Novel Protein–Protein Interaction Between Escherichia coli SoxS and the DNA Binding Determinant of the RNA Polymerase α Subunit: SoxS Functions as a Co-sigma Factor and Redeploys RNA Polymerase from UP-element-containing Promoters to SoxS-dependent Promoters during Oxidative Stress

SoxS is the transcription activator of the SoxRS regulon. Despite being synthesized de novo in response to oxidative stress and despite the large disparity between the number of SoxS binding sites and the number of SoxS molecules per cell, SoxS-dependent promoters are rapidly activated after the onset of the stress. With the usual recruitment/post-recruitment mechanisms being unsuitable for activating gene expression under these conditions, we previously proposed that SoxS functions by “pre-recruitment”. In pre-recruitment, SoxS forms SoxS–RNA polymerase binary complexes, which use the DNA binding properties of SoxS and σ⁷⁰ to distinguish SoxS-dependent promoters from housekeeping promoters and from the large number of sequence-equivalent but functionally irrelevant SoxS binding sites. With previous work in Escherichia coli having indicated that the most likely target on RNA polymerase for interaction with SoxS is the C-terminal domain of α, we investigated the interaction directly with the yeast two-hybrid system. We found that SoxS interacts with the αCTD and that SoxS positive control mutations disrupt the interaction. Moreover, single alanine substitutions of the αCTD that reduce or enhance SoxS activation in E. coli reduce or enhance the interaction between SoxS and the αCTD in yeast. Significantly, the critical amino acid residues lie in and around the DNA binding determinant of the αCTD, the first example of an activator contacting this determinant. These interactions were confirmed with an affinity immobilization assay. Lastly, we found that SoxS induction interferes with utilization of the UP element of an rRNA promoter. Thus, by functioning as a co-sigma factor that interacts with the DNA binding determinant of the αCTD, SoxS diverts RNA polymerase from UP-containing promoters to SoxS-activatable promoters.

Introduction

The SoxRS system of Escherichia coli is responsible for the cell's defense against superoxide anion, nitric oxide and redox-cycling compounds like paraquat that endogenously generate superoxide.¹ The regulatory process leading to induction of the member genes of the regulon during oxidative stress is unusual in that it occurs in two temporal stages of transcription, the first mediated by SoxR, and the second by SoxS.2, 3 SoxR is expressed constitutively in a form able to bind its sole DNA target, a site in the promoter of the soxS gene, but unable to activate soxS transcription.4, 5, 6, 7 However, oxidation of the 2Fe–2S centers of SoxR induces a conformational change in the protein that allows it to activate soxS transcription.4, 5, 8, 9 Then, SoxS, synthesized de novo in response to the oxidative stress, mounts the defense response by activating the transcription of the ∼40 genes of the regulon.10, 11 Thus, SoxR is the sensor and transmitter of oxidative stress while SoxS is the response regulator.

A number of characteristics distinguish SoxS from most other bacterial transcription activators. (1) As mentioned above, the defense response depends on the de novo synthesis of SoxS, a rather surprising requirement given the rapidity with which potentially lethal lesions are introduced by reactive oxygen species. (2) SoxS is very small, only 107 amino acid residues,12, 13 and it functions as a monomer.¹⁴ Its putatively rapid synthesis (∼seven seconds per monomer at a polypeptide chain elongation rate of 15 amino acid residues per second)¹⁵ compared to larger transcription factors and its ability to function without having to achieve the intracellular concentration for oligomerization may help compensate for the requirement for de novo synthesis. (3) SoxS activates transcription from two classes of promoter.16, 17 In class I promoters the SoxS binding site lies upstream of the −35 promoter hexamer, while at class II promoters the binding site overlaps the −35 promoter hexamer. (4) As shown by screening a library of single alanine substitutions of the αCTD for effects on SoxS-dependent transcription activation of class I and class II promoters, some residues are required for activation of both promoter classes while others enhance activation (K. L. Griffith, T. I. Wood, & R.E.W., Jr, unpublished results). The only alanine substitutions of the αCTD that affect activation are those of the residues within and around the 265 DNA binding determinant (K. L. Griffith, T. I. Wood, & R.E.W., Jr, unpublished results), the determinant that usually binds DNA non-specifically at a site adjacent to the DNA bound activator; with many activators, this binding of the 265 determinant to DNA is responsible for recruiting RNAP to the vicinity of the promoter.18, 19, 20 Thus, no cluster of substitutions of residues comprising either the 287 determinant or the 261 determinant were found that affected SoxS-dependent activation; with many other activators, these determinants participate in protein–protein interactions between the DNA-bound activator and RNAP.20, 21 Moreover, as mentioned above, alanine substitution of some of the residues within and around the 265 DNA binding determinant actually enhance SoxS-dependent activation of both classes of promoters. (K. L. Griffith, T. I. Wood, & R.E.W., Jr, unpublished results). These results suggest that the amino acid residues of the DNA binding determinant of the αCTD may participate not in non-specific DNA binding but in specific protein–protein interactions with SoxS (K. L. Griffith, T. I. Wood, & R.E.W., Jr, unpublished results).

Another property distinguishing SoxS from most other bacterial transcription activators is that its DNA binding site is highly degenerate.17, 22, 23, 24 Information content analysis of the SoxS binding sites and direct analysis by gel shift assay has indicated that fast-growing cells of E. coli contain ∼65,000 SoxS binding sites.24, 25 Since the SoxRS regulon contains only ∼40 genes,¹⁰ which is equivalent to ∼200 SoxS-dependent promoters per fast-growing cell with ∼five genomes per cell,¹⁵ it has been unclear how SoxS is able to mount the defense response against oxidative stress when it is synthesized de novo and the total number of binding sites per cell far exceeds the number of functional binding sites residing in SoxS-dependent promoters.

With one possible solution to the problem being the rapid synthesis of SoxS to a level high enough to saturate or nearly saturate all of the cell's SoxS binding sites, we determined the maximum number of SoxS molecules per cell produced following induction of its de novo synthesis. We found that the maximum is 2500 molecules per cell.²⁶ This 26-fold disparity between the total number of SoxS binding sites per cell and the maximum number of SoxS molecules per cell indicates that transcription activation by SoxS is not likely to occur by the usual recruitment mechanism whereby protein–protein interactions between RNAP and a DNA-bound activator recruits RNAP to the promoter and thereby enhances open complex formation.²⁷ As such, we proposed that SoxS employs “pre-recruitment” as the mechanism to distinguish SoxS binding sites residing at the proper position and orientation in SoxS-dependent promoters from the vast excess of sequence equivalent, but non-functional binding sites.²⁶ In pre-recruitment, de novo synthesized SoxS forms a binary complex with RNAP and the SoxS–RNAP binary complex then scans the chromosome for SoxS-dependent promoters using the σ⁷⁰ factor of RNAP to search for the −10 and −35 promoter recognition elements and SoxS to search for soxboxes located in the proper position and orientation for binding and transcription activation.

In addition to the circumstantial evidence leading to the pre-recruitment hypothesis (i.e. the disparity between the number of SoxS binding sites per cell and the number of SoxS molecules per cell), two lines of experimentation have provided additional support for the pre-recruitment mechanism. First, Martin et al.²⁵ demonstrated that SoxS and MarA, a protein whose amino acid sequence is 42% identical with that of SoxS, binds to the same sequence as SoxS, and activates transcription of the same set of genes, although to different degrees,28, 56 are able to form binary complexes with holo-RNAP in vitro at a K_d≅0.3 μM and that binary complexes can also form with core RNAP, which suggests that the target on RNAP with which the activators most likely interact is the αCTD. On the basis of their work, they proposed that SoxS and MarA activate transcription by “DNA scanning”,²⁵ a mechanism similar to pre-recruitment. Second, we determined the dominance relationships of SoxS mutations defective in either DNA binding or positive control. We found that over-expression of a mutant SoxS protein severely defective in DNA binding is dominant to the wild-type SoxS while over-expression of SoxS containing a positive control mutation is not.⁵⁷ This result is consistent with the pre-recruitment mechanism because the high abundance of mutant SoxS protein defective in DNA binding would be able to titrate all of the RNAP while forming SoxS–RNAP binary complexes that are unable to bind to and activate transcription of SoxS-dependent promoters; as such the DNA binding mutations are dominant, because they would prevent wild-type SoxS protein from forming functional SoxS–RNAP binary complexes. This result is inconsistent with the properties predicted of activators that function by recruitment.27, 29 For such an activator, positive control mutations are dominant to wild-type because the over-expressed mutant protein would be able to bind to its target in the promoter but would be unable to activate transcription; thus, the DNA-bound recruitment activator with a positive control mutation would interfere with the action of the wild-type activator. Moreover, unlike the case with pre-recruitment, DNA binding mutations of an activator functioning by recruitment are predicted to be recessive to the wild-type protein.

As the next step in characterizing SoxS as a transcription activator and the mechanism by which it functions, we addressed in this work the following questions. (1) Can SoxS interact in solution in vivo with a component of RNAP, in particular with the αCTD? (2) Are any of the previously isolated soxS positive control mutations,³⁰ of which some reduce activation of all SoxS-dependent promoters (class I/class II mutations) while others reduce activation only of class II promoters (class II mutations), defective in protein–protein interactions with RNAP, in particular with the αCTD? (3) If SoxS does interact in solution in vivo with the αCTD, does the protein–protein interaction depend on amino acid residues within and around the 265 DNA binding determinant?

We used the yeast two-hybrid system to address these questions and found for each that the answer was “yes”. We then confirmed the interactions and the specificity thereof by incubation of affinity-immobilized wild-type and mutant α subunits with E. coli cell extracts containing either wild-type SoxS or SoxS positive control mutants. Lastly, we demonstrated a profound physiological effect in E. coli of the interaction between SoxS and the DNA binding determinant of the αCTD: by masking the ability of the αCTD to bind DNA, formation of the SoxS–RNAP binary complex diverts RNAP from “strong”, UP-element-containing promoters³¹ to SoxS-dependent promoters and thereby hastens the defense response against oxidative stress. A preliminary account of this work has been presented.³²