Activation by TyrR in Escherichia coli K-12 by Interaction between TyrR and the α-Subunit of RNA Polymerase

ABSTRACT

A novel selection was developed for mutants of the C-terminal domain of RpoA (α-CTD) altered in activation by the TyrR regulatory protein of Escherichia coli K-12. This allowed the identification of an aspartate to asparagine substitution at residue 250 (DN250) as an activation-defective (Act⁻) mutation. Amino acid residues known to be close to D250 were altered by in vitro mutagenesis, and the substitutions DR250, RE310, and RD310 were all shown to be defective in activation. None of these mutations caused defects in regulation of the upstream promoter (UP) element. The rpoA mutation DN250 was transferred onto the chromosome to facilitate the isolation of suppressor mutations. The TyrR mutations EK139 and RG119 caused partial suppression of rpoA DN250, and TyrR RC119, RL119, RP119, RA77, and SG100 caused partial suppression of rpoA RE310. Additional activation-defective rpoA mutants (DT250, RS310, and EG288) were also isolated, using the chromosomal rpoA DN250 strain. Several new Act⁻ tyrR mutants were isolated in an rpoA⁺ strain, adding positions R77, D97, K101, D118, R119, R121, and E141 to known residues S95 and D103 and defining the activation patch on the amino-terminal domain (NTD) of TyrR. These results support a model for activation of TyrR-regulated genes where the activation patch on the TyrR NTD interacts with the TyrR-specific patch on the α-CTD of RNA polymerase. Given known structures, both these sites appear to be surface exposed and suggest a model for activation by TyrR. They also help resolve confusing results in the literature that implicated residues within the 261 and 265 determinants as activator contact sites.

IMPORTANCE Regulation of transcription by RNA polymerases is fundamental for adaptation to a changing environment and for cellular differentiation, across all kingdoms of life. The gene tyrR in Escherichia coli is a particularly useful model because it is involved in both activation and repression of a large number of operons by a range of mechanisms, and it interacts with all three aromatic amino acids and probably other effectors. Furthermore, TyrR has homologues in many other genera, regulating many different genes, utilizing different effector molecules, and in some cases affecting virulence and important plant interactions.

INTRODUCTION

The regulatory protein TyrR plays a central role in the regulation of aromatic biosynthesis and transport in Escherichia coli K-12, being involved in the regulation of at least nine transcriptional units by repression or activation. In E. coli K-12, regulation varies according to whether the coeffector is tyrosine, phenylalanine, or tryptophan (1).

TyrR also functions as a regulator in many other Proteobacteria, as reviewed previously (1), and plays a role in virulence in Yersinia pestis (2) and in uropathogenic E. coli and Salmonella bongori (3) and in biofilm formation and nitrogen fixation in plants (4, 5), possibly regulating up to 150 genes. Manipulation of the regulatory gene tyrR has also shown potential for production of commercially significant compounds like tryptophan (6), phenylalanine (7), and l-dopa (8).

TyrR and its homologues PhhR (9, 10) and GcsR (11, 12) in Pseudomonas are members of the NtrC family and have been classified by Ghosh et al. (13) as atypical bacterial-enhancer-binding proteins (bEBP-like), on the basis of domain structure and the presence of ATP binding sites. Unlike other members, TyrR and PhhR are both active at σ⁷⁰ promoters. The structure of TyrR, the nature and arrangement of the TyrR boxes to which it binds, and its role in regulation have been reviewed elsewhere (1).

The role of TyrR as an activator in E. coli K-12 has been established for the genes tyrP, mtr, aroP P3, and folA (14–16). The mtr gene is activated by tyrosine, phenylalanine, and tryptophan in association with TyrR and is also repressed by tryptophan and TrpR (17, 18). TyrR combined with phenylalanine also activates tyrP expression, whereas TyrR combined with tyrosine represses tyrP expression (19) and activates folA expression (14) and tyrP₊₃ or tyrP₊₄ expression, where repression is negated by insertion of 3 bp or 4 bp between the two TyrR boxes (20).

The amino-terminal domain (NTD) of TyrR is necessary for activation of gene expression, and strains with deletions in this region completely lack activation. Kwok has shown that the isolated N-terminal domain of TyrR exists as a dimer (21), and its structure has been determined by X-ray crystallography (22, 23). The orientation of the two monomers in the dimer is slightly asymmetrical, with the interlinking helix and Per-Arnt-Sim (PAS) domain in one monomer rotated slightly relative to the other. This orientation might change by movement around a hinge, as a result of binding to another protein or interaction with an effector molecule.

By a combination of selection and targeted oligonucleotide-directed mutagenesis, two regions in the TyrR NTD important in activation by tyrosine, phenylalanine, and tryptophan have been identified. The first region includes R2, L3, V5, C7, D9, R10, T14, E16, D19, G37, and M55 (24–26). Residues between 2 and 72 have been postulated to constitute a ligand-binding ACT domain on the basis of iterative database searches (21, 27) and by analysis of mutations in the context of the known structure (23) (Fig. 1).

FIG 1.

Schematic diagram of the NTD of the TyrR protein. The structural model of the TyrR NTD dimer is shown in cartoon representation. The ACT domain at the top of the diagram interacts with effectors. The PAS domain in the bottom of the diagram interacts with the RpoA CTD, and the locations of important amino acid residues involved in activation are indicated as spheres of main-chain atoms, with residue type and number labeled. The structure is from the Protein Data Bank (accession number EBI-30882) (23) and is presented using the PyMOL molecular graphics system, version 2 (Schrödinger, LLC).

The second region in which mutations affect activation (P92, V93, S95, and D103) has been identified as a PAS domain, generally associated with signal transduction (28). This was initially placed between residue 80 and residue 114 (26, 27) but on the basis of the crystal structure was later found to be located between residues 89 and 175 (23) (Fig. 1). This region was a candidate for the interaction with the carboxy-terminal domain (CTD) of the α-subunit of RNA polymerase (RNAP), to bring about activation.

The α-CTD (amino acids [aa] 249 to 329) exists as a discrete domain joined via a flexible linker to the rest of the RpoA protein and appears to be involved exclusively in modulation of gene expression (29). Such modulation includes not only activation by specific activators (e.g., class I activation by catabolite activator protein [CAP]) but also enhancement of transcription by interaction with upstream promoter (UP) sequences, A/T-rich DNA in the −40 to −60 region of many genes (30, 31).

Activation of transcription by activators such as TyrR is generally mediated by binding to a specific binding site upstream of the promoter and subsequent interaction with the CTD of the α-subunit of RNAP (32–35). In the case of TyrR, this was first shown by in vitro experiments with the tyrP gene (33). These results, together with the position of the TyrR Box involved in activation, led to its classification as a class I activator.

Further experiments with tyrP₊₃ and mtr carried out on supercoiled templates showed that when expression was inhibited by type II DNA-binding proteins that induce bending, HU (histone-like protein) and integration host factor (IHF), activation of gene expression occurred with wild-type TyrR and RNAP containing the intact α-CTD but not if activation-defective (Act⁻) TyrR or the truncated α-CTD was used (36). This suggested that DNA bending is important for activation, which is significant because studies with the well-studied CAP have shown that the structure of the DNA-CAP-RNAP transcription complex has an 80° bend in the center of the CAP binding site, as determined by Hudson et al. (34).

Residues in the α-CTD have been implicated in activation in many systems and can be mapped to various regions of the molecule, using the tertiary structure determined from the nuclear magnetic resonance (NMR) solution structure of α-CTD by Jeon et al. (37). Early experiments suggested that mutants defective in class I CAP activation had mutations in the region from aa 265 to 270 (38). Some of these residues and various others were implicated in activation by a range of activators, as reviewed by Ishihama (32). Similarly, in vitro experiments with TyrR, using a limited collection of α-CTD mutants, identified a number of amino acids with some role in activation of mtr by TyrR (36). However, it now seems likely that many of these mutant proteins are altered in the interaction of the TyrR–α-CTD complex with DNA or with σ⁷⁰, rather than in the interaction with TyrR protein itself. This also seems likely in the case of many previously identified activation-negative rpoA mutants for other systems.

It has since been shown that in the case of activation by CAP, three different regions of α-CTD are involved in activation (39–41). The so-called 265 determinant (residues R265, N268, N294, G296, K298, S299, and E302) interacts with the DNA immediately downstream of the CAP site, the 261 determinant (residues V257, D258, D259, and E261) interacts with σ⁷⁰ region 4 of RNA polymerase, bound at the −35 element, and the 287 determinant (residues T285, E286, V287, E288, L289, G315, R317, and L318) interacts with CAP (34, 40–42).

A detailed model for the complex of CAP, the α-CTD of RpoA, the σ subunit (region 4), and DNA upstream of the promoter has been proposed, based on combining structures of single components (42), and a high-resolution cryo-electron microscopy (cryo-EM) structure of the transcription complex was determined more recently (43). Comparison between the three-dimensional (3D) structures of FNR and CAP suggest that the interactions are very similar, suggesting that the model for CAP activation might be a general one (44).

Many previous studies in TyrR and in other systems, purporting to identify Act⁻ RpoA mutants or suppressor mutations, in fact isolated mutants of the 265 determinant (the site for DNA binding) or of the 261 determinant (the site for binding to σ⁷⁰ region 4 of RNA polymerase). However, RpoA mutants with changes in the FNR activator binding site have been successfully isolated (44). Isolation of such mutants provides sound evidence for the interaction between the activator and RpoA.

The aims of this study were to characterize the sites within both RpoA and TyrR that are directly involved in the activation of tyrR-regulated genes and to isolate mutations within TyrR that suppress the Act⁻ RpoA mutations. Such results would provide strong evidence for the interaction between TyrR and RpoA. The availability of structures for both the TyrR NTD (23) and the α-CTD of RpoA (37) has made it possible to define the activation sites and to gain some insight into how activation might occur.

RESULTS

Development of a selection strategy for isolation of RpoA Act⁻ mutants.

In order to isolate mutants in the α-CTD of RNA polymerase altered in activation by CRP, Zou et al. (38) used the pLAMC library carrying rpoA with mutations in the CTD, selecting transformants in a strain with a wild-type rpoA chromosomal allele using complete medium. isopropyl-β-d-thiogalactopyranoside (IPTG)-induced overexpression of a mutant rpoA allele provides enough mutant subunit to effectively outcompete the wild-type α-subunit in the formation of holoenzyme, while in the absence of IPTG, the chromosomal rpoA⁺ allele determines the phenotype.

We were seeking rpoA mutants encoding α-subunits specifically altered in their interaction with TyrR in response to the amino acids tyrosine and phenylalanine, so minimal medium seemed a logical choice. However, control experiments showed that this would be impossible, because elevated levels of the α-subunit (even with only 0.02 mM IPTG) caused major growth defects. Perhaps high levels of α-subunit cause binding of the α-subunit to free and bound RNA polymerase molecules and to regulators, making normal activation impossible. Addition of branched-chain amino acids and methionine improved growth, but activation of mtrA-lac by TyrR was still reduced significantly.

To solve the selection problems with overexpression of RpoA⁺ and potential mutant RpoA proteins, a strategy involving a conditional mutant was devised. A chromosomal rpoA(Ts) allele, kindly provided by A. Ishihama, was introduced into the tyrR parent strain by transduction (see Materials and Methods). Carrying out selections at 42°C eliminates the need to outcompete the chromosomally encoded wild-type α-subunit, and a suitable medium for selection was identified using the rpoA(Ts) host strain carrying various plasmids (see Materials and Methods).

Isolation of an rpoA mutant with defective activation by the TyrR protein.

Transformations were carried out using the pLAMC library, which contains plasmids similar to pLAW2 but with PCR-generated mutations in the α-CTD of rpoA (38), and the new host strain, JP10932 (rpoA101 tyrR366 ΔlacU169 recA56), carrying mtrA-lacZ as the indicator plasmid and pMU1065 (multicopy tyrR⁺).

The mtr gene is activated by TyrR with phenylalanine or tyrosine and is repressed by TrpR with tryptophan (45). The indicator plasmid mtr-lac carries the mtr regulatory region, including a T-rich region (TTTTTTT), immediately upstream of the promoter (45). In the mtrA-lac fusion, this T tract has been altered. Previous results have shown that with the mtrA-lac fusion, the tyrR⁺-activated specific activity (SA) in minimal medium (MM) plus tyrosine or phenylalanine is about the same as in wild-type (WT) mtr-lac, but the SA in a tyrR or a tyrR Act⁻ strain is significantly lower (around 4- or 5-fold lower), increasing discrimination between the WT and the Act⁻ mutant. The T tract, possibly together with a second T tract (TTTTT) one helical turn upstream, may act as an UP sequence (30, 46) in a tyrR mutant, but activation by TyrR with phenylalanine or tyrosine can achieve the same outcome. In rich medium, this effect was less pronounced but still significant.

The selection was for Ap^r transformants on MacPhe medium containing IPTG at 42°C (see Materials and Methods). Approximately 2,000 colonies were screened on this medium, and 16 that appeared paler were selected.

Three promising strains were purified, checked, and, together with the pLAW2 (rpoA⁺) derivative as a control, grown in Luria broth containing 0.2% glucose, 4 mM phenylalanine, 0.3 mM IPTG, ampicillin, kanamycin, and trimethoprim at 42°C. The results are shown in Fig. 2.

FIG 2.

Specific activity of Act⁻ rpoA mutants for lacZ indicator fusions. Two mutant strains, MS1 and MS3, and the pLAW2 (rpoA⁺) derivative as a control, all in the JP10932 (rpoA101 tyrR366 ΔlacU169 recA56) background, with pMU1065 (multicopy tyrR⁺), and carrying various lacZ indicator fusion plasmids. The top left panel shows results for mtrA-lacZ in both tyrR⁺ and tyrR derivatives. Other panels show specific activities using a range of different indicator fusions in both trpR⁺ and trpR backgrounds. Strains were grown in Luria broth containing 0.2% glucose, 4 mM phenylalanine, 0.3 mM IPTG, ampicillin, kanamycin, and trimethoprim at 42°C.

Specific activities (see Materials and Methods) were relatively low in these experiments because growth was in Luria medium. Rapid growth results in more rapid dilution of cellular proteins and in multiple, bidirectional replication forks, from the origin at 83 min, near ilvE, to the terminus at 27 min, close to tyrR. Depending on position and relative gene dosage, some gene frequencies increase and some decrease (47). Plasmids replicate independently, but a decrease in copy number and gene dosage relative to the total of all chromosomal gene frequencies would not be unexpected.

Assays for β-galactosidase activity expressed from mtrA-lacZ showed that in MS1 and MS3, activation (relative to tyrR366) was reduced from 4.5-fold in the case of rpoA⁺ to 0.8- and 0.9-fold in the case of the mutants (Fig. 2, top left). There was no decrease in SA in tyrR strains, confirming that the changes were TyrR related and that constitutive expression was unaltered under these conditions.

When mutant plasmids MS1 and MS3, and pLAW2 as a control, were introduced into strains carrying mtr-lacZ fusions to determine whether the altered α-subunit was also affected in its ability to activate the wild-type mtr promoter, activation again decreased, from 11-fold to 4.7-fold for mtr-lacZ, in the mutant but not in the tyrR strain. Since mtrA-lac and mtr-lac can also be regulated by TrpR (45) the mutant rpoA plasmids were also tested in trpR strains. Introduction of the plasmids into a trpR strain containing mtrA-lacZ or mtr-lacZ demonstrated that the difference was independent of trpR, with activation again reduced in the mutants, from 12-fold to 3.7-fold for mtrA-lacZ and from 3.1-fold to 1.7-fold for mtr-lacZ (Fig. 2).

To determine whether the altered α-subunit was also affected in its ability to activate other TyrR-modulated genes, the mutant plasmid MS1 and pLAW2 were introduced into strains carrying tyrP₊₄-lacZ and aroF-lacZ fusions. Activation of tyrP₊₄-lacZ decreased from 3.1- to 1.3-fold in tyrP₊₄-lacZ. Repression of an aroF-lacZ fusion was not diminished, indicating that TyrR retained its ability to repress; indeed, aroF expression increased from 65-fold to 101-fold (relative to tyrR), probably representing loss of a small amount of activation, as observed previously for Act⁻ tyrR mutants (Fig. 2).

DNA sequencing revealed that MS1 and MS3 carried identical mutations in the α-subunit of RNAP, both DN250, a change of an aspartate to an asparagine residue at position 250. This residue is marked in Fig. 3 in a schematic diagram of the α-CTD, as determined by Jeon et al. (37).

FIG 3.

Structural model of the CTD of the α-subunit of RNA polymerase, shown in cartoon representation. The locations of important amino acids are indicated as spheres of main-chain atoms, with residue type and number labeled. The TyrR determinant is indicated, as are the 265 determinant, which interacts with DNA, and the 261 determinant, which interacts with region 4 of the sigma subunit of RNA polymerase. The structure is from the data of Jeon et al. (37), accessed from the Protein Data Bank (accession number 1COO), and is presented using the PyMOL molecular graphics system, version 2 (Schrödinger, LLC).

Identification of additional Act⁻ RpoA mutants using in vitro mutagenesis.

Examination of the three-dimensional structure of α-CTD shown in Fig. 3 revealed that D250 lies quite close to the linker between the carboxyl end of the protein (α-CTD) and the NTD.

Residues that lie close to D250 include R310 and F249. Mutant plasmids with mutations DN250, RD310, RE310, FA249, and FR249 were constructed and transformed into derivatives of JP10932 carrying mtrA-lacZ, with and without the tyrR⁺ plasmid pMU1065, and assayed as before.

The results in Fig. 4 show that, while there are no significant differences in promoter activity in the tyrR background, the mutant alleles have a pronounced affect on activation. Activation relative to LAW2 (rpoA⁺) is decreased by 70% in the DN250 mutant, by 60% in DR250 and RD310 mutants and by 80% in RE310 mutants. The 4.5-fold activation (relative to tyrR) observed with RpoA⁺ decreased to 1.6-fold in RE310 mutants. Activation in FA249 and FR249 mutants showed a slight and possibly nonsignificant increase. As before, similar results were obtained when these experiments were repeated in a trpR strain and in experiments using the mtr-lacZ fusion (data not shown).

FIG 4.

Activation by TyrR in a range of rpoA mutants. Derivatives of the tyrR host strain JP10932 containing an mtrA-lacZ fusion and rpoA mutant plasmids (as indicated), with and without PMU1065 (mc tyrR⁺), were grown in Luria broth plus 4 mM phenylalanine, 0.2% glucose, 0.3% IPTG, and antibiotics as appropriate and assayed for β-galactosidase specific activity. White bars, mc tyrR⁺ derivatives; gray bars, tyrR mutants.

Activation by the chromosomal rpoA(DN250) mutation.

The strain in which the wild-type rpoA gene on the chromosome was replaced by the DN250 mutation was constructed primarily to assist in the isolation of suppressor mutations, since the identification of the partially suppressed phenotype proved to be difficult in rich medium with high IPTG. It also proved to be invaluable in the isolation and analysis of new mutants encoding Act⁻ RpoA and Act⁻ TyrR proteins. In this background, minimal medium with a low level of IPTG could be used, producing an amount of plasmid-encoded RpoA closer to normal.

Enzyme assays showed that the difference in fold activation (comparing tyrR⁺ with tyrR) is indeed much greater in the chromosomal rpoA(DN250) background with rpoA alleles expressed from the plasmid in minimal medium with low levels of IPTG than in the rpoA(Ts) strains with the same plasmids, grown at 42°C in rich medium with high IPTG.

For mtrA-lacZ, activation of β-galactosidase (tyrR⁺ versus tyrR) in MM plus Tyr is around 114-fold in the chromosomal rpoA(DN250) strain carrying the rpoA⁺ plasmid pLAW2 (Fig. 5), compared to the 4.5-fold activation seen previously in rich medium (Fig. 2). The activation of the indicator fusion tyrP₊₄-lacZ in MM plus Phe was increased 48-fold (Fig. 5), compared to the previous 3.1-fold (Fig. 2).

FIG 5.

Suppression of RpoA(DN250) by tyrR alleles. The specific activity of β-galactosidase is shown for strains carrying tyrR alleles capable of partial suppression of RpoA(DN250) or RpoA(RE310), and a suitable tyrR⁺ control plasmid, in an mtrA-lacZ/JP13002 background [rpoA(DN250) tyrR366]. Cells were grown in minimal medium plus tyrosine.

The results in Fig. 5, obtained with the new strains and media, show that the 114-fold activation of mtrA-lacZ expression seen with wild-type rpoA (pLAW2) is reduced to 1.3-fold by the plasmid carrying rpoA(DN250). For tyrP₊₄-lacZ, the 48-fold activation seen with rpoA (pLAW2) is reduced to 1.1-fold by rpoA(DN250). As expected, there was no defect in repression of aroF-lacZ (results not shown).

Isolation of additional Act⁻ α-CTD RpoA mutants using an aroP1-lacZ indicator fusion.

The aroP1-lacZ indicator fusion, pMU6260, exhibits increased β-galactosidase expression with loss of activation at aroP3, producing an easier phenotype to identify. TyrR, when bound to the cofactor tyrosine, phenylalanine, or tryptophan, binds to aroP3 on the opposite strand of the DNA helix, resulting in repression of aroP1 expression. SA is reduced to 11, 14, and 65 units, respectively, compared to 231 in MM, and to 294 in a tyrR mutant (15).

The new strain carrying DN250 on the chromosome was chosen as the host, so that it would be possible to isolate and analyze Act⁻ rpoA mutants using minimal medium and a lower level of IPTG, circumventing difficulties arising from the need for high levels of expression of the mutant protein to compete with wild-type RpoA. The pLAMC library was used in transformations with pMU6260/pMU1065/JP13002 (DN250 tyrR Δlac recA) and yielded several mutants that were blue on MM containing X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) plus Tyr and red on MacPhe medium. They were compared with control strains carrying either multicopy (mc) rpoA(RE310) or pLAW2 (rpoA⁺) and were similar in phenotype to the Act⁻ control with normal repression, namely, sensitive to 3-fluorotyrosine (FT) and to 1 mM phenylalanine and moderately resistant on β-thienylalanine medium (see Materials and Methods). Like both the wild type and the rpoA(RE310) control, they were able to grow on plates containing 0.5 M NaCl, unlike strains lacking the α-CTD of RpoA (e.g., Δ256), since the α-CTD plays a role in the response to osmolarity (48).

Sequencing revealed that three mutant plasmids with an RpoA Act⁻ phenotype encoded RpoA with the mutations DN250, GD279 IT275, and EG288. This confirmed the importance of D250 and suggested that EG288 and GD279 or IT275 might also interfere with activation. Targeted mutagenesis to construct the EG288 strain yielded a EG288 PP293 mutant, which also had the Act⁻ α-CTD phenotype.

To isolate additional Act⁻ alleles of D250 and R310, libraries of rpoA plasmids containing amino substitutions at position R310 or D250 isolated by in vitro mutagenesis (using RX310 or DX250) were used in transformations with pMU6260/pMU1065/JP13002 (250DN tyrR Δlac recA), as described above. The phenotypes of promising transformants were tested, and those with an Act⁻ phenotype were sequenced, resulting in the identification of DT250 and RS310.

These experiments showed that the DT250, mutant, as well as the DN250 and DR250 mutants (Fig. 4) was Act⁻ and the RS310 mutant, as well as the RD310 and RE310 mutants (Fig. 4), was Act⁻; they also suggest a possible role for E288.

Isolation of tyrR mutants that suppress rpoA(DN250).

Isolation of mutations in tyrR (NTD of TyrR) that suppress the activation-negative rpoA mutations (α-CTD of RpoA) would provide strong evidence for a direct interaction between specific areas of these two protein domains, facilitating the elaboration of a model for activation.

A number of experiments were carried out to isolate tyrR mutants that suppress the chromosomal α-subunit mutation, DN250, initially using mtrA-lacZ as an indicator plasmid. A library of tyrR mutants where the tyrR NTD had been mutated by growth in a mutator strain and then cloned into pSU39tyrR⁺ was transformed into an mtrA-lacZ/JP13002 (DN250 tyrR lacΔ recA) host strain. Selections were made on MacPhe medium containing 4 mM Phe with kanamycin (Km) and trimethoprim (Tp).

The few colonies that were redder than the majority were purified and were then screened on MacPhe, MM with X-Gal plus Phe, MM with X-Gal plus Tyr, and MM plus 3-FT, alongside appropriate control strains, including the tyrR⁺ rpoA(250DN) strain, a tyrR⁺ rpoA⁺ strain, and a tyrR rpoA⁺ control strain.

The first experiment resulted in the isolation of a tyrR(EK139) mutant. When it was grown in MM plus Phe or Tyr at 37°C, there was significantly increased activation of mtrA-lacZ in the rpoA(DN250) background. It can be seen in Fig. 5 that β-galactosidase levels in the EK139 mutant were 460%, relative to the tyrR⁺ strain grown in tyrosine, and activation (relative to tyrR) increased from 1.3-fold to 4-fold. In the rpoA⁺ control strain, there was no significant increase in EK139 relative to the tyrR⁺ strain. A similar result was obtained with an EK139 mutant constructed by in vitro mutagenesis but not with an EA139 mutant (results not shown).

When tested with a tyrP₊₄-lacZ fusion, activation in the tyrR(EK139) mutant was almost double (190%) that of the tyrR⁺ strain in the rpoA(DN250) background. There was a small increase of about 20% in the rpoA⁺ strain (results not shown).

These results indicate that the change from glutamate to lysine at residue 139 in TyrR increases activation in the rpoA(DN250) strain. It can be seen in the structural diagram of the TyrR NTD that residue E139 is contiguous with other known residues implicated in activation and is in an exposed position. It lies within the PAS domain situated between residues 89 and 175 (23).

In a second experiment using a different TyrR NTD mutant library, several isolates were identified with elevated β-galactosidase levels on MacPhe, MM with X-Gal plus Phe, and MM with X-Gal plus Tyr. They were less Phe sensitive than Act⁻ controls (as expected) but were still fully sensitive to 3-fluorotyrosine, indicating normal repression.

Enzyme assays confirmed that all the new isolates had modestly elevated levels of β-galactosidase in the presence of tyrosine. All showed an increase over tyrR⁺; the best was identified by sequencing as RG119, showing a reproducible 43% increase in activation, from 201 to 287 (Fig. 5). This corresponded to a 2.0-fold increase relative to the tyrR control, compared to the 1.4-fold increase in the tyrR⁺ strain. In the rpoA⁺ control strain, RG119 did not cause an increase in activation.

The RC119 allele, isolated as a partial suppressor of RE310 (see below), is also included in Fig. 5. It is not a suppressor of rpoA(DN250) but is Act⁻ when present with rpoA⁺. Note that specific activities in Fig. 5 are for strains that are haploid for rpoA alleles; strains in Fig. 6 carry the rpoA alleles on a plasmid and are multicopy.

FIG 6.

Suppression of RpoA(RE310) by tyrR alleles. The specific activity of β-galactosidase is shown for strains carrying plasmid-borne tyrR alleles capable of partial suppression of RpoA(RE310) or RpoA(DN250), with a tyrR⁺ plasmid as a control, in mtrA-lacZ/JP13002 (DN250 tyrR Δlac recA background), carrying RpoA(RE310) or RpoA⁺ plasmids. Strains were grown in MM Tyr supplemented with antibiotics as required.

It can be seen from Fig. 1 that R119 is in the PAS domain, adjacent to D103, with a known role in activation.

Isolation of tyrR mutants that suppress rpoA(RE310).

The isolation of the activation-negative rpoA RE310 allele provided another target for selection of suppressors in the tyrR NTD. Experiments were carried out using overexpression of RE310 in the strain with the chromosomal DN250 mutation, using the mtrA-lacZ fusion as an indicator. While two rpoA alleles are present in this strain, most of the α-subunit comes from the multicopy plasmid expressing IPTG-induced RpoA(RE310).

Mutant tyrR libraries were used to search for possible suppressors of rpoA(RE310), with selection on MacPhe medium containing IPTG. Several red mutants were screened on MacPhe, on MM plus Tyr or Phe, and on FT and had an Act⁻ but repressible TyrR phenotype. Strains were grown for assays in MM plus Met with 0.3 mM IPTG plus Km, Ap, and Tp.

Sequencing indicated that one suppressor was RC119, a mutation in the same residue as the suppressor of DN250. Specific activities of β-galactosidase in Fig. 6 show that the tyrR RC119 allele partially suppresses rpoA RE310, with 45% more activation in the RC119 derivative than in the tyrR⁺ control. It fails to suppress the rpoA DN250 allele (Fig. 5) and is Act⁻ in rpoA⁺ strains. The RG119 allele (isolated as a suppressor of DN250) also suppresses RE310 slightly.

Other tyrR mutants that showed some suppression of α-CTD (R310) were identified as SG100 (where S100 lies close to D103 and R119) and RA77 (where R77 on the opposite strand of the TyrR NTD dimer lies close by, within the linker between the ACT and PAS subdomains) (Fig. 1).

In an approach to seek tyrR suppressors that were better able to restore activation of rpoA(RE310), R119 was targeted directly using a tyrR(RX119) library generated by in vitro mutagenesis. A tyrR library made using an oligonucleotide with mixed bases (NNN) for the codon for R119 (RX119) was used in transformations similar to those described above.

These experiments yielded two new mutants. The RL119 HQ123 double mutant was isolated on MM with X-Gal plus Tyr, and the RP119 mutant was isolated on MM with X-Gal plus Phe. It can be seen in Fig. 6 that both alleles cause a 2.7-fold increase in SA in tyrosine in the strain carrying multicopy rpoA(RE310) relative to the tyrR⁺ control. The same 2.7-fold increase was observed in the presence of tryptophan (where TrpR represses expression), in the presence of either tyrosine or phenylalanine (results not shown). In the presence of rpoA⁺ (mc), the RL119 HQ123 and RP119 mutants are both defective in activation (7.5-fold lower than with tyrR⁺). Screening in strains with the tyrP₊₄-lac fusion gave results similar to those for the mtrA-lacZ fusion. With pMU6260 (aroP1-lacZ fusion), the results were reversed, as expected since the activation function results in “repression” of aroP1 expression (see above).

The new alleles RC119, RP119, and probably RL119 are therefore suppressors of the α-CTD mutation RE310 and are also tyrR Act⁻ alleles in an rpoA⁺ strain. It seems likely that the significant feature of these mutants might be the loss of the positively charged, bulky arginine side chain. SG100 and RA77 are also new suppressors of α-CTD(RE310) (see above).

Isolation of new Act⁻ tyrR mutants in the PAS domain.

Previously, Act⁻ mutations with changes in the PAS domain had been isolated in residues P92, V93, S95, and D103 (26). This paper has already described roles in activation for E139, R119, S100, and R77. Examination of the model for the NTD of TyrR (Fig. 1) indicates that these residues lie in close proximity to each other.

To further define the patch within the TyrR NTD responsible for interaction with RpoA to bring about activation, new tyrR mutants with the mutations DA97, KD101, KA101, DR118, RD121, and EA141 were constructed by in vitro mutagenesis and cloned into the medium-copy-number vector pSu39. The new tyrR plasmids were transformed into the tyrR366 mutant host JP8042, carrying an mtrA-lacZ fusion, for analysis of activation. The results in Fig. 7 show that these mutants all exhibited decreased activation, with DA97, KD101, and DR118 being the most affected.

FIG 7.

Activation in tyrR mutants. The specific activity of β-galactosidase is shown for tyrR alleles isolated using oligonucleotide-directed mutagenesis and a tyrR⁺ control in an mtrA-lacZ/JP8042 background. Strains were grown in MM Tyr and MM Phe supplemented with kanamycin and trimethoprim.

Figure 1 shows that these residues help define the extent of the activation patch on the TyrR NTD, with roles for D97, K101, D118, R121, and E141 in addition to E139, R119, S100, and R77 (this paper) and the previously identified residues S95 and D103 (26).

An additional Act⁻ TyrR allele, the RC77 allele, was isolated in transformations using the aroP1-lacZ fusion with pMU6260/JP8042, using a library generated in a mutator strain. Colonies were selected on MM containing Met, X-Gal, Tyr, Km, and Tp. The RC77 mutant showed decreased expression of β-galactosidase in the presence of phenylalanine or tyrosine, was growth inhibited by phenylalanine, and remained completely sensitive to 3-FT. While R77 is within the linker between the ACT and PAS subdomains, the residue on the opposite strand lies relatively close to D103 and D97. RA77 had already been identified as a partial suppressor of RE310.

DISCUSSION

The TyrR regulon, with its wide range of mechanisms capable of achieving favorable regulatory outcomes, has been an invaluable target of inquiry and a source of novel opportunities (1). This paper focuses on a missing piece of the puzzle—the nature of the activation process involving the interaction between the N-terminal domain of TyrR and the α-CTD of RNA polymerase.

The isolation of rpoA mutations with changes in residues D250 and R310 with the same phenotype as Act⁻ tyrR mutants, and no obvious pleiotropic effects, has defined a patch in the α-CTD involved in activation by TyrR (Fig. 3). These residues lie close together and are surface exposed (43, 49). The residues D250 and R310 contribute to the TyrR determinant and have not previously been implicated in interactions with other activator proteins or in UP interactions. They lie adjacent to the site proposed for CAP interaction, the 287 determinant (34, 40–42), with E288 possibly playing a role in both sites.

The activation patch in TyrR (NTD) for interaction with α-CTD is situated in the PAS domain in the base of domain B and in the linker. Mutations in contiguous residues in the TyrR PAS domain (V93, V95, D97, K101, D103, D118, R119, R121, and E141) cause defects in activation (Fig. 1).

Isolation of suppressors in the TyrR NTD of the mutations DN250 and RE310 in the α-CTD provides strong evidence for interaction between the TyrR NTD and the α-CTD. TyrR(EK139) partially suppresses DN250; RG119 partially suppresses DN250 and RE310, and RC119, RP119, SG100, and RA77 partially suppress RpoA(RE310), all suggesting protein-protein contact. These residues either overlap or are in close proximity to the TyrR Act⁻ alleles and are part of the activation patch (Fig. 1).

A model for the docking interaction was constructed by placing TyrR EK139 at an appropriate distance from α-CTD D250 and rotating one with respect to the other, looking for a suitable fit and proximity to identified residues. This model for docking is shown in Fig. 8. It seems likely that D250 in the α-CTD interacts with EK139 in the TyrR NTD and that R310 interacts with D97 or K101 via salt bridges, either directly or in a solvent-mediated manner. The TyrR residues R119 and possibly R77, in which suppressor mutations were identified, might be from the other strand of the TyrR dimer (Fig. 8, in green) or may affect structure, perhaps by altering the orientation of the two TyrR strands with respect to each other. In this diagram the 265 determinant lies in the upper region of the α-CTD, where it would interact with DNA upstream of the −35 region (Fig. 8, toward the top right of the diagram), positioning the α-CTD to interact with region 4 of σ⁷⁰, which would sit on the right of the diagram, close to the DNA.

FIG 8.

Docking model for interaction between the TyrR NTD and the α-CTD of RNA polymerase, created by P. Carr at the John Curtin School of Medical Research. The TyrR NTD and α-CTD of RpoA are both shown in cartoon form, with important amino acid residues involved in activation indicated as spheres of main-chain atoms, with residue type and number labeled. The TyrR NTD structure is from the Protein Data Bank (accession number EBI-30882) (23). The structure for α-CTD is from the work of Jeon et al. (37), accessed from the Protein Data Bank (accession number 1COO), and the diagram was constructed using the PyMOL molecular graphics system, version 2 (Schrödinger, LLC).

Previous in vitro studies using mutant proteins identified a number of residues of the α-CTD essential for TyrR-mediated activation at two sites (36). It is now clear that the first corresponds to the 265 determinant and is involved in interaction with DNA, as in the lac system (39, 41), and in regulation involving the UP element (50). The second site corresponds to the 261 determinant and is involved in protein-protein contact with σ⁷⁰ region 4, bound at the −35 element of the promoter, as in the lac system (39, 41), bringing about activation (51). A strain carrying mutations affecting E261 (52) was shown to exhibit defects in TyrR-mediated activation in vivo (our unpublished results).

It now seems that the well-studied picture of activation determined for activation of transcription by CAP might be representative of many systems. The details for this interaction were based first on evidence for protein-protein contact between the activator and the α-CTD, originally from cross-linking studies (53) and subsequently from the three-dimensional EM structure of the CAP-dependent transcription initiation complex, as determined by Hudson et al. in 2009 (34) and by Liu et al. in 2017 (43), which clarified the roles of the various determinants. The three-dimensional (3D) picture of the CAP transcription initiation complex also indicates the extent of bending of the DNA, centered on the activator binding site, and the way in which the DNA helix wraps around the complex.

In the lac system, protection of the region between the CAP binding site and the promoter requires both the activator and the α-CTD (54). However, changes in sequence have little effect on expression (55). Photo-cross-linking studies with RNA polymerase and the lacUV5 promoter also demonstrate binding, specifically in the minor groove (56). When CAP binds to DNA upstream of the promoter, it produces a bend of around 80° and may help to narrow the minor groove and facilitate the interaction with the α-CTD, such that the interaction between the 265 determinant and the DNA can take place. The requirement for structural integrity of the DNA helix in this region, whereby 4-nucleotide single-stranded gaps affect expression (57), may reflect a requirement for a narrowed minor groove.

Similarly, when the 265 determinant interacts with the UP element, it binds specifically to the narrowed minor groove, which results from curvature brought about by the A/T tracts present in the UP element. The α-CTD recognizes the backbone structure of the compressed minor groove, with Arg265 and Asn294 interacting with the DNA backbone in the minor groove and Val264, Gly296, and Lys298 interacting with the phosphate backbone outside the minor groove. Arg265 may interact with groups on both sides of the minor groove (58).

The involvement of residues that are part of the well-characterized DNA-binding 265 determinant in activation by TyrR (36) and the involvement of an A/T rich upstream region in mtr expression support a model where interaction of the α-CTD with DNA upstream of the promoter is an intrinsic part of the activation process.

Perhaps, in all these cases, the interaction of the 265 determinant with the narrowed minor groove might be envisaged as analogous to the tines of a fork being used to immobilize a coiled telephone cord. This interaction might then position the α-CTD for interaction of the 261 determinant with σ⁷⁰ region 4. The close interaction between RNA polymerase and DNA at two points (one where the α-CTD interacts with upstream DNA and the other in the −10 region of the promoter) might enable torsional forces to facilitate opening of the DNA helix in the intervening region, leading to increased isomerization. Such a role for the α-CTD fits well with models for the initiation of transcription, where it is proposed that there is extensive wrapping of upstream DNA around the RNA polymerase (59, 60).

It appears that for class I activation, three sites on the α-CTD are necessary, one of which is for binding of an activator (35). This site may be common to different activators, may overlap, or may be unique, with the specificity of activation residing primarily on the presence of specific operator loci, rather than in unique interactions between the activator and the α-subunit.

Activation by TyrR conforms to the model for activation of lac by CAP; the activation patch on TyrR interacts with the TyrR determinant on α-CTD, promoting the interaction of the 265 determinant with DNA, which is followed by the interaction of the 261 determinant with σ⁷⁰ region 4, to bring about activation.

MATERIALS AND METHODS

Bacterial strains, plasmids, and bacteriophage.

The bacterial strains used in this study were all derivatives of E. coli K-12. Their relevant characteristics are listed in Table 1.

TABLE 1.

Strains, bacteriophages, and plasmids used in this work

Strain, bacteriophage, or plasmid	Relevant characteristics	Source and/or reference(s)
Strains
JP8042	ΔlacU169 recA56 tyrR366	25
JP10923	rpoA101(Ts) zhc::Tn10	This paper
JP10927	rpoA101(Ts) Tets	This paper
JP10932	ΔlacU169 recA56 tyrR366 rpoA101(Ts) Tn10::recA	This paper
JP13001	rpoA(DN250) tyrR+ recA56 ΔlacU169	This paper
JP13002	rpoA(DN250) tyrR366 recA56 ΔlacU169	This paper
JP13003	ΔlacU169 recA56 tyrR366 rpoA101(Ts) trpR Tn10::recA	This paper
RLG731	λc1857 rrnB P1-lacZ	R. Gourse
HN315	rpoA101(Ts)	A. Ishihama (70, 71)
E4291	zhc::Tn10	Rowland et al. (72)
MH513	(ompF′-lac)16–13 λpl209	Slauch et al. (73)
XL1-Red	endA1 gyrA96 hsdR17 mutD5 mutS mutT	Stratagene
Bacteriophages
M13 tg130	M13 with lacPOZ′ and a multicloning site	74
mpMU36	2.2-kb tyrR fragment in M13 tg130	26
Plasmids
pMU1065	Kmr; pACYC177 with tyrR+ gene in HindII site	26
pMU3190	Tpr; plasmid with mtr-lacZ transcriptional fusion	25
pMU5304	Tpr; plasmid with mtrA-lac transcriptional fusion	45
pMU6260	Tpr; plasmid with aroP1-lac transcriptional fusion	15
pLAW2	PINIIIA1 vector overexpressing the rpoA+ gene, under the control of the lpp-lac promoter	A. Ishihama (38)
pLADΔ256	Similar to pLAW2, but truncated, removing sequence from aa 256	A. Ishihama (38)
pLAMC library	Plasmid library, similar to pLAW2, but carrying random mutations in the α-CTD of rpoA	A. Ishihama (29)
pMU6219	pSU39tyrR+ (with HindIII site at codon 144)	28
pAM34	ColE1 type, replication controlled by IPTG	65

Minimal media.

Minimal medium (MM) used was prepared from half-strength buffer 56, described by Monod et al. (61), and was supplemented with appropriate growth requirements, including 0.2% glucose and thiamine, with the addition to 0.1 mg/ml l-methionine, 0.05 mg/ml l-leucine, 0.05 mg/ml l-isoleucine, 0.05 mg/ml l-valine, and 0.1 mg/ml l-threonine as indicated. Methionine was included in all selective MM for strains with high levels of free α-subunit to avoid methionine starvation, probably caused by interaction between MetR (52) and free RpoA. For experiments testing repression or activation, 1 mM or 2 mM l-phenylalanine, 1 mM or 2 mM l-tyrosine, and 1 mM l-tryptophan were added as indicated. IPTG was added at the concentration indicated, and X-Gal was used at a concentration of 25 μg/ml. In MM, antibiotics were added as required at concentrations of 25 μg/ml for ampicillin, 10 μg/ml for trimethoprim and 20 μg/ml for kanamycin. The repressible TyrR⁺ phenotype was tested by sensitivity to the tyrosine analogue 3-fluorotyrosine (0.2 mM) in MM and by growth on MM containing the phenylalanine analogue β-thienylalanine (5 mM), shikimic acid (0.3 mM), and tyrosine, a medium developed by H. Camakaris for positive selection of tyrR⁺(62, 63). The probable explanation is that tyrosine represses aroP, reducing uptake of the analogue, with shikimic acid decreasing any effects on the first step of the aromatic pathway. In rich media, ampicillin was added at 50 μg/ml, trimethoprim at 40 μg/ml, and kanamycin at 15 μg/ml.

Construction of an rpoA(Ts) host strain.

zhc::Tn10 was transferred from E4291 into HN315, the strain carrying the temperature-sensitive rpoA101 allele, by P1 transduction. Using a lysate prepared on this strain, Tn10::rpoA101 was introduced into JP7744, a tyrR366 ΔlacU169 strain, by transduction, where tyrR is a deletion and null mutation. After isolation of a tetracycline-sensitive derivative (64), Tn10::recA56 was introduced by transduction to create JP10932.

Selection in the rpoA(Ts) host strain.

To develop suitable media, the rpoA(Ts) host strain JP10932 carrying pLAW2 (multicopy [mc] rpoA⁺), and either the mc tyrR⁺ plasmid pMU1065 (mc tyrR⁺) or a related plasmid carrying the Act⁻ mc tyrR allele RQ10 was used. The tyrR RQ10 mutation in the TyrR amino acid binding site (ACT domain) is defective in activation, so it presents the phenotype expected for the mutants we hoped to isolate, effectively acting as a surrogate. As in previous work, the indicator fusions mtr-lacZ and mtrA-lacZ were used (26). MacPhe medium was based on MacConkey agar (Oxoid), a rich medium containing lactose and a pH indicator to monitor fermentation. MacPhe also contained 4 mM l-phenylalanine and 0.3 mM IPTG, and required antibiotics were added as indicated. The strain with the activation-defective tyrR allele and pLAW2 grew well and was clearly less red than the corresponding strain with the wild-type tyrR⁺ allele. It was anticipated that selected mutants would not have lost RpoA function entirely, since controls that contained a truncated rpoA (plasmid pLADΔ256) grew very poorly. This is not unexpected, since the α-CTD interacts with activators of genes involved in biosynthesis and other cellular functions and also activates by binding UP sequences, necessary for expression of many genes, including the ribosomal genes rrnB and rnnD, and the plasmid gene encoding the RNA II primer (30, 50).

Construction of the chromosomal DN250 mutant.

A disruption plasmid was constructed by amplifying rpoA(DN250) from MS1, the pLAW2-related plasmid in a strain isolated as defective in activation by TyrR (see Results). It was cut with XbaI, and the rpoA(DN250) fragment was amplified using primers with BamHI sites (Table 2). The PCR fragment was digested with BamHI, and the insert was cloned into the multicopy plasmid pAM34 cut with BamHI. The ligated plasmid was transformed into JM101, selecting Ap^r colonies in the presence of 0.5 mM IPTG, as the replication primer RNA is under the control of the lac promoter/operator (65). This disruption plasmid was transformed into JP10927 [lacΔ tyrR⁺ rpoA(Ts) strain] carrying an mtrA-lac plasmid, selecting first for a single-crossover event producing Ap^r colonies that were temperature sensitive and red (activated) on MacPhe medium. In a second step, temperature-resistant colonies were isolated, and colonies that had become Ap^s and yellow on MacPhe (not activated) were selected as possible rpoA(DN250) derivatives. The presence of the DN250 mutation was verified by sequencing. After constructing this rpoA(DN250) tyrR⁺ lacΔ strain, cotransduction with Tn10 was used to introduce tyrR366, followed by Tet^s selection (64). The recA56 allele was then introduced by cotransduction with Tn10 to make the rpoA(DN250) tyrR366 lacΔ recA56 host strain.

TABLE 2.

Oligonucleotides used in this study

Gene	Mutation	Sequence of primer or oligonucleotide	Restriction site
tyrR	DA97	CTTTTCATGGCGACAGAGA
tyrR	DR118	GCAGGCGACGCAATTTTTG
tyrR	KD101	ATATCCACATCGCTTTTCA
tyrR	RD121	GTATGGTTGTCCAGGCGAT
tyrR	EA141	CTTGCGGGGCGCTTTCCA
tyrR	RG119	GTTGCGCAGGCCATCCAATTTTTG
tyrR	RC119	GTTGCGCAGGCAATCCAATTTTTG
tyrR	RX119	TGGTTGCGCAGNNNATCCAATTTTTG
tyrR	RD110	GTTGCGCAGGACATCCAATTTTTG
tyrR	RA77	CGCCAGATGCTCGGCTTCGGAAGGC
tyrR	RD119	GTTGCGCAGGACATCCAATTTTTG
tyrR	tyrR HindIII	TGCGAAGCTTGCGGTTCGCT	HindIII
rpoA	Primer (F)	TCTGGATCCGCGCCTGGTTGATATCGAG	BamHI
rpoA	Primer (R)	CAGGCGGATCCCCAGAGACAGTCCACGG	BamHI
rpoA	Reverse long	GAAGGATCCTGCCATATTGCGGAAC	BamHI
rpoA	DR250	AGGATCGGACGGAACTCTG
rpoA	RD310	GACAGTCCATCGGAAGCCA
rpoA	RE310	GACAGTCCCTCGGAAGCCA
rpoA	FA249	ATCGGATCGGCCTCTGGTT
rpoA	FR249	ATCGGATCGCGCTCTGGTT
rpoA	DX250	GCAGGATCGGNNNGAACTCTGGTTTC
rpoA	EG288	GGCGTTTTAAGGAGGCCAACCTCGGTACG
rpoA	DX310	GAGACAGTCCNNNGGAAGCCAGCAC
rpoA	RX119	TGGTTGCGCAGNNNATCCAATTTTTG
rpoA	EX288	GGCGTTTTAAGGAGNNNAACCTCGGTACG
rpoA	Disrupt (F)	TCTGGATCCGCGCCTGGTTGATATCGAG	BamHI
rpoA	Disrupt (R)	CAGGCGGATCCCCAGAGACAGTCCACGG	BamHI

Selection and analysis in the chromosomal rpoA(DN250) strain.

The new strain carrying DN250 on the chromosome offered the opportunity to isolate and analyze Act⁻ rpoA mutants using minimal medium and a lower level of IPTG, in addition to isolation using MacPhe medium. The basal medium for these experiments was 99 Thr Met, since the introduction of multicopy plasmids carrying rpoA⁺ (pLAW2) created a requirement for methionine and some sensitivity to leucine in the absence of the other branched-chain amino acids (99 represents a mixture of histidine, proline, arginine, isoleucine, and valine).

Recombinant DNA techniques.

Standard recombinant DNA procedures were carried out as described by Sambrook et al. (66). DNA screening of mutants was carried out by the chain termination method of Sanger et al. (67), modified in that T7 DNA polymerase was used instead of the Klenow fragment and terminated chains were labeled with ³⁵S-dATPαS. DNA sequencing was carried out automatically, using the BigDye chemistry of Applied Biosystems.

Oligonucleotide-directed mutagenesis.

In vitro mutagenesis with synthetic oligonucleotides was carried out using the method of Vandeyar et al. (68) and oligonucleotides from Geneworks, Adelaide, Australia. For isolation of rpoA mutants, the template for mutagenesis was a derivative of M13tg130 carrying the rpoA gene on a 1.25-kb XbaI-BamHI insert. The mutated alleles were transferred into the original pLAW2 plasmid, replacing the wild-type rpoA gene. For isolation of tyrR mutants altered in the N-terminal domain, mpMU36, a derivative of M13tg130 with a 2.2-kb tyrR insert with the codon corresponding to residue 144 altered to become a HindIII site, was used. Mutants altered in the N-terminal domain of TyrR were then cloned into the vector pSU39 carrying the tyrR⁺ gene, carrying the same engineered HindIII site. The oligonucleotides are listed in Table 2.

Construction of library of tyrR mutants using a mutator strain.

Random mutations in the NTD of TyrR were generated using the mutator strain XLI-Red from Stratagene, as described previously (26, 69).

Assay of β-galactosidase.

Specific activity was assayed according to the method described by Miller and is expressed in the units he defined (69). Assays were carried out in duplicate and on several isolates. Results are averages of at least four values, with uncertainty represented by one standard deviation above and below the mean.