Amino Acid Contacts between Sigma 70 Domain 4 and the Transcription Activators RhaS and RhaR

Abstract

The RhaS and RhaR proteins are transcription activators that respond to the availability of l-rhamnose and activate transcription of the operons in the Escherichia coli l-rhamnose catabolic regulon. RhaR activates transcription of rhaSR, and RhaS activates transcription of the operon that encodes the l-rhamnose catabolic enzymes, rhaBAD, as well as the operon that encodes the l-rhamnose transport protein, rhaT. RhaS is 30% identical to RhaR at the amino acid level, and both are members of the AraC/XylS family of transcription activators. The RhaS and RhaR binding sites overlap the −35 hexamers of the promoters they regulate, suggesting they may contact the σ⁷⁰ subunit of RNA polymerase as part of their mechanisms of transcription activation. In support of this hypothesis, our lab previously identified an interaction between RhaS residue D241 and σ⁷⁰ residue R599. In the present study, we first identified two positively charged amino acids in σ⁷⁰, K593 and R599, and three negatively charged amino acids in RhaR, D276, E284, and D285, that were important for RhaR-mediated transcription activation of the rhaSR operon. Using a genetic loss-of-contact approach we have obtained evidence for a specific contact between RhaR D276 and σ⁷⁰ R599. Finally, previous results from our lab separately showed that RhaS D250A and σ⁷⁰ K593A were defective at the rhaBAD promoter. Our genetic loss-of-contact analysis of these residues indicates that they identify a second site of contact between RhaS and σ⁷⁰.

Transcription activation in Escherichia coli often involves the interaction of a DNA-binding activator protein with one of the subunits of RNA polymerase (RNAP), most often the sigma (σ) or alpha (α) subunit. Transcription activators that bind immediately upstream and adjacent to RNAP, in some cases overlapping the −35 promoter hexamer, may interact with the C-terminal domain (domain 4) of the σ subunit of RNAP (8, 27). The cI protein of bacteriophage λ is required for the establishment and maintenance of lysogeny and is perhaps the best-characterized example of a transcription activator that contacts σ⁷⁰. The λ cI protein activates transcription of the P_RM promoter when bound at the O_R2 operator site, which overlaps the P_RM −35 hexamer by 2 bp (30). Current evidence suggests that σ⁷⁰ residues R588, K593, and R596 are required for activation by λ cI (23, 26, 35). Genetic and molecular modeling studies, as well as the recent structure of a ternary λ cI-σ domain 4-DNA complex, indicate that λ cI D38 contacts both σ⁷⁰ K593 (σ^A K418) and R596 (σ^A R421) and λ cI E34 contacts σ⁷⁰ R588 (σ^A R413) (8, 19, 26, 35). Prior to the identification of the ternary complex structure, a molecular model of the interaction indicated that σ⁷⁰ K593 (σ^A K418) contacts DNA but was not positioned to contact λ cI (6, 8, 35). However, the ternary structure showed that the σ^A residue that aligns with σ⁷⁰ K593 has moved away from the DNA (relative to the model of the interaction) and instead makes a protein-protein contact with λ cI D38 (19).

There is also evidence that activation by several transcription activators in the AraC/XylS family involves σ⁷⁰ domain 4. Our lab previously identified two σ⁷⁰ residues, K593 and R599, which are required for full activation by RhaS and further obtained genetic evidence that σ⁷⁰ R599 is directly involved in a contact with RhaS D241 (4). These genetic results are also strongly supported by molecular modeling of the RhaS-σ⁷⁰ complex on DNA (4). Evidence for AraC interactions with σ⁷⁰ come from early σ⁷⁰ mutations (eventually identified at R596) that increased araBAD expression in the absence of activation by cyclic AMP receptor protein (CRP) (18, 39, 44), as well as the finding that araBAD expression in a Δcya strain was increased by σ⁷⁰ E591A and R596A and decreased by σ⁷⁰ K593A (27). In addition, with a DNA that mimicked an open complex, a small amount of DNA-binding cooperativity could be detected between AraC and σ⁷⁰ (7). At the melAB promoter, genetic evidence indicated that σ⁷⁰ R596 interacts with MelR D261 and T265 while σ⁷⁰ R599 also interacts with MelR D261, which aligns with RhaS D241 (16). The Ada protein has two activation domains, one of which is an AraC/XylS family domain which is required to activate transcription of the alkA operon (33). Alanine substitutions of σ⁷⁰ residues K593, K597, and R603 each led to significant defects in Ada-dependent alkA transcription in vivo and in vitro (24).

The transcription activator RhaS, and the closely related RhaR protein, activate transcription of the E. coli l-rhamnose catabolic operons in the presence of the sugar l-rhamnose (10, 11, 42). RhaS activates transcription of the rhaBAD and rhaT operons by binding as a dimer to sites that overlap the −35 hexamers of the promoters by 4 bp and extend upstream to −81 and −82, respectively (see Fig. 1 for the rhaBAD promoter) (10, 45). Similarly, RhaR activates transcription of the rhaSR operon by binding as a dimer to a site that overlaps the RNAP binding site by 4 bp and extends upstream to −82 (Fig. 1) (43). The long binding sites for RhaS and RhaR each consist of two 17-bp imperfect inverted repeat half sites that are separated by 16 or 17 bp of uncontacted DNA (10, 43, 45). Each RhaS and RhaR monomer is predicted to contain two helix-turn-helix DNA-binding motifs and thereby contact two consecutive major grooves of DNA (38). CRP also activates transcription at all three of the rha promoters. At the rhaBAD and rhaT promoters, the CRP binding site is located immediately upstream of the RhaS binding site and is centered at −92.5 and −93.5, respectively (see Fig. 1 for the rhaBAD promoter) (11, 45). The CRP site required for full activation at rhaSR is located upstream but not adjacent to the RhaR binding site and is centered at −111.5 (Fig. 1) (17).

FIG. 1.

(Top) Representation of the divergent rhaSR and rhaBAD promoter regions, showing the approximate positions of the transcription activators RhaS, RhaR, and CRP, as well as RNAP at each promoter. (Bottom) Three consecutive lines of DNA sequence extending from the rhaSR transcription start point to the rhaBAD transcription start point. Binding sites for RhaS, RhaR, and CRP are shown by arrows, and the −35 and −10 hexamers of each promoter are indicated. The upstream end points of promoter fusions used in this study are marked by Δs. Deletion end points, protein binding sites, and numbering relative to the rhaSR promoter are shown below the DNA sequence, while deletion end points, protein binding sites, and numbering relative to the rhaBAD promoter are shown above the DNA sequence.

RhaS and RhaR are members of a subset of the AraC/XylS family that share amino acid sequence similarity with AraC over its entire length (9, 13, 28). Based on this similarity, RhaS and RhaR are predicted to consist of two domains connected by a flexible linker (5, 12, 29, 41). The N-terminal domains are predicted to be responsible for l-rhamnose binding and dimerization, while the C-terminal domains contain the 99-amino-acid region that classifies them as members of the AraC/XylS family. In all studied cases of AraC/XylS family members, including RhaS and RhaR, the characteristic 99-amino-acid region constitutes a DNA-binding domain (3, 10, 43). This DNA-binding domain has also been shown to be involved in transcription activation in a number of AraC/XylS family proteins including Ada, RhaS, AraC, MelR, MarA, SoxS, XylS, and UreR (1, 4, 5, 15, 16, 20-22, 37).

In this study, we further explored the mechanisms of transcription activation by RhaR and RhaS. We identified amino acid residues in the C-terminal domain of σ⁷⁰ and in RhaR that are important for RhaR-mediated transcription activation at the rhaSR promoter. We then used a genetic loss-of-contact approach to identify an interaction between RhaR D276 and σ⁷⁰ R599 that is required for RhaR-mediated activation. We also extended the previous studies by Bhende and Egan (4) of RhaS-mediated transcription activation at rhaBAD. Here we identified a second interaction between RhaS and σ⁷⁰, in this case, RhaS D250 and σ⁷⁰ K593.

MATERIALS AND METHODS

Culture media and conditions.

E. coli cultures for β-galactosidase assays were grown in morpholinepropanesulfonic acid-buffered medium (4, 34). Tryptone-yeast extract liquid medium (0.8% tryptone, 0.05% yeast extract, 0.05% NaCl) was used to grow cells for most other experiments. SacB selection medium (1% tryptone, 0.5% yeast extract, 1.5% agar, 5% sucrose [pH 7.8]) was used to select against sacB⁺ strains (14). Antibiotics were used as indicated at the following concentrations: ampicillin (200 μg/ml), chloramphenicol (25 μg/ml), kanamycin (25 μg/ml), and tetracycline (20 μg/ml).

General methods.

Standard methods were used for restriction endonuclease digestion and ligation using restriction endonucleases and T4 DNA ligase purchased from New England Biolabs (Beverly, Mass.). Transformation was carried out using chemically induced competent cells of E. coli, and plasmid DNA was purified by alkaline lysis. DNA sequencing reactions were carried out using custom-synthesized IRD41 dye-labeled primers (Table 1) from LI-COR Inc. (Lincoln, Nebr.) and the Thermo Sequenase primer cycle sequencing kit from Amersham Life Sciences (Arlington Heights, Ill.). DNA sequences were analyzed by automated dideoxy sequencing on a LI-COR 4000L sequencer (University of Kansas Biochemical Research Service Laboratory). The Expand High Fidelity PCR system (Roche, Indianapolis, Ind.) was used to amplify DNA fragments for cloning as well as to generate templates for DNA sequencing from rhaS and rhaR alleles that were recombined into the chromosome. The DNA sequences of both strands were determined for the entire cloned region of all cloned, mutagenized, and recombined DNA fragments. The QIAquick PCR Purification kit (QIAGEN, Chatsworth, Calif.) was used to clean up PCR products.

TABLE 1.

Oligonucleotides used in this study

Oligonucleotide	Oligonucleotide sequence, 5′-3′a	Use
1170	CCGGAATTCTTGTGGTGATGTGATGCTCAC	Amplify rhaSRT′ (within rhaBAD promoter)
2068	ATGACCGTATTACATAGTGTGGAT	IRD41b labeled, rhaS sequencing
2069	TTATTGCAGAAAGCCATCCCGTCC	IRD41 labeled, rhaS sequencing
2074	TGGTTGCACAGATGGAACAGC	IRD41 labeled, rhaS sequencing
2075	GTTGAGACGTGATGCGCTGTT	IRD41 labeled, rhaS sequencing
2097	CGCGAATTCAAGGGTATGGTTTTGCAG	Amplify rhaSRT from chromosome (within rhaT promoter)
2204	GGTCACCGCGTGATATTCG	IRD41 labeled, rhaR sequencing
2205	ATTCCGGGATTTAACGCCAG	IRD41 labeled, rhaR sequencing
2206	TTAATCTTTCTGCGAATTGAG	IRD41 labeled, rhaR sequencing
2207	CAAACGGCACATGCTGACTA	IRD41 labeled, rhaR sequencing
2208	CGGTCGAAATTGCACTGATT	rhaR D276A mutagenesis
2210	AATAGTTACTTGCTTCAAAGCCAC	rhaR D285A mutagenesis
2292	CCTGGATCCCCGCAAAAGTGAA	Amplify rhaSRT′ from plasmids (within rhaT)
2297	GGTAAGATCTCGGTCATACTGGCCTCCTGATG	Long-way-around PCR to delete rhaSR
2298	GGTAAGATCTTTAATTCGCCATGCCGATGCCGA	Long-way-around PCR to delete rhaSR
2299	GGTAAGATCTCGGACCGGGTCGAATTTGC	Amplify cat-sac cassette
2300	GGTAAGATCTATATCGGCATTTTCTTTTGCG	Amplify cat-sac cassette
2381	ATTCTCTTCGTTGCGGATAGTAACTATTTTTCGGTGGTG	rhaR E284A mutagenesis

^aRegions of oligonucleotides not complementary to the wild-type DNA sequence (encode restriction endonuclease sites and flanking DNA or mutations) are underlined.

^bIRD41, infrared dye from LI-COR.

Strains, plasmids, and phages.

Table 2 lists the strains, phages, and plasmids used in this study. All strains used in β-galactosidase assays were derived from ECL116 (2) and carried lacZ fusions in a single copy on λ phage integrated at attλ (40). P1 phage-mediated generalized transduction was used to move Δ(recC ptr recB recD)::Plac-bet exo kan (from KM22) into SME1216 (selecting for kanamycin resistance) to make SME2417. SME2495 was made using P1 transduction to move Δ(rhaSR)::kan zih-35::Tn10 (from SME2800) into SME1217, selecting for tetracycline resistance and then screening for a Rha⁻ phenotype. SME2496 was made from SME2416 by transformation with a PCR product containing Δ(rhaSR)::cat-sac (amplified from pSE254, described below), which was recombined onto the chromosome by using the recombination genes of bacteriophage λ (encoded by Δ(recC ptr recB recD)::Plac-bet exo kan) (31). The plasmid pSE250 was made by restriction endonuclease digestion of pSE101 with BamHI at the rhaT′ end of the clone and EcoRI (a natural site within the rhaBAD promoter), creating the fragment rhaSRT′, which was purified from an agarose gel by using QIAGEN′s QIAEX Gel Extraction kit. The rhaSRT′ fragment was then ligated to pUC18 (46), which had been digested with BamHI and EcoRI. To make pSE254, long-way-around PCR using pSE250 as the template and primers 2297 and 2298 amplified all of the pSE250 sequence except rhaSR and added a BglII site at each end. Then, PCR with primers 2299 and 2300, using a PCR product containing the cat-sac cassette (provided by Kenan Murphy) (32) as the template, generated a product that was ligated to the BglII sites of the long-way-around PCR product to create pSE254. Plasmids pML148 to -169 (containing mutations in the rpoD gene) were obtained from the laboratory of Carol Gross and were sequenced to ensure that they still carried the expected mutations. Several of the rpoD alleles were initially found to be wild type. Assays involving these alleles were repeated upon obtaining true mutants.

TABLE 2.

Strains, phages, and plasmids used in this study

Strain, phage, or plasmid	Genotype	Source or reference
E. coli strains
KM22	Δ(recC ptr recB recD)::Plac-bet exo kan	31
ECL116	F− ΔlacU169 endA hsdR thi	2
SME1074	ECL116 λ SME106	Laboratory collection
SME1216	ECL116 λ SME103	Laboratory collection
SME1851	ECL116 λ SME104	Laboratory collection
SME2416	SME2417 zih-35::Tn10	Laboratory collection
SME2417	SME1216 Δ(recC ptr recB recD)::Plac-bet exo kan	This study
SME2495	SME2417 ΔrhaSR::kan zih-35::Tn10	This study
SME2496	SME2416 ΔrhaSR::cat-sacB	This study
SME2508	ECL116 λ SME114 recA::cat	Laboratory collection
SME2515	ECL116 λ SME114	Laboratory collection
SME2608	SME1851 rhaS(wild type) recA::kan	Laboratory collection
SME2689	SME1851 rhaS(D250A) zih-35::Tn10 recA::kan	This study
SME2691	SME2515 rhaR(D276A) zih-35::Tn10 recA::kan	This study
SME2692	SME2515 rhaR(D285A) zih-35::Tn10 recA::kan	This study
SME2693	SME2515 rhaR(wild type) recA::kan	This study
SME2800	SME1074 ΔrhaSR::kan zih-35::Tn10	This study
SME2933	SME2515 rhaR(E284A) zih-35::Tn10 recA::kan	This study
Phages
λ RS45	bla′-lacZscatt+imm21ind+	40
λ SME103	λ RS45 φ(rhaB-lacZ)Δ110	11
λ SME104	λ RS45 φ(rhaB-lacZ)Δ84	11
λ SME106	λ RS45 φ(rhaS-lacZ)Δ216	11
λ SME114	λ RS45 φ(rhaS-lacZ)Δ92	17
Plasmids
pGEX-2T σ70	AprrpoD, wild type	27
pML148 to -169	pGEX-2T, rpoD substitutions	27
pSE101	pTZ18R AprrhaSR rhaBA′	Laboratory collection
pSE172	Tetr pALTER-1 rhaS D250A	Laboratory collection
pSE249	pSE101 rhaS D250A	This study
pSE250	pUC18 rhaSRT′, wild type	This study
pSE251	pUC18 rhaSRT′ rhaR D276A	This study
pSE252	pUC18 rhaSRT′ rhaR E284A	This study
pSE253	pUC18 rhaSRT′ rhaR D285A	This study
pSE254	pUC18 ΔrhaSR::cat-sac	This study
pUC18	AprlacZα	46

Mutagenesis of rhaS and rhaR.

The mutant rhaS D250A allele was moved from pSE172 into the context of pSE101 by digesting pSE172 with BstEII and BglII (both sites are native to the wild-type rhaS gene) to create a fragment encoding the RhaS D250A substitution. This fragment was then ligated to similarly digested pSE101 to make pSE249. Genes encoding alanine substitution derivatives of RhaR D276A and RhaR D285A were constructed by oligonucleotide-directed mutagenesis of rhaR in pGEM-11Zf(+) (Promega, Madison, Wis.), using the GeneEditor kit (Promega) and oligonucleotides 2208 and 2210. The mutant rhaR alleles were then subcloned into pSE250, using NheI and SmaI restriction endonuclease sites (both sites occur naturally within rhaR), to make pSE251 and pSE253, respectively. The rhaR E284A mutagenesis was performed using PCR to make oligonucleotide-directed mutations at the desired position with primer 2381, which also contained the recognition sequence for the EarI restriction endonuclease. Second, a nonmutagenized PCR fragment was made, also using a primer with EarI restriction sites. Finally, ligation of the mutant and wild-type DNA fragments allowed seamless reconstruction (25) of the full-length rhaR E284A gene in the context of rhaSRT′ to make pSE252. Oligos Etc (Wilsonville, Oreg.), Integrated DNA Technologies (Coralville, Iowa), and MWG-Biotech (High Point, N.C.) synthesized oligonucleotide primers used in mutagenesis (Table 1). Mutations were initially identified by diagnostic PCR using the following method. The very 3′ nucleotides of the diagnostic oligonucleotide contained the desired substitution(s), such that amplification was possible only (in combination with a suitable downstream primer) when the template DNA carried the desired mutation. Putative mutants identified by this method were confirmed by DNA sequencing of both strands of the entire cloned region (see Table 1 for sequencing oligonucleotides).

Recombination of rhaS and rhaR alleles onto the chromosome.

The mutant rhaS and rhaR alleles constructed as described above were recombined onto the E. coli chromosome such that they replaced the wild-type rhaS or rhaR allele by the following methods. Each rhaS or rhaR mutant was present on a plasmid in the context of rhaSRT′ (pSE249, pSE251, pSE252, or pSE253). Oligonucleotides 1170 and 2292 were used to amplify each mutant rhaSRT′ region by high-fidelity PCR. Approximately 500 ng of the rhaSRT′ PCR product carrying a mutant allele was used to transform either SME2495 [λ Φ(rhaB-lacZ)Δ110 Δ(recC ptr recB recD)::Plac-bet exo kan Δ(rhaSR)::kan, zih-35::Tn10] or SME2496 [λ Φ(rhaB-lacZ)Δ110 Δ(recC ptr recB recD)::Plac-bet exo kan Δ(rhaSR)::cat-sac, zih-35::Tn10]. Since SME2495 and SME2496 both contained P_lac-bet exo, which encodes the λ phage recombination proteins, the frequency of homologous recombination was much higher in these strains than in wild-type E. coli strains (31). The transformants were screened by spread plating on media containing 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) (40 μg/ml) and l-rhamnose (0.2%) to identify functional, or partially functional, rhaSR genes that replaced the ΔrhaSR allele. There was no selection for successful recombinants in SME2495; rather we screened for blue colonies among a lawn of white colonies. When transforming SME2496, which contains a cat-sacB cassette (32) in place of rhaSR, we selected for the sucrose resistance of cells that had lost the sacB gene (which confers sucrose sensitivity) by homologous recombination. However, due to a significant background of spontaneous sucrose-resistant mutants, the transformants were also screened for at least a partially functional rhaS or rhaR gene by adding X-Gal (40 μg/ml) and l-rhamnose (0.2%) to the SacB selection plates. We found that sucrose inhibition of the sacB⁺ cells worked most reproducibly at room temperature, although it took about 3 days for the cells to grow. Phage P1-mediated generalized transduction was then used to transfer the rhaS or rhaR allele of interest (linked to zih-35::Tn10) to either SME1851 [λ Φ(rhaB-lacZ)Δ84] or SME2515 [λ Φ(rhaS-lacZ)Δ92] by selecting for the tetracycline resistance conferred by zih-35::Tn10. Diagnostic PCR, as described above, was used to initially identify transductants that contained the rhaS or rhaR mutation of interest. High-fidelity PCR was then used to amplify rhaSRT′ from the chromosome, using oligonucleotides 2097 and 1170, and the entire 3-kb PCR product was sequenced, as described above, to verify the presence of the desired mutation with no additional mutations. Phage P1-mediated transduction was then used to introduce recA::kan into each strain to make SME2689, -2691, -2692, and -2933. Finally, competent cells of each strain were made and transformed with plasmids containing either the wild type-, K593A-, L595A-, R599A-, or R608A-encoding σ⁷⁰ gene for β-galactosidase assays.

β-Galactosidase assay.

β-Galactosidase assays were performed as previously described (3). In all cases, chromosomal rpoD was expressed from its own promoter, not the trp promoter described by Lonetto et al. (27), and the plasmid-encoded σ⁷⁰ derivatives were expressed in the absence of isopropyl-β-d-thiogalactopyranoside. Under these conditions, the σ⁷⁰ derivatives are expected to account for approximately 50% of the total σ⁷⁰ in the cells (27). Specific activities were averaged from at least three independent assays with two replicates in each assay. The assays were performed on at least two different days, with independent cell growth steps (starter tryptone-yeast extract culture, overnight culture, and final growth culture) for each assay.

RESULTS

σ⁷⁰ derivatives at the rhaSR promoter.

We wished to determine whether any residues near the C-terminal end of σ⁷⁰ were important for transcription activation by RhaR. Lonetto et al. (27) constructed a library of alanine substitutions at 17 different positions near the C terminus of σ⁷⁰ and found that some substitutions resulted in defects at class II activator-dependent promoters. Previous work from our lab found that two of the alanine substitutions in this library were defective at a truncated rhaBAD promoter where RhaS was the only transcription activator (4). We assayed this library of alanine substitutions in σ⁷⁰ at two RhaR-activated single-copy translational fusions, Φ(rhaS-lacZ)Δ216 and Φ(rhaS-lacZ)Δ92. The Φ(rhaS-lacZ)Δ216 promoter contained the RhaR binding site as well as upstream CRP sites, while the Φ(rhaS-lacZ)Δ92 promoter contained only the RhaR binding site (Fig. 1). Since the assays were carried out with a strain that also expressed wild-type σ⁷⁰ from the chromosome (27), we considered values below 80% of wild-type activity to be significant defects. At Φ(rhaS-lacZ)Δ216, σ⁷⁰ derivative L595A had 66% of the activity of wild-type σ⁷⁰ (Fig. 2A) while the remaining σ⁷⁰ derivatives were not significantly defective. When the same sigma derivatives were assayed at Φ(rhaS-lacZ)Δ92, L595A was still significantly defective, with 53% activity compared to wild-type σ⁷⁰ (Fig. 2B). In addition, three alanine substitutions of positively charged amino acid residues, K593A (79%), R599A (48%), and R608A (77%) were defective at Φ(rhaS-lacZ)Δ92. These results suggested that σ⁷⁰ residues K593, L595, R599, and R608 might make protein-protein contacts with RhaR that are required for transcription activation. The lack of defect from σ⁷⁰ K593A, R599A, or R608A at the Φ(rhaS-lacZ)Δ216 promoter is similar to previous findings with RhaS and AraC that substitutions at some σ⁷⁰ residues were defective only in the absence of CRP activation (4, 27).

FIG. 2.

Alanine substitutions within the C-terminal domain of the σ⁷⁰ subunit of RNAP assayed at two rhaS-lacZ fusions, Φ(rhaS-lacZ)Δ216 in SME1074 (A) and Φ(rhaS-lacZ)Δ92 in SME2508 (B). The σ⁷⁰ alanine substitutions were encoded on plasmids, and the rhaS-lacZ fusions were in the chromosome as single-copy λ lysogens. In each panel, the values obtained with wild-type σ⁷⁰ were set to 100% and the activity of each σ⁷⁰ derivative is represented as a percentage of the wild-type σ⁷⁰ value. In panel A, the activity of wild-type σ⁷⁰ was 86 Miller units for the I590A, R596A, L598A, R603A, and R608A derivatives, while the wild-type σ⁷⁰ activity for the other derivatives was 87 Miller units. In panel B, the wild-type σ⁷⁰ activity was 3.6 Miller units for the I590A, R596A, L598A, R603A, and R608A derivatives and 2.2 Miller units for the other derivatives. β-gal, β-galactosidase.

Based on the previous finding in our lab that a contact between RhaS residue D241 and σ⁷⁰ residue R599 is required for full transcription activation by RhaS (4), we predicted that RhaR D276, which aligns with RhaS D241 (Fig. 3), might contact σ⁷⁰ R599 at the rhaSR promoter. Molecular modeling of the RhaR-σ⁷⁰ interaction (Fig. 4) indicates that the negatively charged RhaR residue D276 is very close to the positively charged σ⁷⁰ residue R599. RhaR D276 is also near σ⁷⁰ R608, although in the model they do not appear close enough to interact. The molecular model further shows that two adjacent negatively charged RhaR residues, E284 and D285, are located near σ⁷⁰ K593. Based on these pieces of evidence, we hypothesized that contacts between some or all of the RhaR residues D276, E284, and D285 and σ⁷⁰ might be required for maximal transcription activation by RhaR. We therefore tested alanine substitutions at these positions in RhaR for defects in transcription activation.

FIG. 3.

Alignment of the amino acid sequences of the second helix-turn-helix DNA-binding motifs of RhaS and RhaR. Amino acids shown in bold are those tested in this work for possible interactions with the σ⁷⁰ subunit of RNAP. Identical amino acids are indicated by vertical lines between the two sequences. The boundaries of the first helix (Helix 1), the turn, and the recognition helix (Helix 2) are based on the structure of MarA (38) and alignments between MarA and RhaS and RhaR. The numbers of the first and last residues shown, as well as those of the residues tested for interactions with σ⁷⁰, are indicated.

FIG. 4.

Model of RhaS or RhaR interactions with σ⁷⁰ domain 4. The model of the RhaS or RhaR C-terminal domain (aqua) is based on the crystal structure of the MarA-DNA complex (38), while the model of σ⁷⁰ domain 4 (green) is based on the crystal structure of the same domain of σ^A from Thermus aquaticus on DNA (6). Only the DNA from the MarA structure is shown (white). Amino acid residues in RhaS or RhaR and σ⁷⁰ that are implicated in interactions are shown in a space-filling model and labeled, with the RhaR or RhaS residues colored red and the σ⁷⁰ residues colored dark blue. The unlabeled space filling residues are RhaR E284 (pink), RhaR D285 (which is at the same position as RhaS D250), and σ⁷⁰ R608 (light blue). Since σ⁷⁰ sits in front of RhaS or RhaR when the DNA is shown parallel to the page, the model has been rotated somewhat around the vertical axis to allow a view between the interacting proteins. The modeling was performed using the program Insight II (Accelrys, Inc.) by first manually superimposing the DNAs in the PDF files of MarA on DNA (Protein Data Bank file 1BL0) and σ^A domain 4 on DNA (Protein Data Bank file 1KU7) such that the base pairs that corresponded to the −35 region of each were aligned as closely as possible. The σ^A model was then rotated to minimize clashes with MarA while maintaining the DNA superimposition. Finally, the residues implicated in interactions were highlighted. A second σ^A domain 4 molecule in the 1KU7 structure which does not make specific contacts with the DNA is not shown.

RhaR residues D276, E284, and D285 are important for rhaSR transcription activation.

In order to test whether the side chains of RhaR residues D276, E284, and D285 might play a role in transcription activation by RhaR, we constructed alanine substitutions at each of these positions. If the amino acid residues at these positions are required for transcription activation, the alanine substitution should result in a significant decrease in activation of rhaSR transcription. To assay the RhaR derivatives, the mutant rhaR alleles on plasmids were first recombined onto the chromosome such that they replaced the wild-type rhaR gene (see Materials and Methods). The wild-type and mutant rhaR alleles were then assayed for activation of Φ(rhaS-lacZ)Δ92 (Fig. 5). The results showed that all three of the alanine substitutions in RhaR were significantly defective, highlighting the importance of the wild-type residues at those positions. The especially large defect of RhaR D285A may be partly due to a role in DNA binding based on its alignment with D250 in RhaS (Fig. 3), which makes base-specific contacts with DNA (3). However, a role in DNA binding for RhaR D285 does not rule out interactions with σ⁷⁰; therefore, all three of these RhaR residues are candidates for specific contacts with σ⁷⁰.

FIG. 5.

Transcription activation by RhaR derivatives. β-Galactosidase activity was assayed from a single-copy fusion of the rhaSR promoter with lacZ that included the RhaR binding site but not the CRP binding sites [Φ(rhaS-lacZ)Δ92]. In each case, wild-type RhaR or the alanine substitutions in RhaR were encoded in the chromosome at the natural rhaR locus (strains SME2691, -2692, -2693, and -2933). The value obtained with wild-type RhaR (3.3 Miller units) was set to 100%, and the activity of each RhaR derivative is represented as a percentage of that value.

Evidence for an interaction between RhaR D276 and σ⁷⁰ R599.

We used a genetic loss-of-contact approach to test potential interactions between RhaR D276 and σ⁷⁰ K593 or R599. Using this approach, we separately combined wild-type RhaR or the RhaR D276A derivative with each of three plasmids encoding either σ⁷⁰ wild type, K593A, or R599A in a strain carrying a single copy of Φ(rhaS-lacZ)Δ92. The results shown in Fig. 6A were plotted so that the activity with wild-type σ⁷⁰ was set to 100% for each RhaR derivative, thereby illustrating the relative effects of each σ⁷⁰ derivative. On this graph, therefore, a σ⁷⁰ derivative that does not define a site of interaction with a given RhaR derivative is expected to have the same relative defect when combined with the indicated RhaR derivative as it does with wild-type RhaR, since the defects will be independent of each other. On the other hand, if a σ⁷⁰ derivative does define a site of interaction with a given RhaR derivative, the σ⁷⁰ derivative will confer no further defect when combined with the indicated RhaR derivative, since the interaction would already have been lost with the RhaR derivative. Using this method to analyze the results in Fig. 6A, our first conclusion is that there is no interaction between RhaR D276 and σ⁷⁰ K593 since the σ⁷⁰ K593A derivative had approximately the same relative defect in combination with either wild-type RhaR or RhaR D276A. Therefore, the defects of σ⁷⁰ K593A and RhaR D276A are independent. These results and those in Fig. 2B also show that the σ⁷⁰ R599A derivative by itself (in a wild-type rhaR strain) had approximately 50% activity compared to wild-type σ⁷⁰. However, when σ⁷⁰ R599A was combined with RhaR D276A, the σ⁷⁰ R599A derivative conferred no further defect upon RhaR D276A. In fact, the strain with the combination of σ⁷⁰ R599A and RhaR D276A had approximately 1.7-fold-higher activity than the strain with wild-type σ⁷⁰ and RhaR D276A. These results fit the criteria for an allele-specific contact between σ⁷⁰ R599 and RhaR D276. As mentioned above, molecular modeling is consistent with this interaction since RhaR D276 is in close proximity to σ⁷⁰ R599 in the model (Fig. 4). We also tested the σ⁷⁰ R608A derivative in combination with RhaR D276A and found that it had the same relative defective as it did with wild-type RhaR (79% of wild-type σ⁷⁰ activity in both cases [data not shown]); therefore, there was no indication of an interaction between these two residues.

FIG. 6.

Combinations of RhaR or RhaS derivatives with σ⁷⁰ derivatives. Plasmid-encoded σ⁷⁰ alanine substitutions were combined with chromosomally encoded alanine substitutions in RhaR or RhaS (strains SME2689, -2691, -2692, -2693, and -2933), and β-galactosidase activity was measured from Φ(rhaS-lacZ)Δ92 (A, B, and C), or Φ(rhaB-lacZ)Δ84 (D). The activity of wild-type σ⁷⁰ in combination with each RhaR or RhaS derivative in Miller units is shown on the corresponding bar in each graph and was set to 100% in each case. This representation allows the relative defects of the σ⁷⁰ derivatives to be directly compared. The value for σ⁷⁰ K593A in combination with RhaR E284A was 362% of the wild-type σ⁷⁰ value and is drawn off scale to prevent compression of the remaining bars.

RhaR E284 and D285 and σ⁷⁰.

Using the same genetic loss-of-contact approach, we also tested for potential interactions between σ⁷⁰ and RhaR E284 and D285. The results in Fig. 6B show that the K593A σ⁷⁰ derivative was not defective in combination with RhaR D285A (104% of wild-type σ⁷⁰ activity) but the R599A σ⁷⁰ derivative also became less defective (81% of wild-type σ⁷⁰ activity). In the absence of the results obtained for σ⁷⁰ 599A, one might conclude that RhaR D285 contacts σ⁷⁰ K593, since σ⁷⁰ K593A had no significant defect when combined RhaR D285A; however, the lack of strict allele specificity sheds doubt on this conclusion. To further investigate the non-allele-specific defects of σ⁷⁰ substitutions in combination with RhaR D285A, we tested σ⁷⁰ L595A and R608A derivatives, which were both defective in a wild-type rhaR strain, as shown in Fig. 2B. When combined with RhaR D285A, the σ⁷⁰ L595A and R608A derivatives were not significantly defective, with 86 and 87% of wild-type σ⁷⁰ activity, respectively (data not shown). These results suggest that RhaR D285A may reduce the ability of RhaR to interact with σ⁷⁰ in a non-allele-specific manner; therefore, we can't conclude whether RhaR D285 contacts any of these σ⁷⁰ residues.

Figure 6C shows the results of assays to identify potential interactions involving RhaR E284. Our results showed that neither σ⁷⁰ K593A nor R599A conferred a significant defect on RhaR E284A. In fact, the σ⁷⁰ K593A-RhaR 284A combination gave much higher activity (362%) than the RhaR 284A derivative with wild-type σ⁷⁰. Therefore, as described above, we tested the σ⁷⁰ L595A and R608A derivatives in the rhaR E285A strain and found 139 and 93% activity, respectively, compared to wild-type σ⁷⁰ (data not shown). Thus, we again found that all four of the σ⁷⁰ derivatives that were defective in the wild-type rhaR strain were no longer significantly defective in the rhaR E284A strain. These results suggest that, similar to the RhaR D285A derivative, the RhaR E284A derivative may reduce the ability of RhaR to interact with σ⁷⁰ in a non-allele-specific manner. One explanation for the very high relative activity of RhaR E284A in combination with σ⁷⁰ K593A is that a new interaction may have been created in this case.

Evidence for a specific interaction between RhaS D250 and σ⁷⁰.

Our lab previously identified an interaction between RhaS D241 and σ⁷⁰ R599, and we also found that σ⁷⁰ K593A was defective at Φ(rhaB-lacZ)Δ84 but did not identify an amino acid in RhaS that might contact σ⁷⁰ K593 (4). The molecular model in Fig. 4 shows that the only negatively charged RhaS residue that is in close proximity to the positively charged σ⁷⁰ K593 is RhaS D250, suggesting that these two residues might make a contact. Previous results from our lab showed that RhaS D250A was 12-fold defective for Φ(rhaB-lacZ)Δ84 activation; however, they also indicated that this residue participates in a base-specific DNA contact (3). In contrast to other approaches to identify positive-control mutants, the genetic loss-of-contact approach does not require that the protein have wild-type DNA-binding capability; hence it has the potential to identify residues that have dual DNA-binding and transcription activation functions. We therefore used the genetic loss-of-contact approach to test whether RhaS D250 and σ⁷⁰ K593 might be involved in an interaction. The results (Fig. 6D) support the hypothesis of an interaction between RhaS D250 and σ⁷⁰ K593 since the K593A derivative was not significantly defective when combined with RhaS D250A (87% activity compared to wild-type σ⁷⁰). However, when σ⁷⁰ R599A was combined with RhaS D250A, it maintained approximately the same relative defect as it had with wild-type RhaS. These results suggest that there is an allele-specific interaction between RhaS D250 and σ⁷⁰ K593. Molecular modeling is consistent with this interaction since, as mentioned above, RhaS D250 is in close proximity to σ⁷⁰ K593 in the model (Fig. 4).

DISCUSSION

The C terminus of σ⁷⁰ is important for RhaS- and RhaR-mediated transcription activation.

The binding sites for both RhaS and RhaR overlap the −35 region of their respective core promoters by 4 bp, placing them in ideal positions to interact with the σ⁷⁰ subunit of RNAP. Previous results reported by Bhende and Egan (4) identified two amino acid residues in σ⁷⁰, K593 and R599, that were important for RhaS-mediated transcription activation at rhaBAD and rhaT. In the present study, we identified four amino acid residues in σ⁷⁰, K593, L595, R599, and R608, which were important for RhaR-mediated transcription activation of rhaSR (Fig. 2B). Two of the alanine substitutions in σ⁷⁰, K593A and R599A, were defective at all three of the rha promoters, suggesting similar mechanisms of activation by RhaS and RhaR.

The results reported in this paper (Fig. 2), as well as those from a previous study (4), showed that σ⁷⁰ K593A and R599A were defective only at truncated rha promoters that did not include the upstream CRP binding sites. This is similar to the findings obtained for several other promoters that require multiple activators, such as araBAD, uhpT, and narG (18, 27, 36, 39). Two possible explanations for this trend are that the second activator increases the total number of interactions such that the relative importance of each individual interaction decreases or that the second activator creates redundancies in activation that mask the importance of other interactions. A third possibility is that the second activator alters the orientation of the first activator relative to σ⁷⁰ such that the primary activator is no longer in an ideal position to interact with σ⁷⁰. In the first two models, the activator interaction with σ⁷⁰ occurs in both the presence and absence of the second activator but can be detected only in its absence, whereas in the third model, the interaction between the first activator and σ⁷⁰ occurs only in the absence of the second activator. Further experiments will be needed to distinguish these models. At the rhaSR promoter, the σ⁷⁰ L595A derivative was unique in that it was defective in both the presence and absence of the second activator, CRP, but it's role in RhaR-mediated transcription activation is not yet known.

Specific amino acid contacts between σ⁷⁰ and RhaR.

Previous results showing an interaction between RhaS D241 and σ⁷⁰ R599 at rhaBAD (4) led us to investigate whether an interaction between RhaR D276 and σ⁷⁰ R599 might be required for RhaR activation at rhaSR. We also used a molecular model of the RhaR-σ⁷⁰ domain 4 interaction in which the structure of MarA (38) represented RhaR (Fig. 4) to identify the only two negatively charged RhaR residues, E284 and D285, that were near σ⁷⁰ K593. RhaR E284 and D285 were therefore considered candidates for residues that might interact with σ⁷⁰ K593. After determining that alanine substitutions at RhaR residues D276, E284, and D285 were all defective for rhaSR activation (Fig. 5), we used a genetic loss-of-contact approach to test for specific amino acid interactions between σ⁷⁰ and RhaR.

To carry out a loss-of-contact analysis, one must first identify defective derivatives of each of the potentially interacting proteins. In the simplest case, the full defect of both of the two interacting residues is due to loss of the interaction—in other words, the only role of the two residues is the interaction. The rationale behind this approach in this simple case is that mutation of one or the other of the interacting residues will eliminate the interaction; therefore, the phenotype of a strain carrying both mutations will be the same as the phenotype of the strains carrying the individual mutations. If one of the residues has a second role in addition to the interaction, then the strain carrying both mutations will have a phenotype that is no worse than the more defective of the strains carrying the two individual mutations. This analysis does not provide conclusive results if both residues have roles in addition to the interaction. It is also expected that the predicted interactions will be allele specific. The majority of combinations of defective derivatives are not expected to identify interacting residues, and in these cases the defects resulting from each of the two mutations will at least be additive.

Using this rationale to interpret our genetic loss-of-contact assays, the results in Fig. 6A provide evidence for an interaction between σ⁷⁰ R599 and RhaR D276. This result is similar to previous results from our lab (4) that indicate an interaction between σ⁷⁰ R599 and RhaS D241 and evidence from Grainger et al. that σ⁷⁰ R599 interacts with MelR D261 (16), which aligns with RhaS D241 and RhaR D276. Molecular modeling of the RhaR-σ⁷⁰ complex (Fig. 4) shows that σ⁷⁰ R599 and RhaR D276 are in close proximity, consistent with our interpretation that these two residues interact. Our genetic loss-of-contact results do not provide evidence for an interaction between σ⁷⁰ K593 and RhaR E284 or D285 (Fig. 6C). Instead, our results indicate that alanine substitutions at RhaR E284 and D285 result in non-allele-specific decreases in the defects of all of the σ⁷⁰ alleles tested. One hypothesis is that RhaR E284A and D285A alter the details of the RhaR DNA interaction such that RhaR is no longer in an ideal position to interact with σ⁷⁰ domain 4.

The role of σ⁷⁰ K593 in transcription activation by RhaS.

Residue K593 of σ⁷⁰ has been found to be important for several transcription activators, including AraC, UhpA, λ cI, FNR, Ada, RhaR (this study), and RhaS (4, 24, 27, 35, 36). With the exception of λ cI and RhaS (this study), evidence that σ⁷⁰ K593 directly contacts an activator has not been obtained. Our results indicate that σ⁷⁰ K593 contacts RhaS D250 as a part of the mechanism of activation by RhaS (Fig. 6D). Our molecular model of the RhaS-σ⁷⁰ interaction shows that σ⁷⁰ K593 and RhaS D250 are in close proximity, consistent with this result (Fig. 4). While the binary complex of Taq σ^A domain 4 and DNA shows that the residue that corresponds to σ⁷⁰ K593 contacts DNA, in the λ cI-σ domain 4-DNA ternary complex, this residue participates in a protein-protein contact with λ cI instead (6, 19). These findings indicate that σ⁷⁰ K593 is capable of interacting with an appropriately positioned transcription activator and are consistent with our proposal that σ⁷⁰ K593 may contact RhaS D250.

Comparison of transcription activation by RhaS and RhaR.

In this study we identified an interaction between RhaR D276 and σ⁷⁰ R599 that is equivalent to our previously identified interaction of RhaS D241 and σ⁷⁰ R599. Further, although a RhaR equivalent of the RhaS D250 interaction with σ⁷⁰ K593 was not identified, our results do not rule out that such an interaction occurs with RhaR. Therefore, our current evidence suggests that the RhaS-σ⁷⁰ interface is similar to the RhaR-σ⁷⁰ interface. We certainly expect, however, that not all aspects of RhaS activation and RhaR activation will be identical. For example, we know that the CRP site at rhaBAD is centered at position −92.5, whereas the most important CRP site at rhaSR is centered at position −111.5. It is not possible to draw conclusions about how or whether differences in the RhaS-σ⁷⁰ and RhaR-σ⁷⁰ interfaces might relate to this difference in CRP binding site position since all but one of the σ⁷⁰ derivatives tested were defective only in the absence of CRP. However, it is likely that there is a difference in the mechanisms of RhaS and RhaR activation that relates to this difference in the positions of the CRP binding sites.