressed in E. coli BL21(DE3) and purified as described (29).
Selection of Aptamers to E. coli 70 -The selection of aptamers was performed essentially as described previously (26). The library used for the selection was 5Ј-GGGAGCTCAGAATA-AACGCTCAA-32N-TTCGACATGAGGCCCGGATC, where N is a random nucleotide. The selection was performed in a buffer containing 20 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 , 240 mM NaCl, and 60 mM KCl. The amounts of ssDNA and the 70 subunit were varied from 3 nmol and 500 pmol, respectively, in the first round of selection, to 100 and 10 pmol in subsequent rounds. Selection was performed using N-terminally hexahistidine-tagged 70 . 70 and DNA were incubated for 30 min in 1 ml of the binding buffer, nickel-nitrilotriacetic acid-agarose (Qiagen) was added, unbound DNA was removed by extensive washing, and -DNA complexes were eluted with binding buffer containing 200 mM imidazole. The eluted DNA was amplified using primers corresponding to the fixed regions of the library (5Ј-GGGAGCTCAGAATAAACGCTCAA and BBB-5Ј-GATCCGGGCCTCATGTCGAA, where B is a biotin residue), DNA strands were separated by size on 10% denaturing PAGE, and the nonbiotinylated strand was purified and used for the next round of selection. After ten rounds, the enriched library was amplified with primers containing EcoRI and HindIII sites and cloned into the pUC19 plasmid. Individual aptamers were obtained by PCR or purchased from Syntol (Moscow).
Analysis of -Aptamer Interactions-Determination of equilibrium K d values for binding of aptamers to the subunit was done by the nitrocellulose binding method (30). Oligonucleotides were 5Ј-end-labeled with [␥-32 P]ATP (6000 Ci/mmol, PerkinElmer Life Sciences) and T4 polynucleotide kinase (New England Biolabs) and purified by PAGE. Each aptamer (0.1 nM) was incubated with a series of dilutions of the subunit or its fragments (from 1 nM to 3 M) in 50 l of binding buffer for 30 min at room temperature; the samples were filtered through 0.45-m nitrocellulose filters (HAWP, Millipore), and the filters were washed with 5 ml of the buffer and quantified with PhosphorImager (GE Healthcare). To calculate apparent equilibrium dissociation constants, the data were fit to a hyperbolic equation, where B is a percentage of DNA bound, B max is the maximum binding at infinite concentration of , and K d is the dissociation constant. The fitting was performed by nonlinear regression algorithm using GraFit (Erithacus Software). For each aptamercombination, K d measurements were independently repeated two to three times, and averages were calculated. The experimental variation among replicate measurements usually did not exceed 30% of the average value.
Cross-linking Experiments-Cross-linking was performed in 15 l of the binding buffer. The samples contained 10 nM of 5Ј-labeled sEcap1 or nontemplate oligonucleotide and 100 nM 70 . The nontemplate oligonucleotide was closely related to the nontemplate strand of a lacUV5 promoter: 5Ј-ATTGCG-TATAATGTGTGGA. The samples were incubated for 15 min at 25°C, irradiated for 10 min under a 254 nm UV lamp (4 watts, Spectroline), and separated on 5% SDS-PAGE. For the mapping of the cross-linking sites, the DNA-protein complexes were eluted with 400 l of 0.03% SDS, freeze-dried, and dissolved in 20 l of 50 mM HCl. 10-l aliquots were withdrawn, and 1 M CNBr was added to the remaining samples to 50 mM. The reaction was stopped after 5 min by adding an equal volume of stop buffer containing 2% SDS, 0.5 M Tris-HCl, pH 8.4, 100 mM ␤-mercaptoethanol, and 20% glycerol. The degradation products were separated on 13% SDS-PAGE.
Luminescence Resonance Energy Transfer Distance Measurements-The double-cysteine mutant of 70 was labeled with europium chelate and Cy5 and purified as described previously (23). Distance measurements were conducted at 25°C as described (23,26,31). 70 was taken at 25 nM. Oligonucleotides and the E. coli core RNAP were added to 100 and 60 nM, when present, respectively. The control oligonucleotide that did not bind 70 was an sTap1 aptamer described previously (26): 5Ј-GAGTGTATAATGGGAGCGGTATCGTTCGACATGAG. 70 Subunit of E. coli RNA Polymerase-As the primary target for aptamer selection, we chose the 70 subunit from E. coli, which has a well defined promoter specificity and has been used as a model to study the process of promoter recognition for many years (1). ssDNA aptamers were selected from a library of 75-nucleotide-long ssDNA containing a 32-nucleotide central region of random sequence surrounded by regions with fixed sequences. The initial ssDNA library showed poor affinity for 70 (apparent K d Ͼ 10 M). Control experiments demonstrated that free 70 also did not bind nontemplate oligonucleotides containing the Ϫ10 promoter element (K d Ͼ 10 M). After 10 rounds of selection, the affinity of the enriched library significantly increased (K d ϳ 85 nM). The library was cloned, and insert sequences of 30 individual clones were determined. The resulting sequences were named sEcaps (for sigma E. coli aptamer) (Fig. 1A). All the aptamers exhibited high affinity for 70 and bound it with K d values ranging from 30 to 100 nM. For several clones, it was possible to minimize the size of the aptamers by removing the whole fixed library regions without decreasing the aptamer affinity. One of the minimal aptamers, sEcap1, had even higher affinity to 70 (K d ϳ 20 nM) than the full-length variant and was chosen for further studies.

Structural Features of Aptamers Recognized by the
The majority of the aptamers share the octanucleotide sequence TGTAGAAT, which is located at different positions within the central region of the library (Fig. 1A). In this sequence, the underlined motif is similar to the bacterial primary Ϫ10 promoter element but contains G instead of T at the third position. The dinucleotide TG located to the left of the Ϫ10-like sequence matches the TG motif of the extended Ϫ10 promoters (see below). In all the aptamers, the conserved 8-nucleotide motif is preceded by G-rich regions, which can potentially form G-quadruplexes. The G-quartet structure is important for aptamer binding, because aptamer variants with substitutions in this region did not interact with . The binding of the aptamers also depended on the presence of potassium ions, which are known to stabilize G-quartets (data not shown). The proposed secondary structure of sEcap1 is shown on Fig.  1B. In the structure, the G-quadruplex is followed by a singlestranded region containing the conserved aptamer motif.
Although aptamer ligands are usually highly specific to their target proteins (32), we found that the aptamers selected in this work were recognized with equally high efficiencies by subunits from various bacterial species, including T. aquaticus and D. radiodurans (K d values for sEcap1 were 9.5 and 6.4 nM, respectively). This indicated that the aptamers interact with a highly conserved epitope of .
Aptamers Interact with 70 Region 2-The presence of the conserved Ϫ10-like element in sEcaps suggested that it is essential for aptamer binding and is likely recognized by conserved region 2, similarly to the Ϫ10 element in promoters. Indeed, a mutant variant of sEcap1 bearing an ATCTCG sequence instead of the Ϫ10-like motif did not bind 70 (data not shown). Adenine at the second position of the Ϫ10 element (Ϫ11A) is the most conserved promoter base and was previously shown to be critical for promoter recognition and opening (33). We found that a substitution of C for A at this position of the Ϫ10-like motif in sEcap1 completely abolished aptamer binding (K d Ͼ 3 M). Interestingly, an sEcap1 mutant bearing the perfect Ϫ10 element consensus (TATAAT) also exhibited poor binding to (K d Ͼ 1 M) indicating that the Ϫ10 element in sEcaps is recognized by free 70 with a somewhat different specificity than the Ϫ10 promoter element by holo-RNAP. At the same time, changes at other positions of the TAGAAT motif had less dramatic effect on aptamer recognition (K d for aptamer variants with point nucleotide changes at the first, fourth, and fifth positions of this motif ranged from 20 to 80 nM).
To localize the region(s) involved in interactions with the aptamer, we performed cross-linking of 70 with radioactively labeled sEcap1. Irradiation of the 70 -sEcap1 complex with UV light at 254 nm resulted in the appearance of a cross-linked DNA-protein complex that had lower mobility than free DNA on a denaturing gel ( Fig. 2A, lane 1). The cross-link was highly specific, because its efficiency was dramatically reduced in the case of the sEcap1 variant that contained the Ϫ11A to C mutation ( Fig. 2A, lane 2). For mapping of the cross-linking sites in the 70 subunit we used a method of limited chemical cleavage of modified protein at Met residues (34,35). In a control reaction, we performed cross-linking of 70 in a complex of E. coli holo-RNAP with a single-stranded nontemplate oligonucleotide containing the Ϫ10 element (Fig. 2B). The cross-linking site of this oligonucleotide was previously mapped between 70 Met residues 413 and 456 (9). The treatment of 70 , radioactively labeled with sEcap1, with BrCN in single-hit conditions resulted in the formation of a characteristic pattern of radiolabeled peptides corresponding to the C terminus of the protein (Fig. 2B, lane 3). This pattern was identical to that observed in the case of the control nontemplate oligonucleotide (lane 2). The shortest labeled peptide corresponded to cleavage at Met-413. Shorter peptides resulting from cleavage at and C-terminal to Met-456 were not detected on the autograph and, therefore, did not contain the site of modification (Fig. 2B, see also Ref. 9).  (Table 1). The first and the second of the fragments lacked conserved region 4.2 and regions 3 and 4, respectively. The third fragment lacked regions 3 and 4 and the N-terminal segment, including region 1.1 (amino acids 1-93) and the first 8 amino acids from region 1.2, and corresponded to structural domain 2, which is conserved in all 70 -family proteins (15,16) and whose structure has been solved previously (36). We found that the affinities of all three fragments to sEcap1 did not differ significantly from the affinity of the full-length 70 ( Table 1). All fragments recognized sEcap1 specifically, because the Ϫ11C mutation abolished aptamer binding (K d Ͼ 3 M). This indicates that the presence of regions 1, 3, and 4 in full-length 70 is neither required for nor inhibitory to aptamer binding.

The TG Element in Aptamers Is Recognized by Region 2.4-
The TG motif present in all the aptamers resembles the TG element of the extended Ϫ10 promoters but, in contrast to the promoters, is located immediately upstream of the Ϫ10-like sequence. To check whether the position of TG is essential for aptamer recognition, we studied an sEcap1 mutant, which contained an additional cytosine nucleotide inserted between the TG motif and the Ϫ10-like element ("TGC sEcap1"). This aptamer was recognized by 70 with almost as high affinity as wild-type sEcap1 (K d ϳ 70 nM, Table 1 and Fig. 3A). At the same time, an aptamer mutant in which the TG motif was substituted by CC ("CC sEcap1") did not bind (K d Ͼ 3 M, Fig. 3A). Therefore, the TG motif is essential for aptamer binding and can be specifically recognized by 70 when located at the same position as in the extended Ϫ10 promoters. The observed difference in the TG position in the aptamers and promoters likely reflects some differences in conformations of and/or DNA in the -sEcap complex and in the holoenzyme RNAP-promoter complex during transcription initiation.
Suppression genetics studies initially suggested that TG in the extended Ϫ10 promoters is recognized by region 3.0 of 70 (37). Later, region 2.4 was also shown to play a role in the recognition of this motif (38). We therefore used aptamers as model DNA substrates to analyze the roles of these regions in the TG recognition. We found that 70 fragment 1-448 lacking region 3.0 still bound sEcap1 with high affinity but did not bind the TG-less aptamer mutant (Table 1 and Fig. 3B, see also above). The 1-448 fragment also recognized the TGC sEcap1 variant containing C between TG and the Ϫ10 element with about the same affinity as full-length 70 (Table 1 and Fig. 3B). We therefore concluded that region 3.0 is not involved in the interactions with the TG motif in the aptamers.
To test the role of region 2.4 in the TG recognition, we analyzed whether mutations at this region could suppress TG mutations in sEcap1. In our experiment, we used the wild-type A subunit from T. aquaticus and a A mutant with a Gln to His substitution at position 260 of region 2.4 (26). Previously, point mutations of this amino acid in A and of the corresponding residue of 70 (Gln-437) were shown to suppress TG mutations in extended promoters (26,38). We found that wild-type A bound sEcap1 with high affinity (K d 9.5 nM) but did not bind sEcap1 mutant with the TG to CC substitution (K d Ͼ 3 M) ( Table 1 and Fig. 4). In contrast, the mutant A subunit was able to recognize both the wild-type and the mutant aptamers (K d 20 and 65 nM, respectively) ( Table 1 and Fig. 4). However, the mutant A did not bind sEcap1 with the Ϫ11C mutation at the  Ϫ10-like element (K d Ͼ 3 M). Thus, the suppression effect of the Gln-260 3 His substitution is position-specific, and this residue of A is likely directly involved in the recognition of the TG element in the aptamers.

Binding of the Aptamers Does Not Change the Distance between Regions 2 and 4-Measurements of distances
between different conserved domains of E. coli 70 using luminescence resonance energy transfer technique revealed large conformational changes in 70 upon binding to the core and, in particular, a significant increase in the distance between DNA binding domains 2 and 4 (23). We previously showed that the recognition of the sTap aptamers by the T. aquaticus A subunit was also accompanied by an increase in the 2-4 distance (26). To test whether these changes are absolutely required for the recognition of DNA by , we performed similar experiments with sEcap1. We used a previously described derivative of 70 (23) that contained a pair of cysteine residues in regions 2 and 4 (at positions 442 and 596). The was double-labeled with luminescence donor and acceptor probes and used for luminescence resonance energy transfer measurements. The results are shown in Fig.  5. In agreement with our published data (23), the apparent interdomain distance in free 70 was short, indicating a compact structure of the protein. The binding of the core enzyme led to a substantial increase in the 2-4 distance. A control oligonucleotide, which did not bind 70 , did not cause any changes in the interdomain distance. Remarkably, the addition of sEcap1 also did not alter the distance between 70 regions 2 and 4 (Fig. 5). Thus, specific recognition of the Ϫ10 element in sEcap1 by region 2 does not require separation of 2 from 4.

DISCUSSION
In this work, we report that the free primary subunit of RNAP is able to specifically recognize the Ϫ10 and TG promoter elements in ssDNA aptamers that mimic the nontemplate promoter strand. Because the aptamers contain the Ϫ10 promoter element in single-stranded form, their recognition by the subunit likely models the interactions of RNAP with nontemplate strand in the open promoter complex after RNAPinduced DNA melting has occurred. Remarkably, interacts with the aptamers with high affinity that is comparable to the affinity of holo-RNAP to nontemplate oligonucleotides bearing the Ϫ10 promoter element (7,12,39). Thus, the aptamers represent a novel type of model promoter substrates that can be efficiently recognized by free in the absence of core RNAP. DNA-protein cross-linking experiments and mutation analysis demonstrated that both the Ϫ10 and TG aptamer motifs are recognized by region 2 of . The involvement of this region in the recognition of the Ϫ10 element does not seem surprising, given its well established role in the recognition of this element in promoters. More interestingly, the TG motif in the aptamers is also recognized by region 2.4, whereas region 3.0, which was shown to be involved in its recognition in promoters (37,38), is not important for aptamer binding. We, therefore, propose that, during promoter recognition, regions 2.4 and 3.0 of specifically interact with the TG motif in the nontemplate and template promoter strands, respectively.
In support of this model, a 6-Å resolution structure of a complex of T. aquaticus holo-RNAP with a fork-junction promoter template localizes the TG-like DNA segment between two ␣-helixes formed by regions 2.4 and 3.0 (Fig. 6) (40). In the structure, amino acids from region 3.0 that have been implicated in the TG recognition (His-455 and Glu-458 in E. coli 70 , corresponding to His-278 and Glu-281 in T. aquaticus A , shown in light green in Fig. 6) (37) can potentially contact the template DNA strand. On the other hand, the ␣-helix formed by region 2.4 likely contacts nucleotides from the nontemplate strand (Fig. 6). We showed that a mutation of A Gln260 (Fig. 6, yellow) from this region suppresses substitutions of the  TG motif in the aptamers. Previously, substitutions of this and neighboring amino acids in region 2.4 of 70 were shown to suppress TG mutations in promoters (38). It should be noted that, in the RNAP-DNA model, Gln-260 is located downstream of the TG motif, closer to the first nucleotide of the Ϫ10 element (Fig. 6). In agreement with this, mutations of this amino acid in 70 were previously shown to suppress changes at the first position of the Ϫ10 sequence (41). Thus, the conformation and relative positions of promoter DNA and may change at different steps of transcription initiation to allow specific interactions between region 2.4 and the Ϫ10 and TG elements.
Previous studies suggested that regions 2 and 4 in free are masked for interaction with promoter DNA and that large scale conformational changes of induced by core RNAP are required for DNA recognition. In particular, it was proposed that in free (i) region 1.1 may inhibit DNA recognition by interacting with region 2, region 4, or both (17)(18)(19)(20)(21)(22), and (ii) the close proximity of regions 2 and 4 may mask the DNA binding sites and thus interfere with promoter recognition (24 -26) (see the introduction). In this study, we demonstrate that the aptamers containing the promoter-related motif can be efficiently recognized by region 2 independently of the presence of regions 1.1 and 4. Furthermore, we show that 70 in the complex with the aptamers maintains a compact conformation in which domains 2 and 4 are located at the same distance from each other as in free . In summary, our data demonstrate that (i) regions 1.1 and 4 are unlikely to be directly involved in the masking of DNA binding sites in region 2, and (ii) the increase in the distance between 2 and 4 observed upon holoenzyme formation is not required for the recognition of the Ϫ10 element by region 2.
At the same time, recognition of the Ϫ10 promoter element likely requires some local conformational changes in . In particular, it was shown that the binding of the core polymerase alters the solvent accessibility and chemical reactivity of amino acids in region 2.3 (12,27). The core region sufficient for the stimulation of DNA binding by region 2 was localized to the coiled-coil motif of the ␤Ј subunit (E. coli amino acids 262-309) that interacts with region 2.2 (Fig. 6, dark green) (8). We therefore propose that sEcaps are able to cause similar conformational changes in by interacting with the same region. Based on the proposed structure of sEcap1 and on the position of the nontemplate promoter strand in the promoter complex model (Fig. 6), one can suggest that region 2.2 is likely contacted by the G-quartet portion of the aptamer, which could in this case play a role analogous to that of the ␤Ј coiled-coil region. Thus, structural analysis of aptamer-induced conformational changes of would be an important subject of further studies.
Our results define a minimal fragment capable of specific DNA binding, which corresponds to structural domain 2 and  The nontemplate promoter strand (nucleotides from Ϫ23 to Ϫ7 relative to the transcription start point) is light blue, and the template strand (from Ϫ23 to Ϫ12) is dark blue. The nucleotides of the TG motif in the two strands are yellow and green, respectively; the nucleotides of the Ϫ10 element are orange, the first nucleotide is shown at a larger scale. contains conserved region 2, a part of region 1.2, and a nonconserved linker between them (Fig. 6). Given the small aptamer size (which roughly corresponds to the size of DNA fragment shown on the figure), the complex of sEcap with this fragment represents a very promising minimal system for structural and biochemical studies of the mechanisms of promoter recognition and opening by RNAP.
Recently, we described ssDNA aptamers (sTaps) that were specifically recognized by T. aquaticus A and contained the Ϫ10 promoter element (26). The aptamers described in this work also contain the Ϫ10-like element and are efficiently recognized by s from various bacteria, including T. aquaticus. Brief comparison of the properties of the two types of -specific aptamers reveals several important differences in the mechanisms of their interaction with . First, the Ϫ10 element in sTaps and sEcaps is recognized with slightly different specificity, with different preferences for nucleotide at the third position of the Ϫ10 element. Second, in the case of sTaps, an additional downstream GGGA motif was required for aptamer binding. In contrast, the recognition of sEcaps does not depend on downstream motifs but requires the presence of the TG motif and a G-quartet structure upstream of the Ϫ10-like element. Remarkably, is able to recognize the TG motif located at different positions relative to the Ϫ10 element in sEcaps. Third, the recognition of sEcaps does not involve major conformational changes in that were observed in the case of sTaps (26). This difference may result from the different structure of the two types of aptamers, with sTaps bearing a hairpin structure downstream and sEcaps containing a G-quartet structure upstream of the Ϫ10 element. Thus, the subunit shows a remarkable plasticity in the recognition of the promoter elements in model DNA substrates. This property may underlie the ability of RNAP to recognize a large diversity of promoters that can differ in the sequences of the conserved elements and distances between them. Analysis of different types of -specific DNA ligands, including the aptamers described in this work, opens a way for better understanding of delicate mechanisms of promoter recognition by RNAP.