Protein-Protein Interactions Between σ⁷⁰ Region 4 of RNA Polymerase and Escherichia coli SoxS, a Transcription Activator that Functions by the Pre-recruitment Mechanism: Evidence for “Off-DNA” and “On-DNA” Interactions

Abstract

According to the pre-recruitment hypothesis, Escherichia coli SoxS activates transcription of the genes of the SoxRS regulon by forming binary complexes with RNA polymerase that scan the chromosome for class I and class II SoxS-dependent promoters. Previously, we showed that the a subunit’s C-terminal domain plays a role in activating both classes of promoter by making protein-protein contacts with SoxS; some of these contacts are made in solution in the absence of promoter DNA, a critical prediction of the pre-recruitment hypothesis. Here, we identified seven single alanine substitutions of region 4 of the σ⁷⁰ subunit (σ⁷⁰ R4) of RNA polymerase that reduce SoxS activation of class II promoters. With genetic epistasis tests between these σ⁷⁰ R4 mutants and positive control mutants of SoxS we identified ten pairs of amino acids that interact with each other in E. coli. Using the yeast two-hybrid system and affinity immobilization assays we showed that SoxS and σ⁷⁰ R4 can interact in solution, i.e., “off-DNA”. The interaction requires amino acids of the class I/II but not the class II positive control surface of SoxS and five amino acids of σ⁷⁰ R4 that reduce activation in E. coli also reduce the SoxS-σ⁷⁰ R4 interaction in yeast. One of the epistatic interactions that occur in E. coli also occurs in the yeast two-hybrid system, i.e., off-DNA. Importantly, we infer that the five epistatic interactions occurring in E. coli that require an amino acid of the class II surface occur “on-DNA” at class II promoters. Finding that SoxS contacts σ⁷⁰ R4 both off- DNA as well as on-DNA is consistent with the pre-recruitment hypothesis. Moreover, SoxS is now the first example of an E. coli transcriptional activator that uses a single positive control surface to make specific protein-protein contacts with two different subunits of RNA polymerase.

Introduction

During normal growth of Escherichia coli, most genes are transcribed constitutively and encode the general “housekeeping” functions required for growth. Transcription of these genes is effected by a holo-RNA polymerase (RNAP), which is comprised of the “core” subunits, α2, β, β’ and ω ^{1; 2}, and the predominant σ factor, σ^{70 3; 4}. The function of the core RNAP is to carry out transcription elongation ⁵, whereas the function of σ⁷⁰ is to promote transcription initiation by recognizing and binding to specific DNA sequence elements lying within the various promoters ⁶. The most common recognition elements in these promoters are two hexameric sequences optimally separated by 17 bp and centered at −10 and −35 bp with respect to the start site for transcription initiation ⁷. Promoters may also contain an extended −10 sequence, which compensates for a −35 element with little resemblance to typical hexamers ⁸, and an “UP” element, which provides a third recognition element that enhances RNAP binding to promoters like the rRNA P1 promoters ⁹.

In contrast to the housekeeping genes, a number of other genes are transcribed at high levels only under unique environmental conditions of carbon or nitrogen source, the absence of specific nutrients, or stress. The elevated transcription of these genes is controlled by small molecule effectors or environmental signals that regulate the activity of the transcriptional activators that turn-on their target genes or the transcriptional repressors that turn-off their target genes ^{10; 11}.

Recruitment is by far the most prevalent mechanism of transcription activation. In recruitment, a transcriptional activator binds to a specific DNA sequence within the promoter region of its target gene(s) ¹². In some cases, the activator, e.g., catabolite gene activator protein (CAP, also known as cyclic AMP receptor protein, CRP) only binds its target sequence(s) following activation by an inducing signal, e.g., cyclic AMP ^{13; 14}. Then, the DNA-bound activator “recruits” RNAP by making adhesive protein-protein interactions with it ¹⁵. Once recruited to the promoter, RNAP proceeds to form an open complex, then the initiation complex, and then the form that escapes the promoter ^{7; 16}. Activators functioning by recruitment can stimulate one or more of the steps that lead to transcription initiation and promoter escape ¹⁷. Recently, evidence has been presented for a new mechanism of transcriptional activation, called “pre-recruitment” ^{18; 19} or “DNA scanning” ²⁰. In this mechanism, the transcriptional activator forms a binary complex with RNAP in solution or when the two molecules collide during diffusion along the DNA. The activator-RNAP binary complexes then scan the chromosome for activator-dependent promoters that contain both a properly positioned binding site for the activator and promoter elements recognized by σ⁷⁰. So far, only SoxS ^{18; 19}, MarA ²⁰ and Rob (E. F. Keen, III and R. E. Wolf, Jr., unpublished results) appear to activate transcription by this mechanism.

These three proteins form a closely related subset of the AraC/XylS family of bacterial regulatory proteins ^{21; 22}. The three paralogs function as monomers ^23–25, bind to the same highly degenerate 20 bp DNA sequence ^26–29 and activate the transcription of the same set of ∼50 target genes ^30–32, the SoxS/MarA/Rob regulon. Expression of the regulon provides resistance to superoxide stress, antibiotics, heavy metals, bile salts, and medium-chain fatty acids, as well as tolerance to organic solvents ^{30; 33–42}.

Despite their high degree of amino acid sequence identity, the three proteins differentially activate transcription of the regulon’s genes ^30–32, whose promoters fall into two classes ^{27; 28; 43}. For example, SoxS efficiently activates transcription from class I promoters, wherein the soxbox lies upstream of the −35 hexamer and in the backward orientation, and from class II promoters, wherein the soxbox overlaps the −35 hexamer and resides in the forward orientation ⁴³. In contrast, MarA and Rob efficiently activate transcription from class II promoters (e.g., fumC) but activate class I promoters (e.g., fpr) less well ^{23; 33; 44; 45} (E. F. Keen III and R. E. Wolf, Jr., unpublished results). Note that despite this general distinction between the three paralogs, each activator activates transcription of the individual target genes to different degrees.

Besides appearing to activate transcription by pre-recruitment, the three paralogs differ in another important way from most other transcriptional activators, namely, the means by which they become active and able to activate transcription. Thus, SoxS and MarA activate transcription of the member genes of the regulon only upon their de novo synthesis in the second stage of a two-gene, two-stage system ^{46; 47}. For example, in the first stage of the SoxR/SoxS system, the 2Fe-2S centers of constitutively expressed SoxR become oxidized by the stress imposed by superoxide-generating, redox-cycling compounds like paraquat; this oxidation alters the conformation of SoxR so that it becomes an active activator, which then activates transcription of only one gene, soxS ^48–52. In the second stage, newly synthesized molecules of SoxS form binary complexes with RNAP, which then scan the chromosome for soxboxes that lie within SoxS-dependent promoters, thereby distinguishing them from the sequence-equivalent but non-functional soxboxes that do not lie within SoxS-dependent promoters ^{18; 19}. Thus, in this two-component system, SoxR functions as the sensor-transmitter of oxidative stress, while SoxS functions as the response-regulator.

In contrast to SoxS and MarA, Rob is expressed constitutively ^{53; 54} and, like most activators, it is inactive until activated by an inducing signal ^{40; 45; 55}. What is unique about Rob is that its activity is regulated by a new mechanism, “sequestration-dispersal”. Thus, most of Rob’s 5,000–10,000 molecules/cell are sequestered into 3–5 immunostainable clusters and the clustered molecules are inactive ^{56; 57}. Then, upon exposure to inducing compounds ^{40; 55}, Rob molecules become dispersed and form binary complexes with RNAP, which then scan the chromosome for activatable promoters ⁵⁷ (E.F. Keen III and R.E. Wolf, Jr., unpublished results).

A critical function of activators that bind DNA is how they interact with RNAP, either “on-DNA”, in the case of activators that function by recruitment, or both “off-DNA and on-DNA, in the case of activators that function by pre-recruitment. Much is known about the protein-protein contacts with RNAP made by activators that activate transcription by recruitment: they interact with determinants on either the C-terminal or N-terminal domains of the α subunits (αCTD and αNTD, respectively) or with amino acids lying within σ⁷⁰ region 4 (σ⁷⁰ R4), with the specific contact sites usually depending on the location of the activator binding site within the target promoter ^{10; 58; 59}.

Because the discovery of the pre-recruitment mechanism is relatively recent, less is known about the nature of the protein-protein interactions between these activators and RNAP. However, evidence has been presented that SoxS ⁶⁰, MarA ⁶¹ and Rob (E. F. Keen III and R. E. Wolf, Jr., unpublished results) differ from other activators in yet another mechanistic property: transcription activation of both class I and class II promoters requires previously unknown protein-protein interactions between the activator and the 265 determinant of the αCTD, the set of amino acids known to bind specifically to the UP element within rRNA promoters ^{15; 62}. As such, these interactions enhance the respective defense responses by diverting RNAP from strong, UP element-containing promoters to the activator-dependent promoters ⁶⁰ (E.F. Keen III and R.E. Wolf, Jr., unpublished results). Moreover, with SoxS and Rob, the interactions with the 265 determinant have been shown to require amino acids of the class I/II surface, mutations of which interfere with activation of both classes of promoters, but not amino acids of the class II surface, mutations of which interfere only with activation of class II promoters ^{60; 63} (E.F. Keen III and R.E. Wolf, Jr., unpublished results).

Since almost all transcriptional activators that activate transcription of class II promoters do so by making protein-protein contacts between the activator and σ⁷⁰ R4, we thought it highly likely that SoxS would do so too. Moreover, we expected that, at the least, amino acid residue D75 of the class II surface of SoxS would interact with an amino acid of σ⁷⁰ R4, since the homologous amino acids of AraC/XylS family members ⁶⁴ RhaS (D241), RhaR (D276) and MelR (D261) are all known to require protein-protein contact with R599 of σ⁷⁰ R4 ^65–67. Accordingly, in this work, we used a library of single alanine substitutions of σ⁷⁰ R4 encompassing positions 590 – 613 to identify amino acids in this segment of RNAP that are important to transcriptional activation of SoxS-dependent promoters. Finding several, we then determined which of the positive control mutants of SoxS are epistatic to the substitutions in σ⁷⁰ R4 that confer a defect in activation. The epistatic interactions identified in E. coli could represent interactions between SoxS and RNAP that occur either off-DNA in binary complexes or on-DNA at SoxS-dependent promoters (or both). Then, as a step toward determining whether SoxS can interact with σ⁷⁰ R4 in solution, i.e., not on promoter-specific DNA, we examined the SoxS-σ⁷⁰ R4 interaction in a yeast two-hybrid (Y2H) system ⁶⁸ and by affinity immobilization assays (AIA). Finding that SoxS does interact with σ⁷⁰ R4 in solution in the Y2H system, we then identified the alanine substitutions within the class I/II and class II surfaces of SoxS that disrupt the interaction in yeast and we determined which alanine substitutions of σ⁷⁰ R4 that reduce transcriptional activation by SoxS in E. coli also disrupt the interaction in yeast. Then, we conducted genetic epistasis tests with the Y2H system and determined that one interaction between SoxS and σ⁷⁰ R4 that occurs in E. coli also occurs in yeast. Lastly, we conducted AIA with full-length his₆-σ⁷⁰ and SoxS and found that immobilized his₆-σ⁷⁰ can capture SoxS from a crude cell extract. Together, the results of these analyses provide evidence for the existence of two types of interactions between SoxS and σ⁷⁰R4, with one being able to occur off-DNA, e.g., within a SoxS-RNAP binary complex that scans the chromosome for SoxS-dependent promoters, and the other occurring on-DNA e.g., within a SoxS-RNAP binary complex residing at a SoxS-dependent promoter.

Results and Discussion

Effects of single alanine substitutions in σ⁷⁰ R4 on SoxS-dependent transcription activation at class I and class II promoters

We introduced a library of single alanine substitutions at sixteen positions between amino acid residues 590 to 613 of σ⁷⁰ R4 ^{69; 70} into four derivatives of strain N7840 Δ(araBAD)714 [pBAD33-his₆-SoxS]; two carried single-copy transcriptional fusions of lac to class I promoters fpr and zwf and two carried lac fusions to class II fusions fumC and micF. Wild type σ⁷⁰ and the single alanine substitutions of it are expressed constitutively from the pGEX-2T-σ⁷⁰ plasmids carrying the respective σ⁷⁰ alleles ^{69; 70}. After induction of his₆-SoxS expression with 0.02% arabinose, the effects of the substitutions on SoxS-activated transcription of the four promoters were determined by assay of β-galactosidase activity. Since the abundance of the mutant σ⁷⁰ proteins expressed from the leaky tac promoter of the pGEX-2T-σ⁷⁰ plasmids approximately equals that of the chromosomally encoded wild type protein ⁷⁰, we accepted, as have others ⁶⁶, activity values below 80% of that obtained with wild type σ⁷⁰ as indicating that the substitution within σ⁷⁰ R4 confers a significant defect in transcription activation.

Table 1 shows that single alanine substitutions at positions R593, R596, R599 and R603 of σ⁷⁰ R4 each reduce SoxS-activation of the class II promoters fumC and micF by more than 30%. Thus, as expected from prior work on activation of class II promoters by other members of the AraC/XylS family ^65–67, σ⁷⁰ R4 is required for SoxS-dependent activation of its class II promoters.

Table 1.

The effect of σ⁷⁰ mutants on SoxS-dependent transcription activation at two class I promoters (fpr, zwf) and two class II promoters (fumC, micF).

σ70 Region 4 Substitutions	SoxS Dependent Transcription Activation
	Class I SoxS Promoters		Class II SoxS Promoters
	fpr	zwf	fum C	micF
I590A	87% ±6	144% ±11	151% ±17	111% ±12
E591A	92% ±8	106% ±19	110% ±9	90% ±2
K593A	115% ±4	127% ±23	44% ±8	55% ±3
L595A	115% ±7	117% ±14	137% ±17	106% ±5
R596A	82% ±3	106% ±8	38% ±8	59% ±5
K597A	90% ±5	112% ±10	125% ±8	106% ±11
L598A	136% ±7	109% ±10	129% ±17	105% ±10
R599A	94% ±8	125% ±15	63% ±9	69% ±3
H600A	123% ±9	132% ±11	96% ±12	117% ±9
S602A	101% ±6	107% ±10	116% ±15	115% ±9
R603A	88% ±4	88% ±2	27% ±4	43% ±5
S604A	89% ±2	86% ±4	110% ±6	117% ±6
E605A	85% ±2	70% ±5	88% ±4	103% ±6
R608A	84% ±6	37% ±5	112% ±17	95% ±10
D612A	83% ±3	39% ±7	85% ±5	110% ±
D613A	88% ±1	88% ± 2	103% ±8	115% ±6

The experiments were conducted with four Δ(araBAD) 714 derivatives of strain N7840 [pBAD33-his₆-SoxS] each lysogenic for a single copy λ prophage carrying one of the above SoxS-dependent promoters fused to lacZ and expressing the wild type allele of rpoD from the chromosome. Each of the four fusion strains was transformed with a member of the pGEX-2T-σ⁷⁰ library composed of wild type σ⁷⁰ and single alanine substitutions of σ⁷⁰ R4 mutants; the plasmid-encoded rpoD allele is expressed constitutively. As described in Materials and Methods, the cultures were grown in LB medium in polypropylene blocks to A₆₀₀ = 0.1–0.2, treated with 0.02% arabinose to induce SoxS synthesis and β-galactosidase activity was determined when the A₆₀₀ of the cultures reached 0.6–0.8. The average values and standard errors of the mean were determined. Activity values obtained with the σ⁷⁰ R4 mutants were expressed as a percent of the value produced by wild type σ⁷⁰. As described in the text, when a σ⁷⁰ R4 mutant produced an average activity value that is below 80% of the value produced by pGEX-2T-σ⁷⁰ carrying the wild type rpoD allele, the σ⁷⁰ R4 mutant was considered to confer a defect in SoxS-dependent transcription activation of that promoter. Mutant values are highlighted. The activity values produced by the four fusion strains carrying pGEX-2T-σ⁷⁰ and expressing the wild type rpoD allele were 1333 and 2720 units for the class I fpr and zwf promoters, respectively, and 586 and 2520 units for the class II promoters micF and fumC promoters, respectively.

Importantly, none of the substitutions of σ⁷⁰ R4 reduce SoxS activation of fpr (Table 1), a conventional class I promoter whose binding site lies in the backward orientation and 15 bp upstream of the putative −35 promoter element, presumably because SoxS bound to its binding site is too far from σ⁷⁰ R4 to interact with it even though SoxS bends DNA about 35° ^{23; 25}. This absence of effects of substitutions of σ⁷⁰ R4 on SoxS activation of this class I promoter contrasts with the arrangement of protein-protein contacts at class I CAP-dependent promoters, e.g., lac, where the αCTD is wedged between DNA-bound CAP and promoter-bound RNAP such that the 287 determinant of the αCTD interacts with Activation Region 1 of CAP’s promoter-proximal subunit while the αCTD’s 261 determinant interacts with amino acid residues within σ⁷⁰ R4 ⁷¹. Indeed, the results with fpr are consistent with the fact that SoxS interacts with the αCTD by directly contacting its 265, DNA-binding determinant, thereby placing the αCTD in a position where it cannot interact with σ⁷⁰ R4 ⁶⁰.

Table 1 also shows that alanine substitutions at positions E605, R608 and E612 of σ⁷⁰ R4 significantly reduce Sox-dependent activation of the zwf promoter, even though SoxS binds 7 bp upstream of zwf’s −35 promoter hexamer ^{26; 63}. A similar situation pertains to transcription activation by Ada of the aidB, ada, and alkA promoters wherein different amino acids of σ⁷⁰ R4 are required for activation by Ada, depending on whether the binding site overlaps the −35 region or whether the site is 5–7 bp upstream of the promoter hexamer ⁷². For SoxS, its binding at the zwf promoter is likely able to interact with σ ⁷⁰ R4 because its binding to DNA induces a ∼35° bend in the DNA ^{23; 25} that would bring SoxS and RNAP close enough together to allow protein-protein contact.

Previously, zwf has been classified as having a class I promoter SoxS-dependent promoter because its SoxS binding site lies upstream of the −35 hexamer ⁴³, even though the site lies in the forward orientation like those of canonical class II SoxS-dependent promoters. Now, with the zwf promoter behaving like a class II SoxS-dependent promoter because its activation by SoxS requires amino acid residues of σ⁷⁰ R4, it would be appropriate to reclassify zwf as a non-conventional class II promoter, i.e. class II*.

All of the amino acid residues of σ⁷⁰ R4 that cause a defect in transcription activation by SoxS are charged, with five being positively charged and two being negatively charged. As mentioned above, some of the residues that are defective in SoxS-dependent transcription activation are also important to transcription activation by other AraC/XylS family members at their respective promoters. Thus, residues K593 and R599 of σ⁷⁰ R4 are important to transcription activation by RhaS and RhaR ^{65; 66} while these two residues and R596 are also important for MelR-dependent transcription activation ⁶⁷. In addition, residues R599 and R603 of σ⁷⁰ are important to Rob-dependent transcription activation (E.F. Keen III and R.E. Wolf, Jr., unpublished results). Interestingly, K593 not only makes protein-protein contact with transcription activators but it also interacts specifically with DNA ⁷³, suggesting that certain amino acids of σ⁷⁰ play a dual role of interacting with promoter DNA and interacting with activators. Importantly, several residues of σ⁷⁰ R4 that are required for transcription activation of SoxS promoters are also necessary for transcription activation by CAP, FNR and λcI of their target promoters ^{70; 73–76}. Thus, in agreement with the observation of Lonetto et al ⁷⁰, the same amino acid residues of σ⁷⁰ R4 are used to make specific protein-protein contacts with many different transcription activators

In addition to amino acid residues within the helix-turn-helix motif of region 4.2 of σ⁷⁰, some of the residues important to transcription activation by SoxS reside in the C-terminal tail, viz., R603, E605, R608 and D612. Since most activators that interact with σ⁷⁰ R4 do so by making protein-protein contacts with region 4.2 and do not contact the C-terminal tail, we infer that the mechanism of transcription activation by SoxS at certain promoters requires these additional contacts. For example, the position of the soxbox with respect to the −35 hexamer varies from promoter to promoter ⁴³. Thus, the soxbox of the fumC promoter overlaps the entire −35 hexamer and extends 4 bp beyond it whereas the soxbox of the zwf promoter lies 7 bp upstream of the hexamer. Accordingly, SoxS bound at zwf might only be able to contact amino acid residues of the C-terminal tail while SoxS bound at fumC might only be able to contact a subset of amino acids in region 4.2. Alternatively, activation by pre-recruitment might require two types of protein-protein contacts between SoxS and RNAP, contacts that mediate binary complex formation and contacts necessary for transcription initiation that are made after the binary complex binds the promoter.

Identification of amino acids of σ⁷⁰ R4 that make protein-protein contact(s) with amino acids of the class I/II and/or class II positive control surfaces of SoxS in the activation of transcription from class II and class II* promoters

To identify protein-protein interactions critical to SoxS activation of class II and class II* promoters, we carried out genetic epistasis tests, also called “loss of contact” ^{66; 77} experiments, using the single alanine substitutions of σ⁷⁰ R4 that confer defects in transcription activation at these promoters (Table 1) and the single alanine substitutions of SoxS that display a positive control phenotype ⁶³. Before beginning the epistasis tests, we needed to determine whether the presence of plasmid pGEX-2T-σ⁷⁰ interfered in some way with the reduction in SoxS-dependent transcription conferred by the SoxS positive control substitutions ⁶³. To do this, we transformed plasmid pGEX-2T-σ⁷⁰ into the strains carrying lac fusions to the fumC, micF, and zwf promoters. We then introduced plasmid pBAD33-his₆-SoxS and its mutant derivatives carrying single positive control mutations of the class I/II (K5A, D9A, K30A, S31A, and V45A) or class II (F74A, D75A, M78A, D79A and Q85A) surfaces ⁶³. (The class I/II surface lies above the N- terminal helix-turn-helix DNA binding motif of SoxS while the class II surface lies above the C-terminal motif ⁶³). We then assayed β-galactosidase activity in the three sets of 11 strains. All of the positive control mutants reduced transcription activation at the three promoters and did so by amounts similar to those obtained previously in fusion strains that express σ⁷⁰ only from the chromosomal rpoD gene (data not shown) ⁶³.

For the epistasis tests, we prepared three sets of strains carrying each of the σ⁷⁰ R4 substitutions that reduce transcription activation of a given promoter in combination with each of the ten substitutions of SoxS that confer defects in positive control. The activity of β-galactosidase was determined for each combination of relevant σ⁷⁰ R4 substitution and SoxS positive control mutant at each of the three promoters. For each combination, we also used β-galactosidase assays to determine the effect of the σ ⁷⁰ R4 substitution on activation of the promoter by wild type SoxS and the effect on SoxS-dependent transcription of the SoxS positive control substitution in the presence of pGEX-2T-σ⁷⁰ carrying wild type σ⁷⁰.

The logic underlying the epistasis tests is that if the side group of a specific amino acid residue of one protein, e.g., σ⁷⁰, makes a protein-protein interaction with the side group of a specific amino acid residue of a second protein, e.g., SoxS, then the effect of combining single alanine substitutions at those positions of the two proteins should be no greater than the more severe effect of the two individual substitutions; such a combination of mutations is referred to as being epistatic. On the other hand, if the side group of a specific amino acid residue of one protein does not make protein-protein contact with the side group of a specific amino acid residue of the second protein, then the effect of combining single alanine substitutions at those positions of the two proteins should be independent of each other, with the result that the effect of the combination of mutations is greater than that of either substitution alone; in this case, the combination is non-epistatic. Figure 1 shows a representative example of an epistatic and a non-epistatic interaction at each promoter.

Figure 1.

Representative epistatic and non-epistatic interactions between alanine substitutions of σ⁷⁰ R4 and alanine substitutions of SoxS that individually confer a defect in transcription activation at a given SoxS-dependent promoter. The experiments were carried out as described in Materials and Methods. A. The left panel shows that at the fumC promoter amino acid substitution R599A of σ⁷⁰ R4 is epistatic to amino acid substitution D9A of SoxS, while the right panel shows that R599A is not epistatic to D79A of SoxS. B. The left panel shows that at the micF promoter, amino acid substitution R603A of σ⁷⁰ R4 is epistatic to amino acid substitution D75A of SoxS, while the right panel shows that R603A is not epistatic to M78A of SoxS. C. The left panel shows that at the zwf promoter amino acid substitution D612A of σ⁷⁰ R4 is epistatic to amino acid substitution K30A, while the right panel shows that D612A is not epistatic to S31A of SoxS.

Table 2 summarizes the epistasis data obtained with the seven single alanine substitutions of σ ⁷⁰ R4 that reduce SoxS-dependent transcription activation at the two class II promoters and the one class II* promoter examined. Epistasis at one or more promoters was obtained with five of the seven mutants of σ ⁷⁰ R4. We found a total of ten epistatic interactions when we combined the ten positive control mutants of SoxS with σ ⁷⁰ R4 substitutions K593A, R596A, R599A and R603A and examined their effects at the micF and fumC promoters (Table 2). (The zwf promoter was not included in these tests because, as shown in Table 1, these four σ ⁷⁰ R4 substitutions have no effect on SoxS activation of this promoter.)

Table 2.

Summary of the epistatic interactions in E. coli between amino acid residues of σ⁷⁰ R4 and amino acid residues of SoxS at specific SoxS-dependent promoters.

Promoter (Class)	σ70 R4 Amino Acid Residue Conferring Activation Defect	Interacting SoxS residue
		Class I/II Surface	Class II Surface
fumC (Class II)	K593	S31	M78, D75
micF (Class II)		S31	M78, D79
fumC (Class II)	R596	None	None
micF (Class II)		None	None
fumC (Class II)	R599	D9	D75
micF (Class II)		None	None
fumC (Class II)	R603	None	None
micF (Class II)		None	D75, D79
zwf (Class I)	E605	None	None
zwf (Class I)	R608	D9	None
zwf (Class I)	D612	K30	None
fpr (Class I)	None	None	None

All amino acids of σ⁷⁰ R4 that confer a defect in SoxS-dependent transcription activation at a given promoter when substituted with alanine are listed even if no interacting partners were identified. Epistasis was determined as described in the text. The amino acid residues of the two SoxS positive control surfaces that are epistatic partners of the σ⁷⁰ R4 substitutions are listed according to the promoter at which the epistasis was observed.

Several epistatic interactions between SoxS and σ⁷⁰ R4 are particularly interesting. (1) Substitution S31A of the class I/II surface of SoxS and M78A of the class II surface are epistatic to σ ⁷⁰ R4 K593A at both the fumC and micF promoters; since S31 and M78 lie in two different positive control surfaces of SoxS, their interactions with K593 probably occur in different SoxS-RNAP complexes. (2) K593A is also epistatic to D75A and D79A of the class II surface at the fumC and micF promoters, respectively. Using two different amino acids of SoxS for activation of the two class II promoters may reflect the difference in the position of the two soxboxes, with the 3’ end of the fumC soxbox extending 5 bp further downstream of the −35 region of its promoter than does the micF soxbox ⁴³; moreover, this difference may indicate that these particular interactions occur on DNA. (3) D9A of the class I/II surface and D75A of the class II surface are epistatic to σ ⁷⁰ R4 R599A at the fumC promoter but not at the micF promoter; as with the epistatic interactions of S31A and M78 of SoxS with K593A of σ⁷⁰ R4, these interactions are also likely to occur in different SoxS-RNAP complexes. (4) Since D75A of SoxS is epistatic to both K593A and R599A of σ⁷⁰ R4 at the fumC promoter, it is unclear whether these interactions occur simultaneously or in different complexes. The same is true of D79A of SoxS, which is epistatic to K593A and R603A at the micF promoter. (5) In addition to being epistatic to R599A of σ⁷⁰ R4 at the fumC promoter, D9A of the class I/II surface of SoxS is also epistatic to R608A at the zwf promoter; thus a single amino acid of a positive control surface of SoxS interacts with different amino acids of σ⁷⁰ R4, presumably because the fumC soxbox overlaps the promoter’s −35 hexamer by 4 bp while the zwf soxbox lies 7 bp upstream of its promoter’s −35 hexamer. Indeed, the position of the Ada binding site with respect to the −35 region of its class II-type also affects the specific amino acids of sigma with which it makes protein-protein interactions. (6) No epistatic partners were found for R596A and E605A of σ⁷⁰ R4. However, with charge reversal R596E and the experimental design of Westblade et al. ⁷⁸, we determined that the role of the side group of R596 is to maintain the tertiary structure of σ⁷⁰ R4 so that other amino acid residues are optimally positioned to bind to SoxS (data not shown). The same approach did not indicate a similar role for D605 and thus the basis for its ability to reduce SoxS-dependent transcription at the zwf promoter remains unknown. Lastly, we note that just as the amino acid homologs of D75 of SoxS in RhaS (D241), RhaR (D276), and MelR (D261) contact R599 of σ⁷⁰ R4 at their respective target promoters ^65–67 so too does D75 of SoxS contact R599 at the fumC promoter.

Determination of whether SoxS and σ⁷⁰ R4 can interact with each other in vitro

As described in Materials and Methods, we bound his₆-σ⁷⁰ R4 to Ni-NTA resin and then, after washing, the resin was incubated with the soluble fraction of a cell extract prepared from a culture expressing native SoxS. Figure 2 shows that elution of N-his₆-σ⁷⁰ R4 from the Ni-NTA agarose resin leads to the co-elution of native SoxS. Thus, SoxS must have been captured by resin-bound N-his₆-σ⁷⁰ R4 because SoxS was absent when the resin was initially incubated with soluble material from a control extract lacking N-his₆-σ⁷⁰ R4 (data not shown). Importantly, this experiment demonstrates that SoxS can indeed interact with N-his₆-σ⁷⁰ R4 in solution and in the absence of specific promoter DNA. Accordingly, we proceeded to determine whether SoxS and σ⁷⁰ R4 can interact in the Y2H; finding that they do, we investigated the nature of and the requirements for the interactions.

Figure 2.

AIA of the protein–protein interactions between SoxS and amino acids 541–613 of σ⁷⁰ R4. The immobilization of his₆-σ⁷⁰ R4 on Ni-NTA resin and its incubation with a cell-free extract containing wild type SoxS were carried out as described in Materials and Methods, as was the subsequent SDS-PAGE and western blotting of the respective fractions obtained after application of the SoxS-containing extract. FT, flow-through fraction; W3, third wash fraction; EL, eluate. Arrows indicate the location of his₆-σ⁷⁰ R4 and SoxS. A. The blot probed with antibody directed against the his₆ tag. Samples FT and W3 show that all of the σ⁷⁰ R4 bound to the column. Sample EL shows that all of the σ⁷⁰ R4 was eluted from the Ni-NTA resin upon incubation with 0.5 M imidazole. B. The blot probed with anti-SoxS antibodies. The FT sample shows that a large portion of SoxS was not bound to the column while the W3 fraction reveals that all of the non-specifically bound SoxS was washed away. Sample EL shows that SoxS was eluted from the Ni-NTA resin column with 0.5 M imidazole. Taken together, the two blots suggest that σ⁷⁰ R4 binds the column, that the bound protein retains SoxS, and that both proteins are co-eluted.

Analysis of protein-protein interactions in solution between SoxS and σ⁷⁰ R4 with the Y2H system

In the Y2H system used here, three recombinant plasmids were introduced into strain EGY48, which carries nutritional markers complemented by genes carried on the plasmids. (1) The DNA binding domain (DBD) of the LexA repressor was fused in-frame to SoxS to form the “bait”-expressing plasmid pGILDA-SoxS. (2) An 88 amino acid peptide encoding an activation domain, B42AD, was fused in-frame to amino acids 531–613 of σ⁷⁰ to form the “prey”-expressing plasmid pB42AD-σ⁷⁰ R4. Expression of the bait and prey proteins is under control of a fully functional, galactose-inducible GAL1 promoter. (3) Plasmid p8oplacZ carries a lacZ gene whose transcription is controlled by a minimal GAL1 promoter comprised of a TATA box and eight upstream operators that bind LexA’s DBD. Thus, lacZ serves as the reporter of transcription activation from this promoter, since β-galactosidase will only be produced when the LexA-SoxS bait protein binds the operators and recruits the B42AD activation domain to the promoter through protein-protein interaction with the B42AD-σ⁷⁰ R4 prey protein. Three negative control strains were also prepared as derivatives of strain EGY48[p8oplacZ] carrying the following plasmid pairs: pGILDA + pB42AD; pGILDA-SoxS + pB42AD; pGILDA + pB42AD-σ⁷⁰ R4. Strain EGY48[p8oplacZ] carrying pGILDA-SoxS and pB42AD-αCTD served as the positive control, since the interaction between SoxS and the αCTD was first revealed in Y2H experiments using this strain ⁶⁰. In this Y2H system, expression of β-galactosidase activity can be assessed qualitatively by the color of colonies formed by cells plated on triple drop-out medium containing raffinose (raf) as the carbon source, galactose (gal) as the inducer and X-gal as a chromogenic substrate; expression can also be assessed quantitatively by direct assay of β-galactosidase activity in cells growing under the same inducing conditions. The data presented in Table 3 show that all three negative control strains form white-colored colonies on the raf + gal + X-gal indicator plates and thus none of the negative controls produced false positives. In contrast, the test transformants co-expressing the LexA-SoxS bait protein and the B42AD-σ⁷⁰ R4 prey protein formed pale blue-colored colonies on the indicator plate, indicating that transcription of the 8oplacZ reporter had been activated. Thus, the plate phenotype of the test transformants provides evidence for a protein-protein interaction between SoxS and σ⁷⁰ R4. We note, however, that the strength of the SoxS-σ⁷⁰ R4 interaction appears to be less than that of the interaction between SoxS and the αCTD, since cells co-expressing LexA-SoxS and B42AD-αCTD formed colonies on the indicator plate that were a darker blue color, as observed previously ⁶⁰.

Table 3.

Demonstration of protein-protein interactions between SoxS and σ⁷⁰ R4 in the Y2H system and the effects of positive control substitutions of SoxS on the interaction

LexA Fusion (Bait)	B42 Fusion (Prey)	Colony Color	Induction Ratio	Western Analysis
pGILDA (empty)	pB42AD (empty)	White	2		→ β-Tubulin
pGILDA-SoxS	pB42AD (empty)	White	2	N.T.
pGILDA (empty)	pB42AD-σ70R4	White	2	N.T.
pGILDA-SoxS	pB42AD-αCTD	Blue	25	N.T.
pGILDA-SoxS	pB42AD-σ70R4	Pale Blue	14		→ LexA-SoxS
pGILDA-SoxS I/II− +II− (5, 9, 30, 31, 45, 74, 75 78, 79, 85)	pB42AD-σ70R4	White	2
pGILDA-SoxS I/II− (9, 30, 31, 45)	pB42AD-σ70R4	White	2
pGILDA-SoxS II− (74, 75, 78,79, 85)	pB42AD-σ70R4	Pale Blue	14

Strain EGY48[p8oplacZ] carrying the indicated bait and prey fusions expressed from pGILDA and pB42AD plasmids was prepared as described in Materials and Methods. Colony color was determined after 48 hr of growth at 30 °C on SD/gal/raf/-his/-trp/-ura/X-gal plates. The β-galactosidase activities in the respective strains were determined during growth in liquid inducing medium, SD/gal/raf/-his/-trp/-ura, and liquid non-inducing medium, SD/glu/-his/-trp/-ura, as described in Materials and Methods. The induction ratios were computed by dividing the specific activity values obtained in inducing medium by the specific activity values obtained in non-inducing medium. Western blots were carried out to determine the relative expression levels of wild type and mutant SoxS proteins with antibodies directed against SoxS and β-tubulin. N.T., not tested.

Next, we quantified the strength of the interaction between SoxS and σ⁷⁰ R4 by determining the effect of co-expressing the bait and prey fusion proteins on transcription of the lacZ reporter gene and comparing the results to the amounts of reporter gene transcription produced by the negative and positive control strains. Two exponential phase cultures of each strain were prepared, one growing under inducing conditions and the other growing under non-inducing Then, we determined the specific activity of β-galactosidase in each culture and calculated the induction ratio for each strain (the ratio of the amount of β-galactosidase activity produced under inducing conditions to the amount of activity produced under non-inducing conditions) The induction ratio for the test cultures co-expressing the LexA-SoxS and B42AD-σ⁷⁰ R4 fusion proteins was 14, which is about sevenfold higher than the ratio obtained with the three negative control strains and about half of that obtained with cultures of the positive control strain that co-expressed the SoxS and αCTD fusion proteins (Table 3). Thus, the extent of activated transcription of the reporter gene by the co-expression of the LexA-SoxS and B42AD-σ⁷⁰R4 fusion proteins relative to that of the negative and positive control strains correlates well with the respective plate phenotypes (Table 3). More importantly, both the qualitative and the quantitative tests of protein-protein interactions indicate that SoxS and σ⁷⁰ R4 can interact in the absence of complete, holo-RNAP and without the specific binding of SoxS to soxbox DNA. Accordingly, SoxS and σ⁷⁰ R4 appear to be able to interact in solution, i.e., off-DNA. In addition, the results indicate that the interaction between SoxS and σ⁷⁰ R4 is only half as strong as the SoxS-αCTD interaction.

Effects of positive control mutations of SoxS on the interaction between SoxS and σ⁷⁰ R4 in the Y2H system

As mentioned above, the characterization of our library of single alanine substitutions of SoxS enabled us to identify two classes of mutants that bind DNA normally but are defective in transcription activation, i.e., positive control mutants ⁶³. The five class I/II mutants reduce transcription activation of both class I and class II SoxS-dependent promoters while the five class II mutants are defective only in activation of class II promoters. All of the positive control mutants bind and bend target DNA normally but they are leaky with respect to their effects on transcriptional activation of SoxS promoters: substitution K30A reduces transcriptional activation by threefold while the remaining positive control mutants are even more leaky, reducing activation by only about twofold ⁶³.

The availability of the positive control mutants provided several important benefits to our analysis of the interaction of SoxS and σ⁷⁰ R4 in the Y2H system. First, if mutants can be found that disrupt the interaction, then the interaction occurring in yeast is likely to be specific. Second, finding substitutions that prevent the interaction will identify amino acids important to complex formation in yeast. Third, since positive control mutants of SoxS that reduce complex formation in the Y2H system would do so by interfering with protein-protein contacts between SoxS as bait and σ⁷⁰ R4 as prey, the defect in transcription activation conferred by these substitutions in E. coli would almost certainly be due to an alteration in protein-protein interaction between SoxS and σ⁷⁰ R4. Lastly, if the SoxS-σ⁷⁰ R4 interaction that occurs in solution in yeast is disrupted by positive control mutants of SoxS, the corresponding amino acids in wild type SoxS are likely to be involved in forming the binary complexes between SoxS and RNAP that play a critical role in the pre-recruitment mechanism.

Since the interaction of SoxS with σ⁷⁰ R4 could use the class I/II surface, the class II surface or both surfaces, and since the individual alanine substitutions are leaky, we used the same three pGILDA plasmids carrying multiple positive control substitutions for this analysis as we used previously in our characterization of the interaction between SoxS and the αCTD ⁶⁰. The SoxS protein in plasmid pGILDA-SoxS-class I/II⁻ carries alanine substitutions D9A, K30A, S31A and V45A of the class I/II surface and was shown to be totally defective in transcription activation of both class I and class II promoters in E. coli ⁶⁰. The SoxS protein in pGILDA-SoxS-class II⁻ carries substitutions F74A, D75A, M78A, D79A and Q85A and is totally defective in transcription activation of class II promoters in E. coli but activates transcription of class I promoters nearly as well as wild type SoxS ⁶⁰. Plasmid pGILDA-SoxS-classI/II⁻ + class II⁻ carries the nine substitutions listed above plus K5A of the class I/II domain; it is totally defective in transcription activation of both classes of promoter in E. coli ⁶⁰. Importantly, the three multiply mutant proteins display the same over-expression toxicity in E. coli as wild type SoxS, an indication that they retain normal DNA binding in vivo ⁶³, and as purified proteins they bind and bend DNA in vitro as well as wild type ⁶⁰. These latter two tests indicate that the conformation of the proteins with the multiple positive control substitutions is like that of wild type SoxS ⁶⁰.

Each of the three pGILDA-SoxS plasmids carrying multiple positive control substitutions of SoxS were co-transformed with pB42AD-σ⁷⁰ R4 into strain EGY48[p8oplacZ]. The plate phenotypes of the respective strains and the induction ratios of β-galactosidase expression from the reporter gene in the three strains were determined as described above. The results are shown in Table 3 and are compared to the data obtained with wild type SoxS. No interaction with σ⁷⁰ R4 was observed when the SoxS moiety carried the five substitutions of the class I/II surface and the five substitutions of the class II surface, as evidenced by formation of white-colored colonies and an induction ratio as low as that obtained with the two empty vectors. Thus, one or both of these surfaces of SoxS is required for the interaction with σ⁷⁰ R4. The properties conferred by the other two mutant LexA-SoxS proteins enabled us to determine that only the class I/II surface is required for the SoxS-σ⁷⁰ R4 interaction. Since co-expression of pGILDA-SoxS-class I/II⁻ with pB42AD-σ⁷⁰ R4 produced white-colored colonies and an induction ratio of 2, the class I/II surface is required for the interaction. In contrast, co-expression of pGILDA-SoxS-class II⁻ with pB42AD-σ⁷⁰ R4 had no effect on the interaction, since the colonies formed were of the same light blue color as with wild type LexA-SoxS and the induction ratio was the same as with wild type SoxS.

Thus, these data show that the interaction between SoxS and σ⁷⁰ R4 is specific, because it is disrupted by a set of positive control mutants. In addition, the interaction depends on amino acid residues within the class I/II surface but not those within the class II surface; in turn, these data indicate that the substitutions of the class I/II surface that reduce transcription activation in E. coli do so by disrupting protein-protein interactions. Moreover, since interactions in the Y2H system occur in solution and in the absence of specific DNA binding, the ability of positive control substitutions of the class I/II surface to disrupt the interaction suggests that in E. coli the reduction in transcription activation by these positive control substitutions could be due to the disruption of interactions between SoxS and the σ⁷⁰ R4 region of RNAP that occur in solution and in the absence of specific DNA binding. Accordingly, the positive control mutants could be interfering with the formation of the SoxS-RNAP binary complexes required for activation by pre-recruitment. Remarkably, positive control mutants of SoxS have the same effect on the interaction of SoxS with the αCTD ⁶⁰ as with σ⁷⁰ R4: positive control mutants of the class I/II surface disrupt the interactions with both the αCTD and σ⁷⁰ R4 whereas positive control mutants of the class II surface have no effect. Thus, by this criterion, the same class I/II surface on SoxS makes protein-protein contact with both the αCTD and σ⁷⁰ R4. Moreover, the existence of the two complexes in yeast suggests that the same two, mutually exclusive binary complexes form in E. coli, one through interaction between the class I/II surface and the αCTD and the other through interaction between the class I/II surface and σ⁷⁰ R4. Logically, one binary complex might be used to find and bind to class I promoters while the other finds and binds to class II promoters.

The failure of the positive control mutants of the class II surface to disrupt the SoxS-σ⁷⁰ R4 interaction in the Y2H system is also important with respect to the steps in transcription activation in E. coli. Since the class II mutants reduce transcription activation of class II promoters in E. coli but have no effect on the SoxS-σ⁷⁰ R4 interaction in yeast, and since epistatic interactions were found in E. coli between some of these class II mutants and mutants of σ⁷⁰ R4, we infer that the interactions between σ⁷⁰ R4 and the class II surface of SoxS occur on SoxS-activatable promoter DNA.

Determination of whether single alanine substitutions of σ⁷⁰R4 that reduce transcription activation in E. coli interfere with the interaction of SoxS and σ⁷⁰ R4 in the Y2H system

Seven amino acid residues of σ⁷⁰ R4 reduce SoxS-dependent transcription activation at the class II or class II* promoters in E. coli (Table 1). Although epistatic interactions in E. coli were found for only five of the σ⁷⁰ R4 substitutions (Table 2), we wanted to determine which of the seven substitutions that reduce transcription activation affect the SoxS-σ⁷⁰ R4 interaction in the Y2H system. Finding such an effect in yeast would identify σ⁷⁰ R4 substitutions that are necessary for SoxS-σ⁷⁰ R4 interactions in solution in E. coli and thus could be involved in the formation of binary complexes that would lead to transcription activation at one or more class II or class II* promoters.

To conduct this experiment, we first introduced the seven substitutions of σ⁷⁰ R4 into prey plasmid pB42AD-σ⁷⁰ R4 by site-directed mutagenesis and we transformed these plasmids into reporter strain EGY48[p8oplacZ] along with wild type bait plasmid pGILDA-SoxS. We then determined their plate phenotypes as described above. None of the plate phenotypes conferred by the respective mutant strains differed significantly from that of the strain co-expressing the wild type fusion proteins, i.e., all formed light blue-colored colonies, although some were slightly lighter and some were slightly darker than the strain with wild type σ⁷⁰ R4.

To determine whether the σ⁷⁰ R4 substitutions have a quantitative but not a qualitative effect on the SoxS-σ⁷⁰ R4 interaction, we measured β-galactosidase activity under inducing conditions. Figure 3 shows that five of the σ⁷⁰ R4 substitutions reduced transcription of the reporter to levels significantly below that of the wild type while substitutions K593A and R603A enhanced transcription of the reporter by 76% and 36%, respectively. The effects of the mutants were not due to differences in the abundance of the respective σ⁷⁰ R4 proteins, as shown by qualitative western blots that detect the HA tag appended to B42AD (Figure 3). These data indicate that in E. coli amino acids R596, R599, E605, R608 and D612 of σ⁷⁰ R4 are probably required for interaction with SoxS in solution, i.e., in binary complex formation. Because K593A and R603A produce enhanced interaction in yeast, they probably relieve inhibitory interactions conferred by the wild type amino acid residues. Nonetheless, the fact that they alter the interaction between SoxS and σ⁷⁰ R4 in the Y2H system suggests that they too can function in solution. As discussed below, amino acid substitutions that produce values higher than that of wild type give inconclusive results in epistasis tests.

Figure 3.

Effects in the Y2H system of single alanine substitutions of σ⁷⁰ R4 that cause a defect in transcription activation at SoxS-dependent promoters in E. coli. Cultures of strain EGY48[p8oplacZ] carrying pGILDA-SoxS and pB42AD-σ⁷⁰ R4 with wild type and mutant alleles of σ⁷⁰ R4 were grown under inducing conditions and assayed for β-galactosidase activity as described in Materials and Methods. The specific activity of the strain expressing wild type σ⁷⁰ R4 was taken as 100% and the activity values of the strains expressing the mutant alleles are given relative to wild type. As depicted at the bottom of the figure, cell extracts of each culture were subjected to western blotting with antibodies against the HA tag fused to B42AD-σ⁷⁰ R4 and antibodies against β-tubulin, which served as an internal loading control.

Determination of whether single alanine substitutions of the class I/II surface of SoxS that are epistatic in E. coli to substitutions of σ⁷⁰R4 interfere with the interaction of SoxS and σ⁷⁰ R4 in the Y2H system

We wanted to conduct epistasis tests in the Y2H system to identify amino acid residues of the class I/II surface of SoxS that interact in solution with amino acid residues of σ⁷⁰ R4. Since the work described in Table 3 only tested multiple positive control mutants in the Y2H system, we needed to introduce into pGILDA single alanine substitutions that were shown in the epistasis experiments conducted in E. coli to have an interacting partner in σ⁷⁰ R4. We only used substitutions of the class I/II surface of SoxS and not those of the class II surface because the data in Table 3 show that the introduction of five substitutions of the class II surface into prey fusion B42AD-σ⁷⁰ R4 has no effect on the interaction between SoxS and σ⁷⁰ R4 in the Y2H system. Accordingly, we used site-directed mutagenesis to introduce the single alanine substitutions D9A, K30A and S31A into bait plasmid pGILDA-SoxS. We then co-transformed these singly mutant plasmids along with prey plasmid pB42AD-σ⁷⁰ R4 into reporter strain EGY48[p8oplacZ]. After determining that the single positive control mutants have little effect on the plate phenotype, we then determined their effect on activated transcription of the reporter. We also conducted qualitative western blots with anti-SoxS antibodies and determined that the positive control substitutions do not substantially alter the abundance of SoxS in yeast.

As shown in Figure 4, substitutions K30A and S31A reduce activated transcription to 78% and 62% of wild type SoxS, respectively, and thus they play a role in the protein-protein interaction between SoxS and σ⁷⁰ R4 in solution. Moreover, they are candidates to be tested for epistasis in the Y2H system. On the other hand, D9A enhances activated transcription by about 50%, probably by relieving an inhibitory interaction conferred by D9. As discussed below, conclusive results in epistasis tests cannot be obtained if one of the two substitutions produces a level of activated transcription significantly above that of wild type. Nonetheless, since D9A affects the interaction in the Y2H system between SoxS and σ⁷⁰ R4, we infer that in E. coli it likely functions in solution (and perhaps also on-DNA).

Figure 4.

Effects in the Y2H system of single alanine substitutions of the class I/II surface of SoxS that confer defects in transcription activation at SoxS-dependent promoters and which display epistasis in E. coli with alanine substitutions of σ⁷⁰ R4. Cultures of strain EGY48[p8oplacZ] carrying pB42AD-σ⁷⁰ R4 and pGILDA-SoxS with wild type and mutant alleles of SoxS were grown under inducing conditions and assayed for β-galactosidase activity as described in Materials and Methods. The specific activity of the strain expressing wild type SoxS was taken as 100% and the activity values of the strains expressing the mutant alleles are given relative to the wild type. Cell extracts of each culture were subjected to western blotting with affinity-purified antibodies against SoxS and antibodies against β-tubulin, which served as an internal loading control.

Epistasis in the Y2H system

Having identified single alanine substitutions of SoxS and single alanine substitutions of σ⁷⁰ R4 that interact epistatically in E. coli and also affect the interaction between SoxS and σ⁷⁰ R4 in the Y2H system, we wanted to see whether the substitutions display epistasis in the Y2H system. If so, we could argue that an epistatic interaction that occurs in both yeast and E. coli likely occurs off-DNA in E. coli, i.e., in solution. Accordingly, our candidates for epistasis tests in the Y2H system were the single alanine substitutions of the class I/II surface of SoxS and the single alanine substitutions of σ⁷⁰ R4 that participated with them in epistatic interactions in E. coli (Table 2) and which conferred defects in β-galactosidase expression from the reporter gene of the Y2H system (Table 3 and Figure 4).

In E. coli, epistasis involving positive control mutants of the class I/II surface was obtained with S31A + K593A, D9A + R599A, D9A + R608A and K30A + D612A. However, in the Y2H system, substitutions D9A of SoxS and K593A of σ⁷⁰ R4 both enhance activated transcription of the lacZ reporter to levels considerably above that of the strain co-expressing the wild type fusion proteins and as such epistasis with them cannot be evaluated. This is because the activity produced by bait and prey substitutions acting non-epistatically, with one, e.g., D9A, producing an activity higher (e.g., 148%) than that of the wild type activity, and the other, e.g., R599A, producing an activity below (e.g., 72%) that of wild type, will be greater than the least active of the two (i.e., 107% vs. 72%); as such, a non-epistatic interaction cannot be distinguished from an epistatic interaction. Accordingly, we only conducted epistasis tests with the combination of LexA-SoxS-K30A and B42AD-σ⁷⁰ R4-D612A; as a negative control, we also conducted the test on the combination of LexA-SoxS-S31A and B42AD-SoxS-D612A that was expected to be non-epistatic. Since the single mutants of SoxS and σ⁷⁰ R4 did not produce a significant effect on the plate phenotype, we only assessed epistasis by assay of β-galactosidase, a measure of activated transcription of the lacZ reporter gene.

Figure 5 shows that K30A of SoxS and D612A of σ⁷⁰ R4 are indeed epistatic to one another in yeast, since the β-galactosidase activity produced by the strain co-expressing the bait and prey fusions carrying the K30A and the D612A substitutions, respectively, is not statistically different from the amount of activity produced by the strains co-expressing one wild fusion (e.g., LexA-SoxS) and one mutant fusion (e.g., B42AD-σ⁷⁰ R4-K30A). As expected, S31A of SoxS and D612A of σ⁷⁰ R4 are not epistatic to one another, since the β-galactosidase activity produced by the strain co-expressing the putative non-epistatic substitutions, S31A and D612A, respectively, produced a level of β-galactosidase that is statistically lower than either of the mutant fusions in combination with a wild type fusion (Figure 5). Thus, among the ten epistatic interactions in E. coli obtained between substitutions of SoxS and substitutions of σ⁷⁰ R4 (Table 2), only the interaction between K30A of SoxS and D612A of σ⁷⁰ R4 (e.g., 72%) also occurs in the Y2H system and therefore only this specific interaction has been demonstrated to occur in solution, i.e., off-DNA. As stated above, epistasis in E. coli that involves SoxS positive control mutants residing in the class II surface probably represents interactions that occur on promoter DNA, but substitutions of SoxS (e.g., D9A) or σ⁷⁰ R4 (e.g., K593A) that reduce transcription in E. coli but enhance reporter gene transcription in yeast probably represent interactions that occur off-DNA in E. coli.

Figure 5.

Representative epistatic and non-epistatic interactions in the Y2H system between alanine substitutions of SoxS and alanine substitutions of σ⁷⁰ R4 that individually confer a defect in transcription activation of the lacZ reporter in the Y2H system. Cultures of strain EGY48[p8oplacZ] carrying plasmid pGILDA-SoxS with wild type and mutant alleles of SoxS and plasmid p42AD-σ⁷⁰ R4 with wild type and mutant alleles of σ⁷⁰ R4 were grown under inducing conditions and assayed for β-galactosidase activity as described in Materials and Methods. The specific activity of the strain expressing wild type SoxS and wild type σ⁷⁰ R4 was taken as 100% and the activity values of the strains expressing the various mutant alleles are given relative to the wild type. An epistatic interaction was observed between σ⁷⁰ substitution D612A and SoxS K30A, a mutant of the class I/II positive control surface. A two tailed t-test was carried out to determine whether the values for the double mutants are significantly different from the values for the corresponding single mutants. The value for the SoxS-K30A + σ⁷⁰ D612A double mutant was not significantly different than that of the single mutant SoxS + σ⁷⁰ D612A (p>0.05) and thus this interaction is epistatic; in contrast, the value for the SoxS-S31A + σ⁷⁰ D612A double mutant was significantly different than that of the single mutant SoxS + σ⁷⁰ D612A (p<0.05) and thus this interaction is non-epistatic.

It is important to note that even though only one epistatic interaction was found that occurs in both E. coli and yeast (K30A of SoxS with D612A of σ⁷⁰ R4), this single example demonstrates that SoxS can make a specific protein-protein interaction with an amino acid of σ⁷⁰ R4 in solution, i.e., off-DNA. Moreover, we argue that although only one specific interaction has been identified, we still know that SoxS interacts with σ⁷⁰ R4 (Table 3) and full-length σ⁷⁰ (see below) in solution, that this interaction requires amino acids of the class I/II of SoxS but not the class II surface (Table 3), that amino acid residues R596, R599, R605, R608 and R612 of σ⁷⁰ R4 are required for the interaction (Figure 3), and that substitutions K593A and R603A enhance the strength of the interaction. In addition, we argue that even though the single epistatic interaction that has been shown to occur in solution does so in activating transcription from the atypical class II* zwf promoter (Table 2), the fact that amino acid substitutions K593A, R596A, R599A and R603A of σ⁷⁰ R4 that reduce transcription activation of the fumC and micF promoters in E. coli also affect transcription activation of reporter gene expression in the Y2H system indicates that these residues play a role in binary complex formation in E. coli. As such, the data presented here make it clear that SoxS forms a binary complex off-DNA with σ⁷⁰ R4 in yeast and with full-length σ⁷⁰ in vitro (see below). Moreover, we point out that experiments supporting the conclusion that a given interaction occurs off-DNA do not rule out the possibility that the same interaction also occurs on-DNA.

Protein-protein interactions between SoxS and full-length σ⁷⁰ demonstrated by AIA

The Y2H experiments (Figures 3–5) and the AIA (Figure 2) presented above have demonstrated specific protein-protein interactions between SoxS and σ⁷⁰ R4 that occur in solution. Since the results could be an artifact of using σ⁷⁰ R4 and not full length σ⁷⁰ or because the results were obtained in yeast, a heterologous system, we conducted AIA using full-length σ⁷⁰. Obtaining specific protein-protein interactions with SoxS and σ⁷⁰ would not only rule out the possibility that the Y2H results are artifactual but would also provide independent verification of the interactions observed with σ⁷⁰ R4 and the inferences drawn from them. In particular, if SoxS and full-length σ⁷⁰ can be shown by AIA to make specific protein-protein contacts, then additional evidence will have been provided that the interactions can occur in solution and in the absence of SoxS-specific promoter DNA.

The following AIA experiments were carried out like those shown in Fig. 2 except that full-length his₆-σ⁷⁰ expressed from plasmid pHMK-his₆-σ⁷⁰ was used and it was unnecessary to run separate western blots to detect his₆-σ⁷⁰ and SoxS because they migrate to different positions during SDS-PAGE. Figure 6 shows that SoxS was present in the eluate when his₆-σ⁷⁰ was released from the Ni-NTA resin by treatment with a high concentration of imidazole. Thus, his₆-σ⁷⁰ immobilized on the Ni-NTA resin is able to capture wild type SoxS from an E. coli cell-free extract. Moreover, just as SoxS is able to interact with σ⁷⁰ R4 in the Y2H system, so too is SoxS present in a crude cell extract able to interact with the full-length σ⁷⁰ in AIA.

Figure 6.

AIA of the protein–protein interactions between SoxS and full-length σ⁷⁰ The immobilization of wild-type σ⁷⁰ on Ni-NTA resin, the application of cell-free extracts containing wild type and SoxS proteins with multiple positive control substitutions, the recovery of the FT, W3 and EL fractions and western blotting of the fractions with antibodies directed against SoxS and the his₆ tag of σ⁷⁰ were carried out as described in Materials and Methods.

As a control for the specificity of the interaction between his₆-σ⁷⁰ and SoxS, we conducted AIA with the same three plasmids carrying multiple positive control substitutions of SoxS as were used in the Y2H assays (Table 3), i.e., (i) SoxS with the ten positive control substitutions that completely inactivate the class I/II surface and the class II surface; (ii) SoxS with the four positive control substitutions that completely inactivate the class I/II surface; and (iii) SoxS with the five positive control substitutions that completely inactivate the class II surface. The SoxS protein with the multiple mutations of the class I/II and class II surfaces was unable to interact with immobilized his₆-σ⁷⁰ (data not shown) as was the protein containing substitutions only of the class I/II surface (Figure 6). Importantly, the quintuple substitutions of the class II surface of SoxS did not interfere with the interaction (Figure 6). Thus, these results obtained with AIA fully agree with the corresponding assays conducted in the Y2H system. Moreover, they indicate that the class I/II domain is essential to the interaction with σ⁷⁰ R4 even when it is part of full-length σ⁷⁰. Indeed, when we prepared a mutant of his₆-σ⁷⁰ that carries two consecutive stop codons after position 529 and thus lacks region 4, i.e., his₆-σ⁷⁰₍₅₂₉₎, we found that after immobilization on Ni-NTA resin, it was unable to capture wild type SoxS in AIA (data not shown). This shows that σ⁷⁰ R4 is necessary for the interaction in the AIA with full-length σ⁷⁰ and the AIA experiment in which σ⁷⁰ R4 was able to capture wild type SoxS (Figure 2) shows that σ⁷⁰ R4 is sufficient for the interaction. Unfortunately, because of the relative weakness of the interaction between SoxS and σ⁷⁰ R4, e.g., compared to the interaction of SoxS and αCTD, the effects of single alanine substitutions of σ⁷⁰ R4 that reduce transcription in the Y2H system are too small to produce a significant effect in AIA (data not shown). Nonetheless, the data presented above demonstrate that SoxS makes protein-protein interactions with σ⁷⁰ R4 and that the interaction requires the amino acids residues of the class I/II surface.

Concluding remarks

The data presented here and in the previous work by Shah and Wolf ⁶⁰ are consistent with the pre-recruitment hypothesis in that they show that SoxS can indeed form binary complexes with RNAP in solution and in the absence of specific DNA binding. Of course, just because a binary complex between SoxS and RNAP can form in solution does not mean that it does form and then scans the chromosome for SoxS-dependent promoters. Hence, it will be interesting to determine whether activators known to function by recruitment, e.g., MelR ⁷⁹, can also form binary complexes with RNAP in solution.

The data presented here and in the paper by Shah and Wolf ⁶⁰ also indicate that SoxS makes two protein-protein contacts in solution with RNAP, one through an interaction with the αCTD and the other through an interaction with σ⁷⁰ R4. However, since the same class I/II surface of SoxS is used for both interactions, the two interactions must lead to two mutually exclusive binary complexes. Thus, we infer that if SoxS functions by pre-recruitment, then one binary complex, e.g., the complex formed by the interaction with the αCTD, would scan the chromosome for class I SoxS-dependent promoters, while the other binary complex, e.g., the complex formed by the interaction with σ⁷⁰ R4, would scan the chromosome for class II and class II* SoxS-dependent promoters. (We chose this partitioning because mutants of σ⁷⁰ R4 have no effect on activation of class I promoters.)

Although it now appears that SoxS forms two different, off-DNA, binary complexes with RNAP as an initial step in transcription activation and that amino acids of SoxS, the αCTD and σ⁷⁰ R4 participating in this initial step have been identified, determining the nature of the complexes in the subsequent steps leading to transcription initiation remains a challenge for future work.

Materials and Methods

Plasmid library of mutants of σ⁷⁰ R4

Table 1 shows the library of 16 single alanine substitutions of σ⁷⁰ R4 that was used for this study; the substitutions lie in the region of σ⁷⁰ between positions 590 and 613. The gene encoding σ⁷⁰ amino acids from positions 8 to 613 and its mutant derivatives was carried in plasmid pGEX-2T-σ⁷⁰ and was obtained from C.A. Gross via S.M. Egan ^{69; 70} . After sequencing σ⁷⁰ R4 of each member of the library at the UMBC DNA Sequencing Facility and finding several incorrect clones, we received replacements from V. A. Rhodius and W. Ross.

Determining the effect of single alanine substitutions of σ⁷⁰ R4 on SoxS-dependent transcription in E. coli

The σ⁷⁰ R4 library of single alanine substitutions was transformed into Δ(araBAD) 714 derivatives of strain N7840 ( Δ(argF-lac)U169 Δmar carrying transcriptional fusions of lac to the zwf, fpr, fumC and micF promoters ⁶³ and plasmid pBAD33-his₆-SoxS; the transformants were recovered by plating on LB plates containing ampicillin (100 µg/ml) and chloramphenicol (25 µg/ml). σ⁷⁰ is expressed constitutively from the pGEX-2T plasmids while the expression of his₆-SoxS is under control of the arabinose-inducible araBAD promoter. Plasmid pBAD33-his₆-SoxS was prepared by subcloning the XbaI/HindIII fragment of high-copy plasmid pBAD18-his₆-SoxS ⁶³, which contains his₆-SoxS, into the low-medium copy plasmid pBAD33 ⁸⁰ that had been digested with the same two enzymes. Triplicate cultures of the four fusion strains carrying the respective members of the σ⁷⁰ R4 library and pBAD33-his₆-SoxS were grown overnight at 37 °C in LB medium containing the two antibiotics in 96-well polypropylene blocks ⁸¹. Each overnight culture was diluted 1:200 in triplicate into the same medium and grown until the A₆₀₀ of the cultures reached 0.1–0.2, at which point SoxS synthesis was induced with 0.02% arabinose, a concentration previously determined to be sufficient for maximum induction ⁶³. Then, when the A₆₀₀ of the cultures reached 0.6–0.8, samples were removed from each culture and the β-galactosidase activity was determined by our previously described high-throughput method ⁸¹. At least three independent experiments were carried out on each strain and the average values in Miller units for each experiment were averaged. The average values obtained with the mutants are presented as a percentage of the value obtained with the corresponding strain carrying the wild type plasmid. Standard errors of the means were calculated.

Epistasis tests in E. coli

The method described above for determining the effect of single alanine substitutions of σ⁷⁰ R4 on SoxS-dependent activation of transcription of a specific promoter-lac fusion was also used for identifying epistatic interactions between σ⁷⁰ R4 mutants and positive control mutants of SoxS. Thus, in these experiments, the strains harbored pBAD33-his₆-SoxS carrying a positive control mutant of SoxS as well as pVR-σ⁷⁰ carrying either the wild type or a mutant allele of σ^{70 82}. Since all of the single alanine substitutions of SoxS that confer a positive control phenotype were prepared in pBAD18-his₆-SoxS ⁶³, we subcloned each substitution into pBAD33 as an XbaI/HindIII fragment as described above. Assays of β-galactosidase activity were done as described above and epistatic interactions were identified as described in the text.

Interaction between wild type and positive control mutants of SoxS and σ⁷⁰ R4 or full-length σ⁷⁰ as assessed in vitro by AIA

AIA was carried out by the method used previously for studying the interaction between wild type and mutant SoxS proteins and the full-length a subunit of RNAP ⁶⁰. In the first set of AIA experiments, σ⁷⁰ R4 was expressed with an N-terminal his₆ tag from plasmid pSKB2-σ⁷⁰ R4, kindly provided by L. F. Westblade. Since expression of his₆-σ⁷⁰ R4 is under control of a phage T7 promoter, the plasmid resided in E. coli strain BL21 (λDE3) and was maintained by growth in the presence of kanamycin (50 µg/ml); expression of T7 RNAP and thus his₆-σ⁷⁰ R4 was induced by treatment of the culture with 1 mM IPTG when the A₆₀₀ of the 50 ml culture reached 0.2. The rest of this phase of the experiment with his₆-σ⁷⁰ R4 was carried out as described previously ⁶⁰, except that the cell extract containing his₆-σ⁷⁰ R4 was incubated for 90 min with 250 µl of Ni-NTA agarose resin (Qiagen).

In the second phase of the AIA with his₆-σ⁷⁰ R4, the cell extracts containing untagged, wild type SoxS and untagged SoxS carrying multiple alanine substitutions of the class I/II and/or class II surfaces were prepared and incubated with the Ni-NTA resin to which his₆-σ⁷⁰ R4 had been bound as described previously ⁶⁰, except that 50 ml cultures were used and buffer A contained 50 mM NaCl. Unbound material was removed by centrifugation and saved as the flow-through (FT) fraction. The resin was then washed three times with the last wash being saved as a control (W3). The resin was then incubated with buffer containing 500 mM imidazole to release N-his₆-σ⁷⁰ R4 and any proteins bound to it, i.e., SoxS; this elution fraction (EL) was also saved. Since SoxS and his₆-σ⁷⁰ R4 have about the same MW, ∼13 kD, and thus cannot be easily separated by SDS-PAGE and identified by western blotting, duplicate aliquots of the FT, W3 and EL samples were loaded onto two halves of the same 15% gel and after electrophoresis, the gel was subjected to western blotting as described ⁶⁰. After transfer, the membrane was cut in half and one side was probed with mouse monoclonal antibodies directed against the his₆ tag (Sigma) while the other half was probed with rabbit polyclonal anti-SoxS serum ¹⁸. The blots were then treated with secondary antibodies and developed as described ⁶⁰. The resulting chemifluorescent signals were detected with a Molecular Dynamics Storm 860 PhosphorImager in fluorescence mode with a blue (450 nm) laser.

In the second set of experiments, plasmid pHMK-his₆-σ⁷⁰ encoding full-length σ⁷⁰, kindly provided by R.R. Burgess ⁸³, was transformed into strain BL21(λDE3) and the AIA was carried out as described above for the experiments with his₆-σ⁷⁰ R4. As a negative control, QuikChange mutagenesis of pHMK-his₆-σ⁷⁰ was used to introduce dual stop codons (TAGTGA) after codon 529 of σ⁷⁰. AIA was carried out with this construct as described above.

Y2H system

The Y2H system used here was the same as that used in our previous work except that the prey was σ⁷⁰ R4 rather than the αCTD ⁶⁰. Thus, the parental plasmids for the bait and prey fusions were pGILDA and pB42AD, respectively, and the plasmid carrying the lacZ reporter was p8oplacZ; all were obtained from Clonetech, as was strain EGY48 (MATα his3 trp1 ura3 LexA_op(x6)-LEU2), the strain of Saccharomyces cerevisiae used for all the Y2H experiments, and pop8lacZ, the plasmid encoding the gene that reports the transcriptional activity conferred by the bait and prey fusions.

As described by Shah and Wolf ⁶⁰, plasmid pGILDA-SoxS expresses the bait as an in-frame fusion of the DBD of LexA to the N-terminus of SoxS. The following plasmids with multiple positive control mutations served as specificity controls for the interaction of SoxS with σ⁷⁰ R4: pGILDA-SoxS-Class I/Class II⁻ + Class II⁻ with ten alanine substitutions of the class I/II and class II surfaces; pGILDA-SoxS-Class I/Class II⁻ with four alanine substitutions in the class I/II surface; and pGILDA-SoxS-Class II⁻ with five alanine substitutions in the class II surface ⁶⁰. The single positive control substitutions D9A, K30A and S31A were introduced into pGILDA-SoxS by the same methods used to prepare the plasmids with the multiple positive control substitutions.

Plasmid pB42AD-σ⁷⁰ R4 expresses the prey as an in-frame fusion between the B42AD activation peptide and σ⁷⁰ R4. The DNA segment encoding σ70 R4 (amino acids 530–613) of plasmid pGEX-2T-σ⁷⁰ was amplified by PCR with sense primer Sσ⁷⁰R4–5’-CTTAGTGAATTCCTGCCGCTGGATTCTGCGACC-3’ and antisense primer ASσ⁷⁰R4–5’-TACCGTCTCGAGCGATTAATCGTCCAGGAA-3’. The primers add-on restriction sites EcoRI and XhoI, respectively, whose recognition sequences are underlined. PCR was carried out for 30 cycles of 95 °C for 1 min, 55 °C for 30 sec and 72°C for 1 min. The PCR product was purified with a PCR cleanup kit (Qiagen) and digested with restriction enzymes EcoR and XhoI (New England BioLabs). The digestion product was subjected to gel electrophoresis and the DNA was extracted and purified with a Wizard Miniprep kit (Promega). The purified fragment was ligated to plasmid pB42AD digested with the same two enzymes and an aliquot of the ligation mixture was used to transform strain DH5α, which was plated onto LB agar medium containing ampicillin (50 µg/ml) and incubated at 37 °C. Colony PCR with vector-specific primers was used to identify clones containing insert fragments of the correct size. Then, plasmids were purified from these clones with Wizard minicolumns (Promega) and the DNA sequence of the candidates was determined. A clone with the correct sequence was named pB42AD-σ⁷⁰ R4.

Six of the seven single alanine substitutions of σ⁷⁰ R4 that reduce transcription activation in E. coli (K593A, R596A, R599A, R603A, E605A and R608A) were also PCR-amplified with the same primers as above from the members of the pGEX2T-σ⁷⁰ library carrying the respective mutations; they were then cloned and their sequences determined as described above. Preparation of pB42AD-σ⁷⁰ R4 carrying the D612A substitution required a different antisense primer, ASσ⁷⁰R4-D612A-5’-TACCGTCTCGAGCGATTAATCAGCCAGGAA-3’, but otherwise the cloning, identification, and sequencing used the same procedures as for the other mutants.

The procedures for growth and manipulation of yeast were those described in Clonetech’s Matchmaker system manual as modified by Shah and Wolf ⁶⁰. S. cerevisiae strain EGY48 was grown at 30 °C in SD medium, a complete minimal medium prepared to contain all the nutrients required for growth of yeast or to lack specific nutrients to allow selection for nutritional markers. Thus, EGY48 carrying the reporter plasmid pop8lacZ and plasmids pGILDA and pB42AD and their various wild type and mutant bait and prey derivatives were grown in SD medium lacking uracil, histidine and tryptophan (“triple drop-out medium”) and containing 2% (w/v) glucose as the carbons source for non-inducing conditions or 2% (w/v) gal plus 1% (w/v) raf for growth under inducing conditions.

Determination of plate phenotypes and assay of β-galactosidase in the Y2H system

The method for determining on plates whether co-expression of a given combination of bait and prey fusions produced a functional activator able to activate expression of the lacZ reporter was carried out as described previously by Shah and Wolf ⁶⁰, as was quantification of the extent of lacZ reporter gene transcription under inducing and non-inducing conditions. The plate phenotypes were determined on SD/gal/raf/-uracil/-histidine/-tryptophan/X-gal plates. Quantification of reporter gene expression following co-expression of the various bait and prey fusions was done by determining the specific activities of β-galactosidase in Miller units from duplicate assays of samples taken from duplicate cultures, with the final average values calculated from at least three experiments conducted on different days. The induction ratios were calculated by dividing the average specific activity values obtained for cells grown under inducing conditions by the average specific activity values obtained for cells grown under non-inducing conditions.

Statistical Analyses

The analysis toolkit of Microsoft Excel was used to carry out two-tailed t-tests of the significance of differences in the effects of mutations on transcriptional activation. For example, a t-test was conducted to determine whether the effect of a double mutant combination of putative epistatic partners, e.g., σ⁷⁰ D612A + SoxS K30A, differs significantly from the effect of the single mutant of the pair that conferred the greatest defect e.g., D612A. The null hypothesis was that the effects of the single mutant and the double mutant combination are not significantly different at a confidence interval of 95% (α=0.05). For epistatic interactions, the calculated value was p>0.05 and therefore the null hypothesis was accepted. For the non-epistatic interactions, the calculated value was p<0.05 and therefore the null hypothesis was rejected.

Western blotting of extracts prepared from strains used for reporter gene assays in the Y2H system

Western blots were used to determine qualitatively whether the effect of mutations of SoxS or σ⁷⁰ R4 that either reduced or enhanced reporter gene expression in the Y2H system was due to a corresponding alteration in the abundance of the mutant protein compared to the wild type. The procedure employed was that described by Vallabhaneni and Farabaugh ⁸⁴. Cultures were grown under conditions identical to those used when the various strains were grown for determination of the effect of mutations of SoxS and σ⁷⁰ R4 on reporter gene expression. Thus, an overnight culture of each strain grown in non-inducing medium was diluted 1:200 into 10 ml of inducing medium and the culture was grown to A₆₀₀∼0.8. The cells from each culture were collected by centrifugation at 5000g, resuspended in 1 ml water, transferred to a 1.5 ml microfuge tube and collected again by centrifugation. After adding acid-washed glass beads (425–600 mm) (Sigma-Aldrich) in an amount approximately equal to the volume of the cell pellet, the mixture was resuspended in 100 µl of lysis buffer (25mM Tris pH=7.5, 1mM DTT, 0.1% Triton X-100, 0.15M NaCl, 1mM EDTA, and a protease inhibitor cocktail (10 µl/ml) (Sigma-Aldrich). The mixture was subjected to 7 rounds of vortexing for 30 sec with rests of 30 sec on ice between each treatment. The cell lysate was cleared by centrifugation at 16000g for 5 min at 4 °C in a Heraeus Instruments Biofuge with a PP2/97 rotor. The supernatant fluid was transferred to a 1.5 ml microfuge tube and the protein concentration of each sample was determined by Bradford assay (Sigma). Equal amounts of protein (40 µg) were subjected to SDS-PAGE on 12% gels and western blotting was carried out as described above using the following primary antibodies: rabbit antibodies against SoxS ¹⁸; mouse monoclonal antibodies directed against the HA tag (Invitrogen) appended to the B42AD peptide fused to σ⁷⁰ R4; and rabbit antibodies against β-tubulin (ANASPEC), which served as a loading control. The procedure for development of the blots was done as described ⁶⁰. The blots were then treated with secondary antibodies and developed as described ⁶⁰. The resulting chemifluorescent signals were detected with a Molecular Dynamics Storm 860 PhosphorImager in fluorescence mode with a blue (450 nm) laser.

Abbreviations used

RNAP

RNA polymerase

CAP

catabolite activator protein

αCTD

C-terminal domain of RNAP α subunit

αNTD

N-terminal domain of RNAP α subunit

σ⁷⁰ R4

region 4 of σ⁷⁰

Y2H

yeast two-hybrid system

AIA

affinity immobilization assays

his₆

hexa-histidine tag

FT

flow-through

W3

third wash

EL

eluate

DBD

DNA binding domain

AD

activation domain

raf

raffinose

gal

galactose

X-gal

5-bromo-4-chloro-3-indolyl-β-D-galactoside