Ada Protein-RNA Polymerase ς Subunit Interaction and α Subunit-Promoter DNA Interaction Are Necessary at Different Steps in Transcription Initiation at the Escherichia coli ada andaidB Promoters*

The methylated form of the Ada protein (meAda) binds the ada andaidB promoters between 60 and 40 base pairs upstream from the transcription start and activates transcription of theEscherichia coli ada and aidB genes. This region is also a binding site for the α subunit of RNA polymerase and resembles the rrnB P1 UP element in A/T content and location relative to the core promoter. In this report, we show that deletion of the C-terminal domain of the α subunit severely decreasesmeAda-independent binding of RNA polymerase toada and aidB, affecting transcription initiation at these promoters. We provide evidence thatmeAda activates transcription by direct interaction with the C-terminal domain of RNA polymerase ς70 subunit (amino acids 574–613). Several negatively charged residues in the ς70 C-terminal domain are important for transcription activation by meAda; in particular, a glutamic acid to valine substitution at position 575 has a dramatic effect onmeAda-dependent transcription. Based on these observations, we propose that the role of the α subunit atada and aidB is to allow initial binding of RNA polymerase to the promoters. However, transcription initiation is dependent on meAda-ς70 interaction.

Transcription activation is one of the principal strategies used by bacteria to respond to external stimuli and to adapt to a changing environment. Most Escherichia coli activators stimulate transcription by establishing protein-protein interaction with RNA polymerase. Different subunits of RNA polymerase can be a target for transcription activators; however, the majority of activators interact with either the ␣ or the 70 subunits (1,2). ␣ and 70 are also the subunits of RNA polymerase responsible for specific binding to promoters; 70 contacts the Ϫ35 and Ϫ10 promoter elements (core promoter elements), whereas ␣ interacts with UP elements. At the strong rrnB P1 promoter, an UP element stimulates transcription initiation 30-fold through direct interaction with the ␣ subunit C-terminal domain (␣CTD), 1 in the absence of any other protein factors (3).
Exposure of E. coli to sublethal concentrations of methylating agents such as methyl methanesulfonate (MMS) activates expression from three promoters: the ada promoter (which also controls expression of the alkB gene), the alkA promoter, and the aidB promoter. This process is called the adaptive response (4 -7). The product of the ada gene, the Ada protein, plays a dual role in the adaptive response; Ada transfers methyl groups from DNA to two of its cysteine residues, thereby functioning as a DNA repair protein. Upon self-methylation, Ada is converted into an activator able to stimulate transcription of the adaptive response genes, including its own (8 -10). Ada is a 39-kDa protein organized in two independently structured domains, each with one methyl-acceptor cysteine (11). Methylation of Cys-69, in the N-terminal domain, triggers specific DNA binding by Ada and is required for transcription activation. In contrast, Ada protein singly methylated at Cys-321 is not capable of specific DNA binding activity. However, the Ada Cterminal domain does play a role in transcription activation; deletions in Ada CTD affect transcription of ada, even though they do not affect specific DNA binding (12)(13)(14). The methylated Ada protein ( me Ada) recognizes the sequence AAT(N) 6 GCAA, which at ada and aidB is located 5 and 7 base pairs upstream of the Ϫ35 sequence, respectively (10,15).
In the model for me Ada activation previously proposed (16), me Ada contacts RNA polymerase through protein-protein interaction with ␣CTD and recruits RNA polymerase to the promoter region (15,16). However, in a previous report (17), it was shown that RNA polymerase binds the ada and aidB promoters via ␣CTD, regardless of the presence of Ada. In the absence of me Ada, RNA polymerase binds to the Ϫ60 to Ϫ40 region, which also includes the me Ada binding site. This region closely resembles the UP element of the rrnB P1 promoter in A/T content, in location, and in its function as a binding site for the ␣ subunit of RNA polymerase. Mutations in ␣CTD that abolish ␣ binding to the rrnB P1 UP element also affect binding to the Ϫ60 to Ϫ40 region of ada and aidB (17) as well as transcription activation by me Ada (16,18). Due to these similarities, the Ϫ60 to Ϫ40 regions of ada and aidB can be considered as "UP-like elements." UP-like elements promote RNA polymerase binding to the ada and aidB promoters in the absence of me Ada, but the resulting RNA polymerase-promoter binary complexes can only initiate transcription with poor efficiency. Binding of me Ada promotes the formation of a ternary complex that is proficient in transcription initiation (17).
In this report, we show that although ␣CTD is responsible for the formation of the initial RNA polymerase-promoter binary complex, transcription activation requires protein-protein interaction between me Ada and the C-terminal domain of the 70 factor of RNA polymerase ( 70 CTD). Our observations suggest that ␣ and 70 CTD are necessary for different steps in transcription initiation at the ada and aidB promoters.

Construction of aidB Hybrid Promoters and Promoter Activity in
Vivo-The minimal sequence requirement for Ada-dependent activation of the aidB promoter is shown in Fig. 1. This fragment was subcloned into the multiple cloning site of the plasmid pSL1180 (Amersham Pharmacia Biotech), producing plasmid pPL115. The substitutions of the Ϫ10 and Ϫ35 promoter elements to consensus sequences, and of the aidB Ϫ60 to Ϫ40 region to the rrnB P1 UP element (see "Results"), were introduced by using double strand oligodeoxynucleotides. Plasmid pPL115 was digested with either BamHI and NcoI or NcoI and MluI and religated in the presence of complementary double strand oligodeoxynucleotides carrying the desired substitutions and the corresponding restriction sites. The ligation mixtures were used to transform E. coli strain RB791 (19), and recombinant plasmids were sequenced using the T7 sequencing kit (Pharmacia). The promoters were then tested for their in vivo activity by ␤-galactosidase assays in the rpoS strain MV2792. Deletion of the rpoS gene completely abolishes ada-independent regulation of aidB (20,21). Strains were grown to 0.2 A 600 nm in LB medium supplemented with 0.2% glucose, 20 g/ml tetracycline, and 80 g/ml ampicillin and then rediluted 1:50. At an A 600 nm of about 0.02, the cultures were divided in two aliquots, and one was supplemented with 0.04% MMS to activate the adaptive response. Samples were taken 2 h after induction, and ␤-galactosidase activity was measured as described in Ref. 22. For ␤-galactosidase experiments with wild type and mutant rpoD alleles, strain MV3766 (alkB:: PSG1 cam R lacZ) was used. ␤-Galactosidase activity was measured as described above, except that 25 g/ml chloramphenicol was added to the medium, and MMS induction was started at 0.1 A 600 nm .
In Vitro Transcription-Reconstitution of RNA polymerase with histidine-tagged full-length ␣ or histidine-tagged truncated ␣-235 was performed as in Ref. 23. No contamination from wild type ␣ was detectable by SDS-polyacrylamide gel electrophoresis in the ␣-235 RNA polymerase preparation. For reconstitution of RNA polymerase with wild type factor or the E575V mutant, histidine-tagged was purified using Ni-NTA columns (Quiagen), using the standard protocol provided by the manufacturer. Purified factors were added at a 4:1 ratio to core RNA polymerase (Epicentre). Ada was purified as in Ref. 14 and methylated prior to use by the method reported in Ref. 9; when necessary, 0.2 M of me Ada was added to the transcription reaction mixture. Singleround in vitro transcription experiments from linear templates were performed as in Ref. 18. 5 pmol (0.1 M) of reconstituted RNA polymerases was used, except for ␣-235 RNA polymerase, when 12.5 pmol (0.25 M) was used. For the experiments in Fig. 3, the DNA templates were EcoRI-ScaI fragments from pPL115 (wild type aidB promoter sequence) or from pPL116 (aidB derivative with a perfect consensus Ϫ10 sequence). Both fragments are 166 base pairs long and produce an RNA transcript of 40 nucleotides. A DNA fragment carrying the lacUV5 promoter was used as internal control. The fragment is 205 base pairs long and produces a transcript of 65 nucleotides (9). For the experiments in Fig. 8, the template was a HindIII-EcoRI fragment from pYN3066 for the ada promoter (9) and an EcoRI-ScaI fragment from pPL115, described above, for aidB. The amount of transcription was quantified after normalization to the lacUV5 transcript using a phosphoimager (Molecular Dynamics).
Gel Retardation Assays-Fragments for gel retardation assays were the same as those used for the in vitro transcription experiments.
Fragments were labeled, and 5,000 cpm/sample was used in 20 l final volume of reaction buffer (50 mM Tris-HCl, pH 7.6, 50 mM NaCl, 2.5 mM dithiothreitol, 6.25% glycerol, 10 g/ml herring sperm DNA). Wild type and mutant RNA polymerases, factors (purified as in Ref. 24), and me Ada were added as described in the figure legends. Samples were incubated 20 min at 37°C and loaded either on to a 4% (for experiments with RNA polymerase) or a 6% (for experiments with factors) native polyacrylamide gel. Gels were run at 10 V/cm in 0.25 ϫ TBE (22.5 mM Tris borate, 0.5 mM EDTA), 1.25% glycerol. Bands were visualized by autoradiography.
Screening of rpoD Mutations-To screen for rpoD mutant alleles affecting ada-dependent transcription, a plasmid library carrying mutations resulting in single alanine substitutions at 17 amino acids of 70 C-terminal domain (obtained from C. Gross, University of California, San Francisco; plasmids are derivatives of pGEX-2T 70 (25)) was used to transform MV3766 (ada-alkB::lacZ). Strain MV3766 carries a lacZ transcriptional fusion in the alkB gene, which lies downstream of the ada gene and whose transcription is driven by the ada promoter. Additional mutagenesis of the terminal segment of the rpoD gene was performed by two independent polymerase chain reactions from an XhoI site (corresponding to amino acid 528 of 70 ) to a HindIII site immediately downstream of the stop codon. Strain MV3766 was transformed with the plasmid library. Transformants were plated on LB medium and replica-plated onto MacConkey medium both in the presence and in the absence of 0.02% MMS. In the presence of MMS, colonies with normal levels of ada-dependent transcription are pink, whereas colonies with decreased levels are white. The replica-plating step is necessary to avoid exposure of the colonies to the mutagenic effects of MMS. Colonies corresponding to the mutant candidates were picked from the MacConkey plates with no MMS and tested for ␤-galactosidase activity as described above. Fig.  1 shows the sequence of the ada and aidB promoters. The Ϫ60 to Ϫ40 regions of the two promoters show a high degree of similarity to the UP element of the rrnB P1 promoter. This region is also the binding site for me Ada, and its deletion abolishes ada-dependent activation (Ref. 26; Fig. 2). Although both promoters have recognizable Ϫ35 and Ϫ10 sequences, they differ from the consensus hexamers TTGACA and TATAAT. At positively controlled promoters, one of the functions of activator proteins is to improve recognition of one or more weak promoter elements by RNA polymerase through either proteinprotein interaction or modification of local DNA structure. To understand which of the weak promoter elements is the target of me Ada activation, we constructed a set of derivatives of the aidB promoter in which either the Ϫ35 or the Ϫ10 elements were substituted by perfect consensus hexamers, and the UPlike element was substituted by the UP element of rrnB P1. We investigated the effects of these substitutions on both basal and ada-activated levels of transcription in vivo.

In Vivo Transcription from aidB Promoter Derivatives-
Results shown in Fig. 2 suggest that at the rrnB UP element/ aidB hybrid promoter (aidB "rrnB UP"), ada-independent transcription levels are only slightly higher than for the wild type aidB promoter (1.6-fold, Fig. 2). Thus, me Ada does not activate transcription by converting the ␣ binding site of aidB into a better UP element. Because this substitution also replaces the Ada binding site, ada-dependent activation is almost completely abolished. Changes to consensus Ϫ35 or Ϫ10 sequences did have more pronounced effects on transcription; the consensus Ϫ10 increased ada-independent transcription levels by almost 5-fold, and introduction of a consensus Ϫ35 resulted in a more than 12-fold increase. However, neither substitution completely relieved the dependence on me Ada for optimal promoter expression; me Ada activates the aidB promoter derivative with a consensus Ϫ10 element (aidB "Ϫ10 con") by 5-fold, and the derivative with a consensus Ϫ35 element (aidB "Ϫ35 con") by 2.3-fold (Fig. 2). These observations suggest that me Ada activates transcription by affecting RNA polymerase interaction with both the Ϫ35 and Ϫ10 elements of aidB.
In Vitro Transcription-Sakumi et al. (16) had proposed that me Ada activates transcription by direct contact with the CTD of the ␣ subunit. We further investigated the role of ␣CTD in activation by me Ada by performing in vitro transcription experiments using two forms of reconstituted RNA polymerase different with respect to their ␣ subunits: one form carried wild type ␣, the other a mutant ␣ deleted of the C-terminal 94 amino acids (␣-235). Although ␣-235 RNA polymerase is impaired in UP element utilization and in transcription from some activatordependent promoters, it is proficient in transcription from factor-independent and promoters dependent on activators that do not interact with ␣ (1).
We tested the two forms of reconstituted RNA polymerase for transcription from the wild type aidB promoter (aidB wt) as well as from an otherwise identical promoter in which the Ϫ10 sequence was changed to consensus (aidB "Ϫ10 con"). The latter promoter shows 5-fold higher basal transcription level in vivo compared with the wild type aidB promoter but is still dependent on me Ada for maximal promoter expression (Fig. 2). Deletion of the ␣CTD dramatically affects both me Ada-dependent and independent transcription from the wild type aidB promoter, even though activation by me Ada is not totally abolished (Fig. 3, lanes 5 and 6). Transcription from aidB "Ϫ10 con" by ␣-235 RNA polymerase is also affected; however, at this promoter me Ada activates transcription by both forms of RNA polymerase to a similar extent (5.2-fold for wild type ␣and 4-fold for ␣-235 RNA polymerase; Fig. 3, lanes 7-10). These results strongly suggest that, although ␣CTD is necessary for efficient transcription at the wild type aidB promoter, it is not required for activation by me Ada.
The results of in vitro transcription experiments raise the possibility that ␣-235 RNA polymerase is not proficient in carrying out transcription from the wild type aidB promoter because it is affected in RNA polymerase-promoter interaction, rather than interaction with me Ada. To investigate this possibility, we tested both wild type and ␣-235 RNA polymerase for their ability to bind the wild type aidB promoter in the absence of me Ada. As shown in Fig. 4, wild type RNA polymerase binds aidB wt at 0.04 -0.08 M, whereas ␣-235 RNA polymerase fails to bind the promoter at concentrations up to 0.32 M. A similar result was obtained for the wild type ada promoter (data not shown). Both forms of RNA polymerase were equally as efficient in binding both the lacUV5 and the galP1 promoters under the same experimental conditions (data not shown). The above results show that ␣CTD promotes recruitment of RNA polymerase to the ada and aidB promoters independently of me Ada.
Gel Retardation Studies with me Ada and 70 -The location of me Ada binding sites at the ada and aidB promoters is consistent with the possibility of interaction with the 70 factor of RNA polymerase. To investigate this possibility, gel retardation experiments were performed with me Ada and purified 70 . 70 is capable of specific DNA binding only when assembled into RNA polymerase holoenzyme (27). Indeed, no binding of 70 alone to either the ada or aidB promoters could be detected; however, addition of 70 (0.5 M) to a me Ada⅐promoter complex resulted in a supershift at both the ada (Fig. 5) and the aidB (data not shown) promoters. This suggests that 70 is able to form a ternary complex with me Ada at either the ada or the aidB promoter DNA. The presence of multiple shifts is possibly due to partial dissociation of the me Ada⅐DNA complex in the experimental conditions used and is a typical pattern for me Ada (10). In control gel retardation experiments, no significant binding by me Ada and 70 was detected with unrelated DNA fragments; 70 was unable to promote binding of the unmethylated Ada protein to the ada promoter. Finally, 70 did not produce any supershift when added to a CRP⅐gal P1 promoter complex (data not shown).
Gel retardation experiments were performed with two 70 deletion mutants, 574 and 529 ; these mutants, deleted of 39 and 84 amino acids in the C-terminal domain, respectively, are  [3][4][5][6] or the aidB "Ϫ10 con" promoter (paidB Ϫ10 con, lanes 7-10). Fold-activation by me Ada is shown below the transcripts. impaired in interaction with some activators when assembled into RNA polymerase holoenzyme (24). Concentrations up to 5 M of the two mutant factors promote little formation of ternary complexes at either the ada (Fig. 6) or the aidB (data not shown) promoter; in the same experiment, complete supershift of the two complexes occurred at 1 M with wild type 70 .
Substitutions in 70 Affect me Ada-dependent Transcription-The above gel retardation experiments suggest that the Cterminal domain of 70 could be the target for activation by me Ada. Therefore, we expected mutations in the terminal segment of the rpoD gene to specifically affect ada-dependent transcription. We screened for this class of rpoD mutants using a plasmid library carrying rpoD alleles in which the segment encoding for amino acids 528 -613 of 70 had been mutagenized by polymerase chain reaction. The plasmid library was used to transform strain MV3766. This strain carries a lacZ transcriptional fusion in the chromosomal alkB gene, whose transcrip-tion is driven by the ada promoter. Over 1700 colonies were screened and four mutant colonies were isolated. The mutant rpoD alleles were sequenced; three carried nonsense mutations, which resulted in truncations of 70 at amino acid 560 (1 clone) or at amino acid 584 (2 clones). The fourth candidate carried a missense mutation, resulting in a glutamic acid to valine change at position 575 of 70 (E575V).
In addition to polymerase chain reaction-directed mutagenesis, we also tested a set of plasmids (kindly given by C. Gross, UCSF) carrying rpoD alleles with single alanine substitutions at 17 amino acids of 70 CTD (Fig. 7). ␤-Galactosidase assays were performed to quantify the effect of the alanine substitutions, as well as the E575V mutation, on transcription from the ada promoter. In strain MV3766, wild type 70 from the chromosomal rpoD gene is normally expressed, and mutated factors are present in only a slight excess to wild type 70 (data not shown). Nevertheless, expression of several mutant rpoD alleles resulted in altered levels of in vivo transcription at the ada promoter: E574A, E575V, I590A, E591A, E605A, and D612A substitutions significantly decreased ada-dependent transcription, with E575V having the most severe effect (Fig.  7). Some mutations, such as R596A, K597A, R608A, and D613A, resulted instead in an increased level of ada-dependent transcription (120 -150%, Fig. 7).
To verify that inhibition of ada-dependent transcription in vivo is indeed due to disruption of me Ada-70 interaction, we purified both wild type and E575V factors and reconstituted RNA polymerase in vitro. As shown in Fig. 8, E575V 70 -RNA polymerase was able to carry out transcription from the lacUV5 promoter with the same efficiency as wild type 70 -RNA polymerase, which suggests that the E575V mutation does not affect either core enzyme-70 interaction or factorindependent transcription; however, me Ada-dependent transcription by the mutant RNA polymerase at both ada (Fig. 8) and aidB (data not shown) was drastically impaired. DISCUSSION In a previous report, we showed that RNA polymerase binds to the ada and aidB promoters via the ␣ subunit, independently of the Ada protein. Thus, me Ada does not recruit RNA polymerase to the promoters but rather converts the RNA polymerase-promoter complex into a ternary complex proficient in transcription initiation (17). The location of the me Ada binding site suggests the possibility of interactions with either ␣CTD or ; alternatively, the mechanism for activation by me Ada could involve turning the ␣-binding sites of ada and aidB into more efficient UP elements. Our experiments with hybrid promoters show that when the UP-like element of aidB is substituted by the rrnB P1 UP element, no significant stim- ulation of transcription occurs (Fig. 2). In contrast, changing either the Ϫ35 or Ϫ10 sequences of aidB to consensus results in a significant stimulation of ada-independent transcription in vivo, suggesting that me Ada improves RNA polymerase interaction with the core promoter region (Fig. 2). Thus, me Ada activates transcription either by improving initial binding of RNA polymerase to the Ϫ35 or by facilitating a later step in transcription initiation, such as isomerization to open complex. Deletion of ␣CTD severely affects me Ada-activated transcription (Ref. 16; Fig. 3) and prevents RNA polymerase binding to the ada and aidB promoters (Fig. 4). However, at an aidB derivative in which the Ϫ10 sequence was changed to consensus (aidB "Ϫ10 con"), me Ada is able to activate transcription by ␣-235 RNA polymerase (Fig. 3) to roughly the same extent as wild type RNA polymerase. Thus, dependence on ␣CTD for transcription at Ada-dependent promoters can be by-passed by strengthening the core promoter, presumably by providing an alternative binding site for ␣-235 RNA polymerase that compensates for its loss of interaction with the UP-like element. Although it is possible that altering the Ϫ10 sequence of the aidB promoter also modifies the interaction between me Ada and RNA polymerase, the results at the aidB Ϫ10 con promoter clearly demonstrate that me Ada can activate transcription by RNA polymerase containing an ␣ subunit deleted of its CTD. Although we cannot rule out the possibility of direct ␣-me Ada interaction, we propose that inefficient transcription by RNA polymerase deleted of its ␣CTD results from the loss of the ␣-UP-like element interaction and consequent inability of RNA polymerase to bind the promoter, rather than from lack of interaction with me Ada.
Several lines of evidence show that me Ada interacts with 70 . Gel retardation experiments (Figs. 5 and 6) indicate that the terminal 39 amino acids of 70 are necessary for this interaction. This region contains determinants for the recognition of the Ϫ35 element (region 4.2), followed by the so-called "basic cluster" (28). Several amino acids in both region 4.2 and in the basic cluster have been found to be important for interaction with transcription activators, such as PhoB, bacteriophage cI protein, and FNR (29,30). 2  Fold-activation by me Ada is shown below the transcripts. 70 CTD suggests that me Ada interacts with a set of negatively charged residues both in and downstream of region 4.2 (Fig. 7). With the exception of I590A, all the substitutions that significantly affect ada-dependent transcription involve negatively charged amino acids (Glu-574, Glu-575, Glu-591, Glu-605 and Asp-612); substitution to alanine of several positively charged residues (Arg-596, Lys-597, Arg-608) results in increased levels of ada-dependent transcription (Fig. 7), supporting the hypothesis that me Ada interacts with a negatively charged patch in 70 CTD. One of the residues important for me Ada-70 interaction, glutamic acid at position 575 (Glu-575), also plays a role in transcription activation by PhoB (29), suggesting that the location of this residue allows interaction with different transcription activators. The Glu-575 residue does not seem to be necessary for activator-independent transcription; neither the E575V mutation described in this report nor the E575K mutation described in Ref. 29 has any effect on transcription from the lacUV5 promoter in vitro (Fig. 8), nor do they affect growth rates in any of the reporter strains tested (data not shown; Ref. 29).
The specificity of me Ada-70 interaction, the location of its determinants in the terminal 39 amino acids of 70 , and the strong effect of the E575V mutation in vitro point to 70 CTD as the principal target of activation by me Ada at the ada and aidB promoters. Interestingly, me Ada is also able to activate transcription by s -RNA polymerase at both promoters (21,32). s is an alternative factor, highly expressed in the stationary phase of bacterial growth (33). s is homologous to 70 in region 4.2 (28), and the amino acids important for me Ada-dependent transcription are either conserved residues (Glu-574/Glu-289 and Ile-590/Ile-305) or conservative substitutions (Glu-575/ Asp-290, Glu-591/Gln-306, and Glu-605/Gln-320) in s . Thus, it is possible that me Ada could contact amino acids conserved in both factors; future experiments of site-directed mutagenesis in s will allow better understanding of activation of s -dependent transcription by me Ada.
A model for me Ada-activation of ada and aidB is presented in Fig. 9. In the absence of me Ada, RNA polymerase can bind to the ada and aidB promoters via its ␣CTD but fails to establish any strong interaction with the core promoter (17). me Ada does not recruit RNA polymerase to the ada and aidB promoters, but upon binding of ␣ to the UP element it interacts with 70 CTD, triggering transcription initiation; therefore, ␣ subunitpromoter and me Ada-70 interactions act at separate but interdependent steps of transcription initiation. It is possible that Ada CTD is responsible for this interaction; deletions in Ada CTD abolish transcription activation at the ada promoter without affecting DNA binding by me Ada (14).