Effects of Nonbulky DNA Base Damages on Escherichia coli RNA Polymerase-mediated Elongation and Promoter Clearance*

DNA base damage products either formed spontaneously or as a result of exposure to various genotoxic agents were examined for their effects on Escherichia coli RNA polymerase-mediated transcription in vitro. Uracil, O6-methylguanine (O6-meG), and 8-oxoguanine (8-oxoG) were placed at specific sites downstream from the transcriptional start site on the transcribed strand of a duplex template under the control of the strong tac promoter.In vitro, single-round transcription experiments carried out with purified E. coli RNA polymerase revealed efficient bypass at the three lesions examined and subsequent generation of full-length runoff transcripts. Transcript sequence analysis revealed that E. coli RNA polymerase inserted primarily adenine into the transcript opposite to uracil, uracil opposite to O6-meG, and either adenine or cytosine opposite to 8-oxoG. Thus, a uracil in the DNA template resulted in a G-to-A transition mutation in the lesion bypass product whereas O6-meG produced a C-to-U transition mutation and 8-oxoG generated either the correct transcriptional product or a C-to-A transversion mutation. When 8-oxoG was placed within close proximity to the transcriptional start site (within the region required for effective promoter clearance), a reduced of full-length, runoff transcript was observed, indicative of lower promoter clearance. Taken together, these results demonstrate that the DNA base damages studied here may exert significant in vivo effects on gene expression and DNA repair with respect to the production of mutant proteins (transcriptional mutagenesis), or decreased levels of expressed proteins.

There are three major steps for Escherichia coli RNA polymerase (RNAP) 1 to properly effect DNA template-dependent transcription: promoter binding and initiation, RNA chain elongation, and chain termination (1). Any of these steps could potentially be perturbed by DNA base damage occurring on the template strands of transcribed gene sequences, leading to various deleterious consequences for a cell. For example, certain bulky lesions such as UV light-induced cyclobutane pyrimidine dimers and psoralen adducts if located on the transcribed strand of DNA have the ability to permanently arrest RNA polymerase at the site of damage during elongation (2)(3)(4). If left unrepaired, these lesions block gene expression. In E. coli, it is has been demonstrated that RNAP arrested at a DNA lesion is recognized by factors that may displace the arrested polymerase and recruit the appropriate repair machinery components to the damaged site in order to facilitate nucleotide excision repair (2,5,6). It has also been shown that damages such as UV light-induced cyclobutane pyrimidine dimers, when located on the template strands of actively transcribed genes, are repaired preferentially compared with the nontemplate strand as well as nonexpressed regions of the genome (7)(8)(9). Thus, transcription-coupled repair is dependent on the ability of a specific DNA lesion to arrest an elongating RNAP. However, there exist several examples of DNA lesions such as abasic sites, uracil, dihydrouracil, and certain types of strand breaks that are efficiently bypassed by the E. coli transcription elongation complex, and thus presumably not subject to transcription-coupled repair (10 -13). Furthermore, these damages have been shown to be miscoding lesions for RNAP in vitro, causing base substitutions in the resulting transcripts.
In addition to effects on RNAP elongation, it can be envisioned that certain base damages could also affect transcriptional initiation by altering the efficiency of the transition of the RNAP from the initiation state to an elongation complex (promoter clearance). Several reports have shown that ability of RNAP to clear the promoter may be dependent on the nucleotide sequence of the initially transcribed region of DNA (14 -16). Damaged bases within this initial region could affect promoter clearance and subsequently production of a fulllength transcript. The effects of DNA lesions on this process are unknown.
The goal of this work was to extend our previous studies concerning the effects of nonbulky DNA base lesions on elongation of various phage RNA polymerases by both expanding the range of lesions studied, and also by examining these effects on the more complex, multi-subunit E. coli RNAP. Uracil, 8-oxoG, and O 6 -meG are important DNA base damage products formed cytosine deamination, oxygen radical attack of guanine, and alkylating agent-induced methylation of guanine, respectively (17). Each of the lesions studied are repaired by distinct base excision repair enzymes (uracil and 8-oxoG) or direct reversal proteins (O 6 -meG) with specificity for that base damage (17). We examined whether these lesions allowed bypass of RNAP and, if so, whether these lesions were mutagenic at the level of transcription. In addition to further generalizing our previous elongation studies, we wished to study the effects of these damages on promoter clearance. From these experiments, we find that (i) uracil, 8-oxoG, and O 6 -meG are lesions that do not block E. coli RNAP during elongation, (ii) these lesions cause mutagenic base insertion RNAP during transcription, and (iii) when placed near the transcriptional start site, 8-oxoG decreases promoter clearance. These findings have broad implications for the effects of DNA damage on gene * This work was supported by National Institutes of Health Grant CA73041. 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.

EXPERIMENTAL PROCEDURES
Materials-The pGEM-2 in vitro transcription vector was purchased from Promega. Undamaged oligonucleotides for use as PCR primers for transcription duplex DNA template synthesis were synthesized by the Emory Microchemical Facility. Oligonucleotides containing uracil, 8-oxoG, and O 6 -meG were obtained from Life Technologies, Inc. and National Biosciences, Inc. The 5-propynyl cytosine-containing oligonucleotide was obtained from Genosys, Inc. Heparin and RNase inhibitor were purchased from Sigma and Promega, respectively. High performance liquid chromatography-purified nucleotide triphosphates, [␣-32 P]UTP (3000 Ci/mmol), and [␥-32 P]ATP (3000 Ci/mmol) were from Amersham Pharmacia Biotech. QIAEX DNA extraction kit was from Qiagen.
Enzymes-Purified E. coli RNAP (21) was a generous gift from Charles Turnbough (Birmingham, AL). Moloney murine leukemia virus reverse transcriptase was purchased from Stratagene. T4 polynucleotide kinase was purchased from New England Biolabs. E. coli endonuclease III was a gift from Richard P. Cunningham (Albany, NY). Uracil DNA N-glycosylase was purchased from Epicenter Technologies. E. coli endonuclease IV and Fpg protein were gifts from Yoke W. Kow (Atlanta, GA). Taq DNA polymerase for PCR amplifications was purchased from Promega.
Construction of E. coli RNA Polymerase Transcription Templates-A series of oligonucleotides were used as PCR primers to generate linear duplex transcription templates for elongation and promoter clearance experiments. For elongation studies, four different 139-bp duplex transcription templates were generated via PCR amplification using a modification of a method originally described by Zhou and Doetsch (10) employing the plasmid pGEM-2, an upstream primer oligonucleotide (68-mer, containing the tac promoter) and a downstream primer containing either no damage (ND-38-mer), 5-propynyl cytosine (PROC-38mer), uracil (U-38-mer), 8-oxoG (8OG-31-mer), or O 6 -meG (O6MEG-31mer). The resulting products (Fig. 1A) were gel-purified and contained no damage (template ND-elong), 5-propynyl cytosine (template PROCelong), or uracil (template URA-elong) 56 nt downstream from the transcription start site or 8-oxoG (template 8OG-elong) or O 6 -meG (template O6MEG-elong) 63 nt downstream from the transcription start site. For promoter clearance studies, two different 94-bp duplex transcription templates were generated via a similar strategy except that a segment of duplex template ND-elong was PCR-amplified with upstream primer oligonucleotide 21-mer and downstream undamaged oligonucleotide C-45-mer or 8-oxoG-containing oligonucleotide 8OG-45mer. The products (Fig. 1B) contained either no damage (template ND-clear) or 8-oxoG (template 8OG-clear) 4 nt downstream from the transcription start site.
Transcription Experiments-Single-round transcription reactions with E. coli RNAP were carried out as described previously (18) with several modifications. Briefly, the transcription buffer contained 40 mM Tris-HCl, pH 7.95, 50 mM KCl, 10 mM MgCl 2 , 2 mM ATP, 5 mM dithiothreitol, and RNasin at 0.8 unit/l. Reaction mixtures containing 0.5 pmol of DNA template, RNAP, and 1.3 M [␣-32 P]UTP were incubated at 37°C for 10 min to form initiation complexes. After preincubation, heparin (250 g/ml) and GTP, CTP, and cold UTP (0.4 mM each) were added (20 l final reaction volume), and 4-l aliquots were removed at various time intervals. Transcripts were analyzed on denaturing 10% polyacrylamide gels and detected by autoradiography and quantified by PhosphorImager analysis (Molecular Dynamics model 445 SI).
RNA Sequencing-Full-length RNA transcripts were gel-purified and cDNAs were generated with MMLV reverse transcriptase and a primer covering the 5Ј end of the RNA transcript under conditions recommended by the supplier (Stratagene). The resulting cDNAs were PCR-amplified with both 5Ј and 3Ј primers covering the 5Ј and 3Ј ends of the transcript using Taq DNA polymerase. These amplified products were subsequently sequenced by dideoxy sequencing using the 32 Plabeled 5Ј primer under conditions recommended by the supplier (U. S. Biochemical Corp.) and as described previously (10).

Construction of Transcription Templates Containing Various
Base Damages at Specific Sites-Duplex DNA templates were constructed by PCR amplification of a segment of the pGEM-2 vector (see "Experimental Procedures"). Transcription templates for elongation studies (139 bp) contained the E. coli tac promoter (49 bp) and a transcribable segment (90 bp) with various base modifications on the transcribed strand (Fig. 1A). Transcription templates for promoter clearance experiments were similar to the elongation templates except that the transcribable segment was shorter (45 bp) and contained 8-oxoG 4 nt downstream from the transcription start site (Fig. 1B). To confirm the identity of each of the templates, PCR reactions were conducted with 5Ј end labeled downstream primers in order to generate labeled DNA templates containing no damage, uracil, 8-oxoG, or O 6 -meG. Template URA-elong was treated with uracil glycosylase and subsequently with E. coli endonuclease IV, which generated the appropriate DNA cleavage product when analyzed on a DNA sequencing gel (data not shown). Likewise, templates 8OG-elong and 8OG-clear were treated with E. coli Fpg, which resulted in DNA cleavage at the sites of 8-oxoG and generation of the predicted, appropriate length DNA strand scission products. Template O6MEG-elong was confirmed by primer extension sequencing by DNA polymerase which yielded the appropriate misinserted base (T) opposite to O 6 -meG into the extended fragment (19). Furthermore, both the damaged primer and template exhibited altered gel mobilities due to the presence of O 6 me-G lesion and the melting temperature of the primer was significantly reduced as would be expected in the presence of this base damage product (20).
E. coli RNAP Bypass of Nonbulky Base Damage during Elongation-Damaged DNA templates were used in in vitro transcription experiments with purified E. coli RNAP (21). In order to determine the interaction of E. coli RNAP with each of the damaged bases during elongation, comparative single-round transcription experiments were carried out with templates CON-, PROC-, URA-, 8OG-, and O6MEG-elong (Fig. 1A). In these single-round transcription experiments, each template molecule is utilized only once by a single molecule of RNAP and the transcription products represent a single, promoterdependent elongation event (10 -12). Measurement of transcript formation at various times during a single round of elongation will directly reveal template sites that cause temporary pausing or permanent arrest of RNAP at a lesion site (10). The amount of transcript generated was monitored from 0 to 2 min following the start of the transcription elongation phase. The undamaged control template CON-elong, which contains cytosine 56 nt downstream and guanine 63 nt downstream from the transcriptional start site, produced a 90-nt full-length runoff transcript as expected ( Fig. 2A, lanes 1-5). Base damage-containing templates URA-, 8OG-, and O6MEGelong were also efficiently transcribed and generated fulllength runoff transcription products (Fig. 2, lanes 6 -20), indicating efficient bypass of uracil, 8-oxoG, and O 6 -meG by E. coli RNAP. No temporary pausing or permanent arrest of RNAP at the lesion sites was observed as determined by the absence of shortened transcripts in the 56 -63-nt size range. However, a template (PROC-elong) containing 5-propynyl cytosine at nt position 56 efficiently blocks transcription at this site indicating that this type of base modification is an effective block to transcriptional elongation under single-round transcription conditions (Fig. 2B). With the exception of PROC-elong, each template generated the 90-nt full length, runoff transcript with similar kinetics (Fig. 3). Thus, unlike our previous studies, which examined abasic sites and dihydrouracil effects on phage and E. coli RNAP (10), there was no evidence of temporary pausing of the polymerase at sites of uracil, 8-oxoG, and O 6 -meG. We conclude that, when encountered by an elongating E. coli RNAP, uracil, 8-oxoG, and O 6 -meG are efficiently bypassed with no detectable pausing or arrest.
Sequence Analysis of Transcripts from Elongation Stud-ies-In living cells, uracil is formed through the spontaneous deamination of cytosine (17). Therefore, if any base other than guanine is inserted opposite to uracil by RNAP, the resulting transcript will contain a base substitution mutation (10). Sequence analysis of transcripts generated from template URAelong indicated the expected adenine insertion opposite to uracil (Fig. 4B). In the case of 8-oxoG, transcript sequence analysis revealed that E. coli RNAP inserted either adenine or cytosine opposite to this lesion with approximately equal frequencies (Fig. 4C). This result is similar to that previously observed for T7 RNA polymerase and 8-oxoG (22), indicating that this base damage may affect both types of RNA polymerases in a similar fashion. Sequence analysis of transcripts generated from template O6MEG-elong indicated preferential insertion of uracil FIG. 1. A, E. coli RNAP transcription elongation templates. The control, undamaged duplex template (CON-elong) was generated via PCR utilizing the indicated upstream (68-mer) and downstream (C-38-mer) primers as described under "Experimental Procedures." Duplex transcription templates PROC-elong, URA-elong, 8OG-elong, and O6MEG-elong contain 5-propynylcytosine, uracil, 8-oxoG, and O6-meG on the transcribed (bottom) strand 56, 56, 63, and 63 nt downstream from the start of transcription, respectively (vertical arrows). The boxed areas indicate the E. coli RNAP tac-1 promoter. The bold horizontal arrow indicates the transcription start site and the direction of transcription. Upstream (68-mer) and downstream (31-or 38-mer) PCR primers used to generate the templates are indicated by the solid lines above and below nontemplate and template strands. The resulting 139-bp transcription templates are composed of the promoter (48 bp) and transcribable (90 bp) regions as indicated. B, E. coli RNAP promoter clearance templates. The control, undamaged template (CON-clear) was generated via PCR utilizing the indicated upstream (21-mer) and downstream (C-45-mer) primers as described under "Experimental Procedures." 8OG-clear contains 8-oxoG (vertical arrow) on the transcribed strand 4 nt from the transcriptional start site (horizontal arrow) and was generated by using oligonucleotide 8OG-45-mer in place of C-45-mer. opposite to O 6 -meG (Fig. 4D). This insertion preference is similar to the effect of O 6-meG on DNA polymerases (19). Thus, these three DNA base damages are highly mutagenic at the level of transcription in vitro, with uracil causing G 3 A transition mutations, 8-oxoG causing C 3 A transversion muta-tions or normal base insertions, and O 6 -meG causing C 3 U transition mutations.
Promoter Clearance Studies-An important step in transcriptional initiation is promoter clearance, the transition step where RNAP switches from an initial transcribing stage to an elongation stage. During this step, the holoenzyme is involved in generating short RNA oligomers from the promoter-proximal sequence (initial transcribed sequence (ITS)) in abortive initiation. In order to determine whether nonbulky DNA base damages in the ITS could affect promoter escape and consequently the extent of gene expression, a transcription template containing 8-oxoG 4 nt downstream from the start of transcription was constructed (Fig. 1B). This damage was placed in close proximity to the promoter because it had been shown previously that alterations in the ITS contribute to variations in promoter activity (1,16). In vitro transcription experiments were carried out to assess the effect of 8-oxoG on the extent of transcription associated with efficient promoter clearance. In order to accurately compare transcription in damaged versus undamaged templates, the amount of DNA used in each reaction was carefully controlled (see "Experimental Procedures"). The extent of promoter clearance was assessed as a function of the amount of full-length runoff transcript that was generated from either the undamaged (CON-clear) or damaged (8OGclear) template. Transcription products were quantified by polyacrylamide gel electrophoresis followed by PhosphorImager analysis (see "Experimental Procedures"). It was found that runoff transcript, and hence promoter clearance, was reduced by approximately 50% when 8-oxoG was placed near the promoter (Fig. 5). From these data, we conclude that oxidative base damage, if located near the promoter, can significantly reduce gene expression at the level of transcription.

DISCUSSION
The results of this study demonstrate that various types of DNA damage can have different effects on transcription by E. coli RNAP when such damages are located on the template strand of the transcribed DNA segment. First, we have shown that several nonbulky lesions affect the quality of the transcription products (i.e. mutation-containing) by allowing for efficient bypass and mutagenic base insertion during elongation. We have also demonstrated that when certain base damages such as 8-oxoG are located close to the transcriptional start site, there is a significant negative effect on promoter clearance by RNAP and a corresponding decrease in the level of transcript produced. It is conceivable that these lesions, which are efficiently bypassed by RNAP, may negatively influence their own repair if situated on the template strand of a transcribed segment of DNA. Therefore, if lesions that do not block RNAP accumulate and/or remain unrepaired, they could alter gene expression in several ways. As each of the lesions studied in this work are recognized by different repair enzymes, it is possible that a defect in any one of these enzymes could lead to the generation of significant levels of mutant protein or reduced gene expression.
Uracil, 8-oxoG, and O 6 -meG are all efficiently bypassed by E. coli RNAP during elongation in vitro. Transcriptional bypass is accompanied by mutagenic base insertion events opposite to these lesions. Specifically, uracil inserts an adenine (assuming cytosine to be in the undamaged template), whereas 8-oxoguanine also frequently miscodes for adenine, and O 6 -methylguanine allows for the insertion of uracil. All of these events are equivalent to a base substitution mutation at the level of RNA synthesis (transcriptional mutagenesis). These findings have several important implications for DNA repair and mutagenesis in general. More bulky lesions such as cyclobutane pyrimidine dimers cause the permanent arrest of RNAP at the dam- . Levels of transcription products were determined at various times (0 -2 min) following the start of elongation. B, kinetics of full-length transcript formation between templates containing control (Ⅺ) and 8-oxoguanine (q). Three separate single-round transcription experiments were conducted for each template, and the error bars represent standard deviations. age site, thus signaling the cellular proteins which direct various components of the DNA repair machinery to the damage site in order to carry out repair (2). It is the arrest of an elongating RNAP that is thought to be an important signal for initiating transcription-coupled repair of DNA damages located on the template strands of actively transcribed genes (7)(8)(9). However, the nonbulky base damages studied in this work as well as other lesions previously studied in our laboratory (10,11) do not arrest an elongating RNAP, but rather allow for efficient bypass. This leads to the prediction that these lesions may not be subject to the transcription-coupled repair system involved in removing other DNA lesions that cause permanent RNAP arrest.
With regard to transcriptional mutagenesis, it is evident that if these lesions remain unrepaired, a population of mutant RNA transcripts will be generated, subsequently leading to translation and production of mutant protein that could have toxic or other types of effects on cellular physiology. For example, transcriptional mutagenesis could lead to the production of a mutant protein that could cause the aberrant initiation of replication and cell division in a previously quiescent, nondividing cell. As these lesions also cause mutagenic insertions for DNA polymerases, a mutagenic lesion for RNAP that leads to mutant protein through transcription could force a cell to divide, which subsequently, through the miscoding of DNA polymerase, could fix the mutation permanently into the cellular genome (10,23).
We have also shown that 8-oxoG, if located near the promoter of an actively transcribed DNA segment, causes decreased promoter clearance and hence lower amounts of expressed fulllength transcript. Promoter clearance is a target of gene regulation, as several recent studies have shown (24 -28). It has also been demonstrated recently that changes in the ITS in E. coli can markedly affect the ability of the polymerase to escape from the promoter (27). Our results provide further insight gained from these previous studies by showing that DNA damage can also have a "regulatory" effect on transcriptional initiation. There are several implications for decreased promoter clearance due to DNA damage. DNA damage could lower gene expression, which could potentially lead to toxic effects or a myriad of other effects due to inadequate amounts of an essential cellular protein. These results highlight the fact that ways in which specific DNA base damages are able to affect gene expression will depend on their exact location on the template strand of a transcribed gene. The extent to which transcriptional mutagenesis may affect gene expression in living cells is currently unknown and will be an important avenue of pursuit for future studies.