The RNA Polymerase α Subunit Carboxyl-terminal Domain Is Required for Both Basal and Activated Transcription from thealkA Promoter*

Expression of the Escherichia coliadaptive response genes (ada, aidB, andalkA) is regulated by the transcriptional activator, Ada. However, the interactions of RNA polymerase and Ada with these promoters differ. In this report we characterize the interactions of Ada, methylated Ada (meAda), and RNA polymerase at thealkA promoter and contrast these interactions with those characterized previously for the ada and aidBpromoters. At the alkA promoter, we do not detect the RNA polymerase α subunit-mediated binary complex detected at theada and aidB promoters. In the presence of either of these two activators, RNA polymerase protects thealkA core promoter, including the elements at −35 and −10, and is more efficient in transcription initiation in vitro. RNA polymerase holoenzyme containing the α subunit mutation R265A is severely impaired in Ada-independent basalalkA transcription, shows no activation by Ada ormeAda, and fails to bind the alkA promoterin vitro. Binding of the purified wild type α subunit toalkA was not detected, but a complex of promoter DNA, Ada or meAda, and α was observed in gel shift assays. These observations suggest that both forms of Ada protein activatealkA transcription by enhancing RNA polymerase holoenzyme and α subunit binding.

In bacteria, transcriptional activation plays a pivotal role in adaptation processes. Changes in the cell's environment stimulate transcriptional activators to bind their target promoters and to induce their expression. This results in adaptation of gene expression to the environmental requirements. Transcriptional activators have recently been classified on the basis of their mechanism of interaction with RNA polymerase, leading to the definition of Class I and Class II activators (1). According to this definition, Class I transcriptional activators work by contacting the ␣ subunit of RNA polymerase, whereas Class II activators interact with the factor. ␣ and are also the subunits of RNA polymerase involved in specific DNA binding and promoter recognition. is responsible for interaction with the Ϫ35 and Ϫ10 sequences, whereas ␣ is able to bind a third element of the bacterial promoter identified at the rrnB P1 promoter region and termed the UP element (2).
The Escherichia coli adaptive response to DNA methylation damage is induced when cells are exposed to sublethal doses of alkylating agents (3)(4)(5). ada is the regulatory gene of the adaptive response. The product of the ada gene, Ada, is a methyltransferase able to transfer methyl groups from damaged DNA to two of its own cysteine residues (6,7). Methylation of Cys-69 converts Ada into a transcriptional activator. The Cys-69methylated Ada protein ( me Ada) 1 is able to recognize specific sequences upstream of its own promoter and at least two other promoters, alkA and aidB, and activate their transcription (8 -10). me Ada has been proposed as a Class I transcriptional activator, since deletion of the carboxyl-terminal domain of the RNA polymerase ␣ subunit results in the loss of transcriptional activation by me Ada at the ada promoter (11).
We have recently shown that the ␣ subunit of RNA polymerase binds to the promoters of the adaptive response genes ada and aidB in the absence of Ada protein (12). RNA polymerase holoenzyme binds the two promoter regions between Ϫ60 and Ϫ40 via its ␣ subunit; this area overlaps the me Ada binding site and closely resembles the UP element of the rrnB P1 promoter with respect to location, high A/T content, and recognition by ␣ (2). This RNA polymerase⅐promoter binary complex shows only basal levels of transcription and is modified by me Ada into a ternary complex competent in transcription initiation at induced levels (12). The single amino acid substitution in the ␣ DNA binding domain, R265A, which is a change from arginine to alanine at residue 265, prevents ␣ from binding the rrnB P1 promoter (13) as well as the Ϫ60 to Ϫ40 region of the ada and aidB promoters (12).
The alkA gene encodes a glycosylase responsible for recognition and excision of methylated DNA bases (14). A number of observations strongly suggest that the mechanism of Ada action on the alkA promoter differs from that of the other adaptive response genes. First, methylation of Ada is not necessary for alkA induction. Overexpression of the unmethylated Ada in vivo is sufficient to induce alkA (i.e. in the absence of alkylation damage) (14). Both Ada and me Ada proteins are able to activate transcription of alkA in vitro (9), although higher concentrations of Ada are required (15,16). In contrast, the ada and aidB promoters are only activated by me Ada protein (6,8), and high concentrations of Ada protein inhibit activation by me Ada (17). Second, the location of the Ada binding site in the alkA promoter region differs from that in the aidB and ada promoters. At the alkA promoter, me Ada protects residues from Ϫ54 to Ϫ29 (16,18), overlapping the Ϫ35 region. In both ada and aidB, me Ada binds farther upstream, protecting sequences from Ϫ62 to Ϫ38 in DNase I protection experiments (10,19) (see Fig. 1). * This work was supported in part by National Institutes of Health, American Cancer Society-Massachusetts Division, Inc. Grants GM37052 (to M. R. V.) and GM37048 (to R. L. G.), and by research funds from the University of Massachusetts Medical Center. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Third, Ada mutants capable of activating alkA but not ada, and vice versa, have been isolated (15,20). Likewise, the isolated, methylated amino-terminal domain of Ada acts differently on the two types of promoters, activating alkA but inhibiting ada (18).
In this report we investigate the interaction between RNA polymerase and the alkA promoter and the effects of Ada and me Ada on this interaction. We find that 1) the RNA polymerase ␣ subunit plays an important role in both basal and Adaactivated transcription, 2) both Ada and me Ada stimulate specific binding of both RNA polymerase holoenzyme and the purified ␣ subunit of RNA polymerase, and 3) the Ada⅐RNA polymerase⅐alkA promoter ternary complex differs from the ternary complex formed by me Ada in the extent of upstream DNA binding. DNase I Protection and Gel Retardation Experiments-An NruI/PstI (Ϫ60 to ϩ296 of the alkA region) fragment from pYN3072 (9) was subcloned into pSL1180 (Pharmacia) using the corresponding restriction sites to obtain plasmid pMV447. To obtain pMV465, pMV447 was cut with AccI, and the vector fragment was re-ligated to delete 253 bp from the polylinker of pSL1180. For the coding strand, a HindIII/AccI (145 bp) fragment from pMV447 that contained the alkA promoter region from Ϫ60 to ϩ24 plus an additional 61 bp of upstream vector DNA was labeled at the AccI site with [␣-32 P]dATP by an end-filling reaction with Klenow enzyme. For the template strand, a HindIII/ EcoRI (169 bp) fragment from pMV465 that contained the same 145-bp fragment as pMV447 plus an additional 24 bp of downstream vector DNA was labeled at the HindIII site. The region downstream of Ϫ60 contained all the necessary elements for transcription and regulation of alkA (Ref. 16 and Fig. 1). Deletion of the region upstream of Ϫ60 resulted in the loss of a second promoter that reads in the opposite direction (14). This promoter is not regulated by the Ada protein and does not affect levels of alkA transcription either in vivo or in vitro. 2 The Ada protein was purified as in Saget and Walker (17), and methylation of the Ada protein was obtained as in Nakabeppu and Sekiguchi (9). DNase I protection and gel retardation experiments were performed in 20 l final volume of binding buffer (12). Samples were incubated for 20 min at 22°C and processed as described previously (12).

Chemicals-Restriction
In Vitro Transcription-Single-round in vitro transcription experiments were performed using the linear DNA templates as follows. For the alkA promoter, a HindIII/EcoRI fragment from pMV465 (169 bp) was used. This fragment contained the same 169-bp alkA promoter fragment used in DNase I protection experiments of the template strand. The RNA transcript obtained from this fragment was 48 nucleotides in length. As a control, we used the lacUV5 promoter, a 205-bp fragment from pYN3077 (9), to produce an RNA transcript of 65 nucleotides. In the experimental conditions, a second RNA transcript approximately 10 nucleotides shorter was also produced, possibly from a cryptic promoter on the same DNA fragment. The sizes of the RNA transcripts were confirmed by DNA sequencing ladders run on the same gels, assuming an electrophoretic mobility 10% slower for RNA than DNA. 0.3 pmol of each promoter region were preincubated for 5 min at room temperature with 4 pmol of either wild type RNA polymerase holoenzyme or ␣ R265A-RNA polymerase (reconstituted as in Ref. 13) in 40 mM Tris-HCl, pH 8.0, 30 mM KCl, 10 mM MgCl 2 , 1 mM dithiothreitol, and 20 g/ml bovine serum albumin. When required, 10 pmol of unmethylated or methylated Ada were added, and the mixture was incubated for 15 min at 37°C. Methylation of Ada was performed as in Ref. 9, except that the amount of methylated calf thymus DNA was lowered to 0.06 g. Transcription reactions were started by the addition of 0.1 mM each ATP, GTP, and UTP, 0.001 mM CTP, 10 Ci of [␣-32 P]CTP, 250 g/ml heparin (final reaction volume, 50 l). After 5 min of incubation at 37°C, 1 mM unlabeled CTP was added, and the reaction was allowed to continue for 3 min and then blocked with 50 l of stop solution (600 mM NaCl, 50 mM EDTA, pH 8.0, 30 g/ml salmon sperm DNA). Samples were precipitated with 2 volumes of ethanol, resuspended in gel loading buffer, and analyzed on a 6% polyacrylamide, 8 M urea gel.

Effects of the ␣ Subunit Mutation R265A on RNA Polymerase
Function at alkA-Consistent with previous reports (8,9,16), we observed a low level of transcription of the alkA promoter with wild type RNA polymerase in vitro in the absence of any activator protein (Fig. 2, lane 2). Both the unmethylated and the methylated forms of Ada (Ada and me Ada) stimulated in vitro transcription, although the highest levels of transcription occurred with me Ada (Fig. 2, lanes 3 and 4). We and others have previously found that the carboxyl-terminal domain of the RNA polymerase ␣ subunit plays a role at other Ada-activated promoters (11,12). However, the binding site for me Ada in the alkA promoter (Ϫ52 to Ϫ30, partially overlapping the Ϫ35 region) differs from that at the aidB and ada promoters that have me Ada binding sites extending from Ϫ36 to Ϫ60 and Ϫ40 to Ϫ64, respectively. This result combined with the results from the genetic and biochemical studies of others (11,17,18,20) suggests that the mechanisms at these promoters may differ. We therefore asked whether the carboxyl-terminal domain of the RNA polymerase ␣ subunit is essential for activation of transcription by Ada at the alkA promoter.
We used an RNA polymerase containing the ␣ R265A mutation. This mutant ␣ subunit was previously shown to be defective in DNA binding and, when assembled into the RNA polymerase holoenzyme, was defective in UP element function at the rrnB P1 promoter (13). Purified ␣ R265A was also shown to be defective in the binding of the upstream regions of the Ada-dependent aidB and ada promoters (12). The ␣ R265A-RNA polymerase was reconstituted in vitro and used in transcription of alkA in the presence or absence of Ada and me Ada (Fig. 2, lanes 5-8). The activator-independent, basal level of transcription of alkA was reduced substantially by the R265A mutation (lane 6), and no activation of trancription was detected by either Ada or me Ada (lanes 7 and 8). The R265A-RNA polymerase transcribed the lacUV5 promoter as efficiently as The single underline indicates the RNA polymerase binding sites in the absence of the Ada protein, as identified by DNase I protection assays (10). The dashed overline indicates the me Ada binding sites (10,18,19). A plus or minus sign indicates the ability or inability of ␣ or me Ada to bind. See text for further details.
wild type RNA polymerase in these reactions (compare lanes  1-4 with lanes 5-8). Another transcript from the placUV5 fragment was also detected with wild type RNA polymerase (lanes 1-4) although not with R265A-RNA polymerase (lanes [5][6][7][8]. This transcript is approximately 52 nucleotides in length and may be the result of either initiation from a weak promoter-like sequence or premature termination. Relative Binding Affinities of Ada and me Ada at the alkA Promoter-Since unmethylated Ada was found to stimulate transcription by wild type RNA polymerase, but less effectively than me Ada ( Fig. 2 and Ref. 9), we tested the possibility that there are differences in the affinity of the two forms of the activator for the binding site in the alkA promoter. We used a gel shift assay to compare the concentrations of Ada or me Ada required to observe formation of a complex with an alkA promoter fragment (Fig. 3). A me Ada⅐DNA complex was formed that was detectable at a relatively low me Ada concentration (0.015 M) and required a concentration of approximately 0.05 M for complete site occupancy. Complexes of higher mobility were also formed at high concentrations of me Ada, but the nature of the interactions in these complexes is not understood. In contrast, the unmethylated form of Ada formed complexes with alkA only at a relatively high concentration (0.17-0.50 M), and full site occupancy was not detected at the maximum concentration used. Thus, the methylated form of Ada binds the alkA site more efficiently.
Formation of Complexes of alkA, RNA Polymerase, and Either Ada or me Ada-We previously observed that RNA polymerase alone, in the absence of me Ada, protected the upstream region of the aidB and ada promoters in a complex detected by DNase I footprinting. The basal level of alkA transcription by the wild type RNA polymerase (Fig. 2) indicates that RNA polymerase is also able to specifically recognize the alkA promoter in the absence of activator. However, we were unable to detect this RNA polymerase⅐alkA complex in a DNase I footprint (Fig. 4, compare lane 1 with 4), presumably reflecting the instability of the complex under these conditions. Consistent with the reduced ability of the ␣ R265A mutant RNA polymerase to carry out basal transcription of alkA (Fig. 2), we were also unable to detect a complex of this mutant RNA polymerase with the promoter (Fig. 4, lane 7).
A stable ternary complex of alkA promoter, RNA polymerase, and either Ada (Fig. 4, lane 5) or me Ada (Fig. 4, lane 6) was observed. The boundaries of protection conferred by the activator protein alone (Ϫ52 to Ϫ30) are seen in the footprint of me Ada (Fig. 4, lane 3), consistent with previous results (18). Although the unmethylated form of Ada does not bind well enough to confer clear protection in the DNase I footprints (Fig.  4, lane 2; see also Figs. 3 and 5), it does support the formation of a ternary complex with RNA polymerase and the promoter. The upstream boundaries of protection observed with Ada and me Ada containing ternary complexes differ (compare lane 5 with 6). The complex containing me Ada had an upstream boundary of Ϫ52, whereas the complex with Ada had a boundary at Ϫ43. Increasing the concentration of Ada to 2.5 M did not convert the protection pattern exhibited by Ada to that of me Ada (data not shown), suggesting that the difference in protection patterns reflects differences in the structure of the complexes, rather than degree of occupancy of the promoter. The Ada or me Ada concentration dependence of formation of these complexes is examined further in Fig. 5.
Consistent with the inability of the mutant R265A-RNA polymerase to be activated by Ada or me Ada at the alkA promoter (Fig. 2), we did not observe a ternary complex of R265A-RNA polymerase, the alkA promoter, and Ada or me Ada in DNase I footprint assays (Fig. 4, lanes 8 and 9). (The protected region in lane 9 reflects binding of me Ada (compare with lane 3).) Thus, the arginine 265 residue in the carboxyl-terminal domain of the ␣ subunit is required for formation of the Ada-or me Ada-stimulated ternary complexes.
Consistent with the gel shift results of Fig. 3 indicating that only a small fraction of DNA molecules were bound by Ada, we did not detect an Ada footprint at concentrations ranging from 0.02 to 0.5 M (Fig. 5, lanes 2-4). Over a comparable concentration range, me Ada fully occupied its alkA promoter binding site (Fig. 5, lanes 5-7). Although binding of unmethylated Ada protein alone was not detectable, it stimulated the formation of a ternary RNA polymerase⅐alkA⅐Ada complex at each concentration tested (Fig. 5, lanes 9 -11), as did me Ada (lanes 12-14).
Thus, it appears that there is cooperative interaction of RNA polymerase and Ada at this promoter.
Formation of Complexes of ␣ Subunit and Either Ada or  (Fig. 2). The effect on basal transcription indicates that the alkA promoter contains a UP element component. Purified ␣ subunit binds to DNA in the UP element region of several promoters, including the me Ada-activated ada and aidB promoters (12). However, we were unable to detect complexes of purified ␣ subunits and the alkA promoter in a gel shift assay at concentrations higher than required for complex formation at other UP elements (0.5 M (13) ;  Fig. 6, lane 4) or in DNase I protection assays (data not shown). In the presence of purified wild type ␣ subunit and either Ada or me Ada, complexes with mobility slower than that of the Ada or me Ada binary complexes were observed, indicating a cooperative interaction of ␣ with both forms of Ada (Fig. 6, compare  lanes 5 and 6 with lanes 2 and 3). The ␣⅐Ada or ␣⅐ me Ada complexes were not detected, however, when purified R265A mutant ␣ subunit was used (lanes 7-9), indicating that the DNA binding function of the carboxyl-terminal domain is necessary for the formation of these complexes as well as for activation of this promoter.

DISCUSSION
In a previous report (12), we showed that RNA polymerase is able to contact the promoters of the adaptive response genes ada and aidB via binding of the ␣ subunit to the Ϫ40 to Ϫ60 region in the absence of the transcriptional activator me Ada. At these promoters, binding of me Ada converts the RNA polymerase⅐promoter complex into a ternary complex competent for induced levels of transcription initiation. At the ada and aidB promoters, only me Ada functions as an inducer, and its role in activation is to isomerize a binary complex of RNA polymerase bound to an upstream site via ␣ subunit contacts; binding of me Ada alters the nature but not the affinity of RNA polymerase binding, leading to complete promoter protection and induced transcription (12). The alkA promoter differs from ada and aidB promoters in several respects. Both forms of the Ada protein activate alkA transcription (9) and stimulate complete protection of the alkA promoter region by RNA polymerase (Fig. 4, lanes 5 and 6). RNA polymerase alone binds the alkA promoter too weakly to afford clear DNase I protection, but both forms of Ada activator enhance RNA polymerase binding to the alkA promoter. Thus, the role for me Ada at the DNase I-hypersensitive sites are indicated by arrows. DNase I protection of this fragment by RNA polymerase alone is likely to reflect end binding by RNA polymerase rather than a promoter-specific interaction (24).
alkA promoter is fundamentally different from its role at other Ada-dependent promoters.
Although RNA polymerase is unable to bind the alkA promoter in the absence of Ada at levels sufficient to afford clear protection from DNase I (Fig. 4, lane 4), it can specifically interact with the alkA promoter region, resulting in low levels of transcription in vitro (Fig. 2, lane 2) and providing a basal level of expression of alkA in vivo (21).
The region of alkA bound by me Ada extends from Ϫ52 to Ϫ30 (Ref. 18 and Fig. 4, lane 3). The weak binding displayed by unmethylated Ada in the absence of RNA polymerase does not allow a determination of its binding site. The Ϫ52 to Ϫ30 area includes the recognition sequence we have recently proposed for methylated Ada, AATnnnnnnGCAA (where n represents the part of the sequence that varies) (10), which in the alkA promoter spans between Ϫ46 and Ϫ34 (Fig. 1). The binding affinity of unmethylated Ada to the alkA promoter region in the ternary complex is increased at least 25-fold (Fig. 5) by the presence of RNA polymerase. Unlike the unmethylated protein, me Ada binding affinity to the alkA promoter region is not affected by RNA polymerase in this concentration range (Fig. 5).
Differences between Ada⅐ and me Ada⅐RNA polymerase⅐alkA promoter ternary complexes are clearly detectable in DNase I protection experiments in the Ϫ40 to Ϫ60 areas (Fig. 4). These differences appear to be qualitative rather than quantitative, since different DNase I protection patterns are produced even in the presence of a large excess of unmethylated Ada (2.5 M, data not shown). We suggest that different DNase I protection patterns reflect different conformations of RNA polymerase and/or promoter DNA in the two ternary complexes. Despite these differences, both unmethylated and methylated Ada share the ability to stimulate ␣ binding and trigger more efficient transcription initiation.
In contrast to what is seen at the ada and aidB promoters, the ␣ subunit of RNA polymerase does not bind the alkA promoter in the absence of the Ada protein; however, the presence of either Ada or me Ada together with ␣ results in formation of ternary complexes detectable in band shift experiments (Fig. 6, lanes 5 and 6). The requirement for the binding of Ada protein to the alkA promoter for ␣ binding raises the possibility that alkA promoter activation may result from recruitment of ␣ to its binding site, which in turn allows RNA polymerase to bind the appropriate core promoter elements. Ada-and me Adastimulated binding of ␣ could be mediated by protein-protein interaction of ␣ with Ada and me Ada; an alternative possibility is that binding of Ada or me Ada might facilitate binding of ␣ to a specific DNA site in the alkA promoter. The observation that the DNA binding-defective R265A mutant ␣ subunit is not able to bind the alkA promoter region even in the presence of me Ada is consistent with this hypothesis. Promoters have already been described in which the ␣ subunit is able to specifically bind DNA in the Ϫ40 to Ϫ60 area upon binding of transcriptional activators: in the gal P1 promoter, for instance, binding of the catabolite repressor protein induces upstream binding of ␣ (22). DNase I protection assays performed on the me Ada⅐␣⅐alkA promoter ternary complex indicate that ␣ does not extend the area of protection provided by me Ada (data not shown), suggesting that the ␣ binding site overlaps the me Ada binding site and is therefore indistinguishable in DNase I protection experiments. Overlapping sites for ␣ and me Ada have already been observed at both the ada and aidB promoters (12).
The R265A mutation of ␣ impairs RNA polymerase binding to the alkA promoter in the presence of Ada, as determined by DNase I protection experiments (Fig. 4) and almost completely abolishes basal and activated transcription from the alkA promoter (Fig. 2). These observations suggest that the ␣ subunit plays an important role in the formation of the RNA polymerase⅐alkA promoter binary complex and that ␣ binding is necessary for uninduced, basal levels of alkA transcription. This finding is apparently a contradiction with the lack of binding observed with the purified ␣ subunit (Fig. 6). However, additional interactions between the promoter and the other subunits of RNA polymerase may enhance or stabilize ␣ binding when it is part of the holoenzyme. Indeed, differences in concentration needed for DNA binding by the ␣ subunit alone and ␣ assembled into RNA polymerase have already been described (2,12). At the alkA promoter, the binding of RNA polymerase subunits in other areas of the promoter might provide stabilization of ␣ binding. Moreover, the requirement of Ada or me Ada for ␣ subunit binding suggests that the different role for Ada in activation of alkA predicted from genetic studies (9, 14 -17) may be that at alkA, Ada and me Ada function to enhance RNA polymerase holoenzyme binding by stimulating ␣ binding, whereas at the ada and aidB promoters, only me Ada functions as an activator, and ␣ binding is unaffected by me Ada; it has been proposed that at ada and aidB, me Ada is required for binding of RNA polymerase to the core promoter elements at Ϫ10 and Ϫ35 (12).
Although the DNA binding function of the ␣ subunit is required for binding of RNA polymerase to the alkA promoter, this requirement does not exclude the possibility that transcription is activated by Ada through a Class I transcriptional activator mechanism, as proposed (1,11). At the alkA promoter, methylated Ada binding is required for ␣ binding, a role consistent with the general model proposed for Class I transcription activators (23). It is possible that interaction of the Ada protein with the ␣ subunit triggers an extensive conformational change in RNA polymerase, so that the Ϫ35 and Ϫ10 elements are accessible to the factor, and transcription initiation can occur. However, direct interaction of Ada and me Ada with other subunits such as cannot be ruled out; the location of the Ada binding site in the alkA promoter would favor such interaction, which might not be possible in the other Ada-dependent promoters where the Ada binding site is 6 -8 base pairs upstream of the Ϫ35 sequence.