Promoter Recognition As Measured by Binding of Polymerase to Nontemplate Strand Oligonucleotide

See allHide authors and affiliations

Science  23 May 1997:
Vol. 276, Issue 5316, pp. 1258-1260
DOI: 10.1126/science.276.5316.1258


In transcription initiation, the DNA strands must be separated to expose the template to RNA polymerase. As the closed initiation complex is converted to an open one, specific protein-DNA interactions involving bases of the nontemplate strand form and stabilize the promoter complex in the region of unwinding. Specific interaction between RNA polymerase and the promoter in Escherichia coliwas detected and quantified as the binding affinity of nontemplate oligonucleotide sequences. The RNA polymerase subunit sigma factor 70 contacted the bases of the nontemplate DNA strand through its conserved region 2; a mutation that affected promoter function altered the binding affinity of the oligonucleotide to the enzyme.

Escherichia coli RNA polymerase exists in two forms (1): a core enzyme (E) that consists of subunits ββ′α2 and is sufficient for elongation, and a holoenzyme (Eσ70) that includes E and a sigma polypeptide required for specific initiation of transcription—usually the sigma factor 70 (σ70). The holoenzyme recognizes primarily two hexameric sequences centered at −10 and −33, with +1 being the start of transcription; the −10 element is included in the region of initial promoter opening. The specificity for promoter recognition is carried by σ70, which can bind to double-stranded promoter DNA (2). It is likely, however, that Eσ70 mediates promoter opening.

70-DNA interactions that form and stabilize the open-promoter complex in the region of unwinding involve primarily the bases of the nontemplate DNA strand and not those of the template strand (3, 4). We used an electrophoretic mobility-shift assay (EMSA) (5) to investigate the specific binding of holoenzyme (2, 6-9) to small segments of single-stranded DNA that contain the −10 nontemplate strand promoter sequence. A 19-base oligonucleotide (oligo) (10, 11) containing variants of the −10 element of the λ late gene promoter PR′ was used as the single-stranded binding species (Fig. 1A). The λ sequences were chosen as a naturally occurring context for the different −10 hexamers. The nontemplate sequence of PR′from −18 to +1 was modified in the −10 region in three ways (Fig.1A). Oligo C contained the nontemplate consensus TATAAT hexamer, whereas oligo A contained nucleotides rarely found at each position in the −10 hexamer (12); oligo M contained a T to C mutation at the −12 position, a strong down mutation in many promoters (13). Oligo T, the complement of oligo C (Fig. 1A), was used to determine possible interaction of template strand with RNA polymerase.

Figure 1

EMSA for oligo binding. (A) Sequences of oligos C, A, M, and T are shown with the −10 element in larger letters. (B) RNA polymerase core (E) or holoenzyme (Eσ70) was assayed for its ability to bind end-labeled oligos.

70 bound oligo C and oligo M but failed to shift oligo A or T (Fig. 1B). Because the surrounding sequences of the three nontemplate oligos are identical, this result implies that the holoenzyme recognizes the single-stranded nontemplate sequence in the −10 region. Core enzyme bound all three nontemplate oligos nonspecifically (Fig. 1B) (14), as would be expected from the existence of numerous specific and nonspecific binding sites in the ββ′ and the α subunits (15, 16). The failure to bind the template oligo could be significant, or it might be an artifact of this particular sequence.

An equilibrium competition assay (17) was used to quantify the relative affinity of Eσ70 for the different oligos. Holoenzyme was incubated with labeled oligo C and increasing concentrations of unlabeled oligo competitor. After electrophoresis, the amount of radioactivity bound to the polymerase was measured (Fig.2). Competition by unlabeled oligo C reflects a stoichiometry of binding of about one oligo per enzyme. Oligo A did not compete detectably even in 20-fold excess over polymerase (200-fold molar excess over oligo C). Oligo M competed poorly but detectably, by a factor of 15 to 20 less than oligo C. This result confirmed that Eσ70 binds specifically to the nontemplate sequence of the −10 element.

Figure 2

Equilibrium competition between labeled oligo C and unlabeled oligo A (triangles), M (squares), or C (circles) for binding to RNA polymerase holoenzyme. Percentage bound is plotted as a function of competitor oligo per RNA polymerase. Open symbols designate Eσ70 and filled symbols designate Eσ70 Q437H.

Several mutations in region 2.4 of σ70 have been isolated as suppressors of mutations in the −10 consensus (13, 18, 19). The σ70 mutation Gln437 → His (σ70 Q437H) suppresses a T to C mutation at the −12 position (13), causing mutant promoter function to be about eight times greater, but not affecting wild-type promoter activity. To confirm that the detected binding reflects interactions important for promoter function, we used EMSA to examine the mutant sigma factor and the oligo containing the −12 T to C mutation (oligo M). σ70 Q437H was constructed with a histidine tag to facilitate purification; the histidine tag does not interfere with transcriptional activity or EMSA when σ70 is complexed with E (14). In vitro transcription with purified polymerase shows that the mutant sigma factor has wild-type activity on wild-type PR′ and partially suppresses a T to C mutation in PR′(14). This is similar to the results of in vivo analysis (13).

In addition to Eσ70, we tested Eσ70 Q437H in equilibrium competition against labeled oligo C (Fig. 2). As expected, oligo A failed to compete, and competition by oligo C showed identical stoichiometry for Eσ70 and Eσ70 Q437H. However, oligo M competed with oligo C about threefold greater for binding to Eσ70 Q437H than to Eσ70. The implication that Eσ70 Q437H binds oligo M with greater affinity than Eσ70 was verified by use of EMSA to measure an apparent binding constant (Kd) for oligos C and M with both mutant and wild-type polymerase (Fig. 3). Both Eσ70 and Eσ70 Q437H bound oligo C with a Kd of ∼3 nM. However, Eσ70 Q437H bound oligo M with twofold greater affinity than Eσ70, with Kd values of ∼7 and ∼15 nM, respectively. The difference was subtle but reproducible. Binding was also qualitatively different for holoenzyme containing the two sigma factors; mutant sigma factor displayed less smearing between the free and bound DNA, possibly reflecting a smaller dissociation rate. We presume that this binding preference accounts for suppression of the −12 T to C promoter mutation by the σ70 Q437H mutation.

Figure 3

Dependence of oligo binding on RNA polymerase concentration. Percentage of bound oligo C (A) or oligo M (B) is plotted against concentration of RNA polymerase. Circles denote Eσ70 and squares denote Eσ70 Q437H. On the right of each graph is an image of a typical (but different and independent) EMSA.

The difference in apparent Kd with the two sigma factors indicated that binding of the nontemplate oligos is directed by σ70. To confirm this, we used an ultraviolet cross-linking assay (20); ultraviolet light is expected to cross-link only polypeptides that are in intimate contact with the DNA. Binding reactions containing labeled oligo C and various forms of RNA polymerase were irradiated and then analyzed on an SDS-polyacrylamide gel (Fig. 4). Although the β and β′ subunits cross-linked slightly, the predominant species had the mobility of the σ subunit. To verify that σ was cross-linked, we used holoenzyme reconstituted with three sigma polypeptides of different molecular mass that recognize the same −10 region, and also a σ70truncation: Bacillus subtilis SigA (EσA; 43 kD), E. coli RpoS (EσS; 41 kD), domain 2 of sigma factor 70 (Eσ70 104–448; 39 kD), and σ70 amino acids 360 to 528 fused to glutathione-S-transferase (GST) (Eσ70 GST 360–528; 40 kD). Both the σ70 fragment, defining domain 2 (21), and the GST fusion product (14) were sufficient to bind oligo C in the context of holoenzyme. Each of the three sigma polypeptides cross-linked efficiently to oligo C (Fig. 4), shifting the mobility of the major cross-linked species according to the molecular mass of the sigma polypeptide used; SigA and RpoS ran larger than expected because full-length sigma polypeptides run anomalously in SDS–polyacrylamide gel electrophoresis. The GST fusion product cross-linked detectably but less efficiently. This may reflect a core binding deficiency, because, unlike the native sigma factors, the GST fusion product could not compete for binding with a (presumptive) contaminant σ fragment in the core preparation (Fig. 4).

Figure 4

UV cross-linking of RNA polymerase subunits to oligo C. Autoradiogram of SDS gel analysis of a UV cross-linking experiment is shown. Lanes 1 to 5, different forms of holoenzyme: Eσ70, EσA, EσS, Eσ70 104–448, and Eσ70 GST 360–528; lane 6, core (E) alone; lanes 7 and 8, overexposure of lanes 5 and 6 to show weakly cross-linking GST fusion sigma polypeptide. Positions of proteins were verified by silver staining and are shown on the left for Eσ70; molecular sizes of protein markers (in kilodaltons) are indicated on the right.

Thus, E. coli RNA polymerase recognizes single-stranded DNA oligos that represent the nontemplate strand of the open region of a promoter. Our findings agree with footprinting data (22, 23), with evidence that nontemplate bases are important for promoter function (3), and with cross-linking of sigma polypeptides to nontemplate sequences in the promoter region (24) and are consistent with proposals that DNA melting proteins might act through specific affinity for a single strand of DNA (25, 26). Furthermore, we have shown that the agent of binding is the σ subunit. By the use of overlapping truncations in the cross-linking assay and a mutation in region 2.4 of σ70 that suppresses a promoter mutant, as well as by homology alignment with B. subtilis SigA (27), we have localized the region of interaction to amino acids 374 to 448 of σ70, encompassing region 2. An atomic structure of region 2 has shown that Gln437 and other residues involved in recognizing the −10 segment and melting the DNA are arrayed on one face of an α helix, where they might contact the bases of the nontemplate DNA strand (28). Whether Eσ70recognizes TATAAT specifically as double-stranded DNA and later transforms this binding to the single-strand interaction we describe, or whether single-strand binding is the only base-specific interaction in the region of melting, arising after other forces (for example, superhelical coiling energy) (29, 30) initiate the opening process, remains unknown.


View Abstract

Stay Connected to Science

Navigate This Article