RNA polymerase signals UvrAB landing sites.

Transcription when coupled to nucleotide excision repair specifies the location in active genes where preferential DNA repair is to take place. During DNA damage-induced recruitment of RNA polymerase (RNAP), there is a physical association of the beta subunit of Escherichia coli RNAP and the UvrA component of the repair apparatus (G. C. Lin and L. Grossman, submitted for publication). This molecular affinity is reflected in the ability of the RNAP to increase, in a promoter-dependent manner, DNA supercoiling by the UvrAB complex. In the presence of the RNAP, the UvrAB complex is able to bind to promoter regions and to translocate in a 5' to 3' direction along the non-transcribed strand. As a consequence of this helicase-catalyzed translocation, preferential incision of DNA damaged sites occurs downstream on the transcribed strand. Because of the helicase directionality, the initial binding of the UvrAB complex to the transcribed strand would inevitably lead to its collision with the RNAP. These results imply that the RNAP-induced DNA structure in the vicinity of the transcription start site signals a landing or entry site for the UvrAB complex on DNA.

Transcription when coupled to nucleotide excision repair specifies the location in active genes where preferential DNA repair is to take place. During DNA damageinduced recruitment of RNA polymerase (RNAP), there is a physical association of the ␤ subunit of Escherichia coli RNAP and the UvrA component of the repair apparatus (G. C. Lin and L. Grossman, submitted for publication). This molecular affinity is reflected in the ability of the RNAP to increase, in a promoter-dependent manner, DNA supercoiling by the UvrAB complex. In the presence of the RNAP, the UvrAB complex is able to bind to promoter regions and to translocate in a 5 to 3 direction along the non-transcribed strand. As a consequence of this helicase-catalyzed translocation, preferential incision of DNA damaged sites occurs downstream on the transcribed strand. Because of the helicase directionality, the initial binding of the UvrAB complex to the transcribed strand would inevitably lead to its collision with the RNAP. These results imply that the RNAP-induced DNA structure in the vicinity of the transcription start site signals a landing or entry site for the UvrAB complex on DNA.
In Escherichia coli, NER 1 requires an ensemble of gene products including the UvrABC endonuclease to specifically incise DNA on both sides of a damaged nucleotide (1)(2)(3)(4). The UvrA subunit forms a dimer in solution when driven by the binding energy of ATP (5)(6)(7). The interaction between UvrA and UvrB results in a significant masking of the UvrA-associated ATPase activity (5,8) and an unmasking of the cryptic UvrB-associated ATPase activity (9). This ATPase provides the energy to drive the UvrAB DNA helicase activity, which can displace short complementary fragments (20 -50 oligonucleotides) and D-looped DNA in a 5Ј 3 3Ј direction on the strand to which the UvrAB complex binds (10,11). The UvrAB helicase activity generates supercoiling of relaxed covalently closed circular DNA duplexes (12). These characteristics suggest that the UvrAB complex tracks along DNA scanning for damage in an ATP hydrolysis-dependent manner (13,14) and that the helix-turn-helix (15) and polyhinge regions (16) of UvrA are required for damage recognition. The UvrAB complex eventually leads to the formation of a highly stable UvrB-DNA nucleoprotein complex (17,18). Dual incision is catalyzed by the UvrBC complex at a damaged site.
The binding of the UvrA to undamaged DNA is 10 3 -to 10 4 -fold weaker than to damaged sites (6,19,20). Hence, the specificity of UvrA for damaged sites is quite limited given that UvrA must be able to recognize a single damaged site per E. coli genome. It seems unlikely that UvrA or UvrAB can locate damage by random passive diffusion, suggesting an alternative scanning mechanism for interaction with damaged substrates (6,14,20). The mechanism by which the UvrAB complex recognizes damaged sites during translocation is not yet understood.
A sophisticated requirement of NER is transcription-coupled repair, which quickly targets the repair machinery to genes that are actively transcribed by RNAP. This process is responsible for the rapid removal of possible transcription blocking lesions from the transcribed strand, while the non-transcribed strand is repaired at a slow rate similar to the overall genome (21). Preferential repair was found to be conserved from E. coli (22) to yeast (23) and to mammalian cells (21,24). The identification of those factors that play a role in transcription-coupled repair is the elucidation of the mechanism of this conserved pathway. In E. coli it has been resolved to a considerable detail. A TRCF was isolated that is necessary and sufficient for transcription-coupled repair in a defined in vitro system (25). TRCF, encoded by the mfd gene, is able to recognize and displace the stalled RNAP complex and to lead the UvrABC complex to the site of the lesion possibly due to its affinity for the damage recognition subunit UvrA (26). One of the astonishing recent discoveries was that a basal transcription factor, TFIIH, contains components that are integral to nucleotide excision repair, thus establishing an important link between transcription and DNA repair in higher organisms (27)(28)(29)(30)(31). It is speculated that the preferential targeting of repair proteins to actively transcribed genes in eukaryotes is facilitated by the obligatory loading of repair proteins onto promoter regions during transcription initiation (26).
It is unclear how the translocation of the UvrAB complex is linked to transcription-coupled repair at the molecular level. From reversible cross-linking experiments and resolution of the resultant components of E. coli RNAP with the Uvr proteins during SOS, it is suggested that a physical interaction between RNAP and repair proteins may be required for the transcription-coupled repair. The effect of this physical interaction is the subject of this study in which its effect on damaged DNA was examined.
Here we show that the UvrAB complex preferentially binds to the promoter regions of DNA in an open complex induced by RNAP resulting in enhancing UvrAB induced supercoiling. Further, incision takes place in the strand opposite to which the UvrABC endonuclease binds. The model proposed for the role of RNA polymerase in the site and specificity of nucleotide excision repair is presented.

EXPERIMENTAL PROCEDURES
Materials-E. coli strain (DH5␣) was from Life Technologies, Inc. Deoxyribonucleoside triphosphates (ATP and GTP) were purchased from Pharmacia LKB Inc. Ribonucleoside triphosphates (CTP and UTP) were obtained from BRL. Radioactive compounds (dNTP: 3000 Ci/ mmol) were purchased from Amersham or ICN. Potassium permanganate was from Aldrich. Urea (Ultra-Pure grade) and acrylamide were from Life Technologies, Inc. DTT, formamide, and glycogen were from Boehringer Mannheim. Tris base was from Sigma. Iso-mercaptoethanol was from J. T. Baker. Other chemicals were purchased from Sigma, J. T. Baker, Boehringer Mannheim, or Life Technologies, Inc.
The UvrA, UvrB, and UvrC proteins were purified according to published procedures (33). E. coli DNA topoisomerase I was purified from E. coli strain (RB 968) harboring the plasmid (pJW 312-Sal) encoding the DNA topoisomerase I gene (kindly provided by Professor James C. Wang, Harvard). Proteinase K was from Sigma. BSA was purchased from Sigma or Miles. T7 RNA polymerase and Sequenase were purchased from U. S. Biochemical Corp. E. coli RNA polymerase was obtained from Boehringer Mannheim. Klenow fragment and restriction enzymes were obtained from New England Biolabs. Calf thymus DNA topoisomerase I was obtained from Life Technologies, Inc. Micrococcus luteus UV-endonuclease (1 mg/ml) was from Applied Biotechnology Inc. Plasmid DNAs, pTZ18R (Pharmacia Biotech Inc.) and pHE6 (34), were prepared using the Qiagen column protocol.
UV Irradiation-DNA molecules were placed on parafilm floating on ice water and irradiated by ultraviolet (UV) light (germicidal lamp) at 130 J/min/m 2 .
Preparation of DNA Fragments Containing the P L Promoter-P L promoter fragments were prepared by cutting the plasmid pHE6 with restriction enzymes and labeled at their 3Ј ends using the Klenow fragment or Sequenase. The 461-bp DNA fragments containing the P L promoter were separated on a native polyacrylamide gel after restriction enzyme digestion with BglII and SmaI. This fragment was digested with HpaII and then labeled by filling in with [␣-32 P]dCTP for the non-transcribed strand (NTS). To label the transcribed strand (TS), BglII-SmaI digested fragments were filled in with [␣-32 P]dGTP and then digested with the HaeIII restriction enzyme. The sample was loaded onto a 5% polyacrylamide gel with 1 ϫ TBE (89 mM Tris, 89 mM borate, and 1 mM EDTA) buffer. The band corresponding to the expected size was excised after electrophoresis and then sliced into small pieces. The labeled DNA fragments were eluted in TE (pH 8.0) overnight at 37°C. The recovery of labeled DNA was about 70%.
Supercoiling Assay-The procedure used for following supercoiling has been described previously (12). Negatively supercoiled plasmid DNA (pTZ18R or p18R) purified using the Qiagen column protocol was relaxed to completion with calf thymus DNA topoisomerase I as follows; DNA was incubated in buffer (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM MgCl 2 , 0.1 mM EDTA, 30 mg/ml BSA) with calf thymus DNA topoisomerase I for 2 h and further purified by double phenol extraction followed by phenol/chloroform (1:1) extraction and ethanol precipitation just prior to reaction. The degree of relaxation of DNA was checked by agarose gel electrophoresis.
Each supercoiling reaction (20 l) contained 30 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 50 mM KCl or 50 mM NaCl, 1 mM DTT, 50 mg/ml BSA, 2-4 mM ATP, and 40 ng of relaxed DNA. Incubation was performed at 37°C for various times in the presence of UvrA, UvrB, and E. coli DNA topoisomerase I. To monitor the superhelical states of DNA templates, the reactions were terminated by adding stop solution (0.5% SDS, 25 mM EDTA) and 6 g of proteinase K and then incubation continued for 2 h. DNA samples were run on horizontal agarose gel (20.5 ϫ 24 cm) electrophoresis (0.9% agarose in 0.5 ϫ TPE (44.5 mM Tris phosphate, 1 mM EDTA) buffer) for 14 -16 h at 50 V at room temperature. After electrophoresis, hybridization was followed as described previously (12). For two-dimensional electrophoresis, DNA samples were loaded in four positions. Once one-dimensional electrophoresis was completed, the gel was soaked in 8 M chloroquine for 2-3 h. The gel was placed on a gel apparatus with 90°rotation. Electrophoresis was carried out in the buffer containing 8 M chloroquine for 8 h at room temperature. The extent of highly positive supercoiled form (ϩ) was quantified by the AMBIS Optical Imaging System.
Footprinting-E. coli RNAP was mixed with 165 pM labeled DNA and 2.5 M unlabeled plasmid DNA (bp) in a reaction buffer containing 30 mM Tris-HCl, 10 mM MgCl 2 , 50 -100 mM KCl, 100 g/ml BSA, 1 mM DTT. The mixture was incubated at 37°C for 15 min. One microliter of DNase I (40 ng/l) was added successively to the incubation mixture, which was mixed rapidly. The mixture was further incubated at 37°C for 30 s, and DNase I digestion was stopped by the addition of 20 l of stop solution (4 M NH 4 OAc and 35 mM EDTA). The DNA was purified by extraction with phenol/chloroform (1:1) and precipitated with 20 g of glycogen (Boehringer Mannheim) and 4 volumes of 100% ethanol. The DNA pellet was washed with 0.5 ml of 70% ethanol. The pellet was dried and resuspended in 4 l of loading buffer (85% formamide, 1 ϫ TBE, 0.01% xylene cyanol, 0.01% bromphenol blue). All samples were heated at 90°C for 3 min and chilled immediately on ice. Samples were loaded onto a 6% urea-sequencing gel (38 ϫ 42 ϫ 0.04 cm, 19:1 (w/w); acrylamide:bisacrylamide, 7 M urea, 1 ϫ TBE). For sequencing comparisons, Maxam-Gilbert sequencing was carried out (35).
KMnO 4 Reactions on Transcription Complexes-Transcription complexes were formed as in the footprinting experiment, and treated with 1 l of 100 mM KMnO 4 and incubated for 2 min at 37°C (36). The reactions were terminated by the addition of 2 l of 8.2 M of ␤-mercaptoethanol and placed on ice, followed by the addition of 23 l of stop solution. After phenol/chloroform extraction, each sample was precipitated with 100% ethanol and 20 g of glycogen and then treated with 1 M piperidine as described by Maxam and Gilbert (35). Sequencing reactions were performed as described by Maxam and Gilbert with the identical substrate used in transcription complexes. After drying in vacuo, all samples were resuspended in loading buffer and electrophoresed through 5% urea-sequencing gels. Gels were dried and exposed to Kodak XAR-5 film using a Dupont Lighting Plus intensifying screen or to a Phosphor imaging plate.
Incision-Incision by the UvrABC endonuclease or M. luteus UVendonuclease was conducted in a volume of 20 l. The final reaction mixtures contained 30 mM Tris-HCl (pH 7.6), 85 mM KCl, 10 mM MgCl 2 , 1 mM DTT, 2 mM ATP. After incubation of the UV-irradiated (400 J/m 2 ) DNA substrates with RNAP at 37°C for 10 min, UvrA (0.1-0.2 pmol) and UvrB (0.3-0.6 pmol) protein were added and then incubated for another 10 min. Incision was initiated by the addition of UvrC(0.3 pmol) protein and incubated for 5 min. The reactions were stopped by the addition of 21 l of stop solution and placed on ice. The samples were extracted with phenol/chloroform (1:1) to remove the proteins followed by precipitation with absolute ethanol. The DNA was pelleted, washed with 70% (v/v) ethanol, and dried. Pellets were resuspended in 4 l of loading solution and heated for 3 min at 90°C. The samples were loaded onto a 5-6% urea-sequencing gel. The M. luteus UV-endonuclease incision was carried out in the buffer as described previously (37) or under the same conditions as in the UvrABC incision reaction.

RESULTS
The Effect of RNA Polymerase on UvrAB Supercoiling-RNA polymerase unwinds DNA as it binds to the promoter unwinding DNA continuously as the enzyme extends the growing nascent RNA chain (38). The localized unwinding results in the transient formation of positive supercoils ahead of the RNAP complex and negative supercoils behind it (39 -41). Simian virus 40 large tumor antigen (SV40 T antigen) acts similarly as a helicase and is able to unwind duplex DNA producing positive and negative supercoils in an ATP hydrolysis-dependent reaction (42). This process relies on the translocation of proteins along DNA which is accompanied by the unwinding of closed circular duplex DNA. Like SV40 T antigen and RNA polymerase, the UvrAB complex can also unwind DNA generating both positive and negative supercoils in an ATP hydrolysis-dependent reaction (12). It seems that the initial entrance into DNA by helicases such as SV40 T antigen and the UvrAB complex is due to localized unwinding because DnaB and UvrD, which cannot unwind duplex DNA by just binding to it, are unable to generate supercoils. 2 However, the efficiency with which the UvrAB complex unwinds duplex DNA has not yet been determined. RNA transcription-dependent supercoiling requires the specific binding of RNAP to its promoter for translocation along a transcription unit, whereas SV40 T antigen and the UvrAB complex can bind to DNA nonspecifically. It, however, has not yet been determined where the UvrAB complex initiates translocation.
The specific interaction between the ␤-subunit of RNAP and UvrA 3 provides insights into the potential interaction between transcription and UvrAB catalyzed translocation. Supercoiling reactions were performed under non-transcribing conditions to avoid the complication of dealing with the supercoiling introduced by RNA transcription. UvrA and UvrB proteins were incubated with the closed circular relaxed plasmid, pTZ18R, in the presence of ATP and E. coli DNA topoisomerase I protein.
The accumulation of highly positively supercoiled species of DNA (marked in (ϩ)) was greatly increased by T7 RNAP (28- Fig. 1B). Supercoiling stimulation may reflect UvrAB tracking along the template DNA because of the supercoiling reaction under these non-transcribing conditions. Enhancement of supercoiling by the UvrAB complex seems to be due to the binding of RNAP to its promoter (this paper).
Topoisomeric Forms Generated by the UvrAB-Two-dimensional gel electrophoresis was used to determine the nature of the DNA species formed in the presence of E. coli DNA topoisomerase I, UvrA, UvrB, and RNAP. The first dimension was carried out in TPE buffer, and the second dimension was performed in chloroquine-containing TPE buffer (Fig. 1C). The electrophoretic positions of relaxed DNA (d), nicked form (a), and highly positively supercoiled forms (c) are shown (1 in Fig.  1C). It was reported in previous studies (12) that highly positive supercoils are formed by UvrAB proteins in this supercoiling reaction (2 in Fig. 1C). When the DNA substrate was incubated in the presence of E. coli DNA topoisomerase I, UvrA, UvrB, and T7 RNAP, highly positively supercoiled DNA was produced as the amount of T7 RNAP was increased (3 and 4 in Fig. 1C). These data are in agreement with those in Fig. 1A and show that the UvrAB complex can generate transient positive and negative supercoils in the presence of RNAP under non-transcribing conditions.
The Effect of UV Irradiation of the DNA on Supercoiling-The supercoiling by the UvrAB complex was greatly increased when UV-irradiated relaxed DNA was used in the supercoiling reaction (12). It has been postulated that DNA damaged sites may provide the UvrAB complex with an anchoring site enhancing the translocation of the UvrAB complex along the DNA. Therefore, the binding of the UvrAB complex to DNA was increased, resulting in increasing the number of plasmid molecules being converted into highly positively supercoiled form. In Fig. 1D, supercoiling by the UvrAB complex is increased (8.3-fold) as a consequence of UV damage (393 J/m 2 ) (lane 4), as reported previously (12,13). The increase (14-fold) in the formation of highly positive supercoils in the presence of RNAP (lane 5) was greater than that of damaged DNA alone. When UV-damaged DNA and RNAP were included in the supercoiling reaction, supercoiling (19.5-fold, lane 6) was greater than that of undamaged DNA (lane 5) and that in the absence of RNAP (lane 4). The enhancement of supercoiling suggests that RNAP and/or UV-damaged sites provide binding sites for the UvrAB complex. The shift of topoisomers (lane 2) is due to the unwinding of DNA by the presence of photodimers.
Promoter Requirement for Enhancement of Supercoiling-T7 promoterless plasmids were constructed to determine if the specific binding of RNA polymerase to its promoter is required for enhancement of the UvrAB supercoiling. The T7 promoter region of pTZ18R was excised as a 365-bp fragment by PvuII restriction enzyme followed by ligation. It can be seen in Fig. 2 that the positively supercoiled form (ϩ) was generated by the UvrAB complex with the DNA template containing the T7 promoter (lanes 1 and 2). Under conditions of preincubation with T7 RNAP and relaxed plasmid pTZ18R, there was no further effect on the extent of supercoiling (lanes 3-7). However, in the reaction with the T7 "promoterless" plasmid DNA, the extent of supercoiling was not affected with synergistic levels of T7 RNAP. There were some changes within the spectrum of DNA topoisomeric species generated by the T7 RNAP alone (lane 12) without the UvrAB complex, suggesting some nonspecific binding of T7 RNAP under these experimental conditions. The data from these experiments suggest that the specific binding of T7 RNAP to its promoter may be related to the enhancement of DNA supercoiling by the UvrAB complex and that the RNAP may exert its influence through its effect on DNA structure. The location and extent of binding of UvrAB may be enhanced by the specific binding of RNAP to its promoter. It suggests that the local DNA structure around the promoter is a potential site for the strand and site preferential binding of the UvrAB protein complex.
E. coli RNAP Footprinting with P L Promoter-DNase I footprinting was performed to localize the binding of the UvrAB complex to DNA influenced by E. coli RNAP. P L promoter fragments isolated from plasmid pHE6 were incubated with E. coli RNAP and then digested with DNase I. E. coli RNAP bound to the P L promoter fragment protects a region from Ϫ48 to ϩ18 on the non-transcribed strand and from Ϫ48 to ϩ20 on the transcribed strand (Fig. 3). This cleavage site is similar to that observed by Oppenheim's group (43). The DNase I footprinting pattern obtained for RNAP at P L is also similar to other strong promoters like those in the lac UV5 and P R promoters. For both promoters, protection by RNAP is extended to about 20 nucleotides downstream beyond the transcription start site. The UV5 promoter was protected upstream of the start site to around Ϫ52 (44) and the P R was protected in the promoter to position Ϫ57 and to Ϫ52 on the transcribed and non-transcribed strands, respectively (45).
In Fig. 3, the DNase I footprinting pattern reflects the location of RNAP, and the UvrAB complex in the presence of ATP. Lanes 6 and 12 are control lanes of free DNA. Lanes 1 and 7 are those regions protected by RNAP. The UvrAB footprints in lanes 4, 5, 10, and 11 show no specific binding by the UvrAB complex under these reaction conditions. When UvrAB and RNAP were incubated together, there is an extended protection region downstream (position a showed 50% protection) beyond the transcription start site at ϩ1 in only the non-transcribed strand (lanes 2 and 3 in Fig. 3B), whereas no like protection is seen in the transcribed strand (lanes 8 and 9). The extended region of protection in the non-transcribed strand is not well defined. It is likely that the UvrAB complex is moving downstream as a consequence of ATP hydrolysis. The extended region of protection in the non-transcribed strand is seen with T7 RNAP as well (data not shown). The direction of the region on the non-transcribed strand of the DNA helix is 5Ј to 3Ј, which is consistent with the directionality of the UvrAB helicase. The UvrAB complex should predominantly bind to this strand to translocate along the strand in a 5Ј to 3Ј direction.
Potassium Permanganate Reactivity-The thymine residues in the single-stranded region of DNA within the open complex (so called transcription bubble) were sensitive to KMnO 4 reactivity. The oxidized thymine residues are detected by piperi- dine-catalyzed cleavage of the DNA strand followed by denaturing gel electrophoresis. KMnO 4 oxidizes the 5-6 double bond of thymine to create a diol. Thus, thymine residues in B-form DNA are protected from attack by the bases stacked above and below them, and only those thymine residues in single-stranded or structurally altered DNA regions are efficiently modified (46,47). Thymines within the single-stranded synthetic bubble or loop regions of the DNA substrate react uniformly with KMnO 4 (59).
To examine directly the formation of a single-stranded regions in the P L promoter, KMnO 4 reactions were performed in the presence of RNAP (Fig. 4A). Thymines in positions Ϫ11T, Ϫ10T, Ϫ2T, and Ϫ51T (lane 5) in the transcribed strand and -7T (lane 10) in the non-transcribed strand were modified by KMnO 4 in the presence of E. coli RNAP (Fig. 4A) and cleaved by piperidine treatment. This cleavage is specific for the active form of RNAP because cleavage is not observed in the presence of heat-denatured RNAP (data not shown) or in the absence of RNAP (lane 3). This indicates that the region between Ϫ11 and Ϫ2 (Fig. 4B) may be uniquely melted generating a unique single-stranded region. Oppenheim's group observed similar KMnO 4 reactivities, but more reactive sites were on the nontranscribed strand in the pL1 promoter (43). Open complexes of RNAP and the lac UV5 promoter (36) and late promoters in phage (48) melt the DNA helix from Ϫ10 to ϩ4 (lac UV5), from Ϫ11 to ϩ3 or ϩ4 (phage ), from Ϫ11 or Ϫ12 to ϩ4 or ϩ5 (phage 82), and from Ϫ12 to no more than ϩ5 (phage 21) are melted since thymines in this region are reactive with KMnO 4 . Those modified bases are detected by primer extension treatment. In the presence of UvrAB (lanes 6 and 11), the reactivities of these sites were significantly diminished (70%) on both the transcribed and non-transcribed strand. However, the extent of reduction is less decreased in the presence at equal molar concentrations of single-stranded binding protein (data not shown). It is likely, therefore, that the binding of UvrAB occurs at the distorted promoter region generated by RNAP binding and such binding allows the UvrAB complex to specifically translocate along DNA in a 5Ј to 3Ј direction on only the non-transcribed strand.
Incision Reactions-The DNA fragments used in the DNase I footprinting were irradiated with UV-light (254 nm) and were employed to determine those sites incised in UV-irradiated DNA by the UvrABC endonuclease which occurs on both sides (3Ј and 5Ј) of the damaged site (1,2). The incision experiments are shown in Fig. 5A. Lane 3 shows the DNase I protection in the promoter region in the transcribed strand by the E. coli RNAP. Heat-denatured RNAP provides no protection (lane 2). Incision at the 3Ј side of the damaged site was detected with the 3Ј end-labeled substrate when catalyzed by the UvrABC endonuclease (lanes 4 and 7). In lane 5, the extent and sites of incision by the UvrABC complex in the presence of heat-denatured RNAP is the same as that in lane 4. In the presence of active RNAP, the two bands (a and b) at the promoter region disappeared, whereas the intensity of incised sites downstream (lane 6 in Fig. 5A) 11 and 13), which is lower than that in the presence of ATP (lanes 4 and 7 in Fig. 5B). It is likely that the UvrAB complex is able to displace stalled RNAP complexes.
The M. luteus UV-endonuclease recognizes UV-damaged sites and hydrolyzes the phosphodiester bond between sugar moieties of pyrimidine dimers in UV-irradiated DNA (37). Incision of the M. luteus UV-endonuclease incision is unaffected by RNAP. However, under transcribing conditions, preincubation with RNAP decreased incision by the M. luteus UV-endonuclease (lanes 3 and 6 in Fig. 6).
Therefore, the increase in incision of damaged sites on the transcribed strand is specific for UvrABC. It is likely that the preferred binding of the UvrAB complex to the non-transcribed strand in the presence of RNAP led to increased incision downstream from the transcription start site in the transcribed strand, suggesting that the UvrAB complex translocates along the non-transcribed strand and recognizes damaged sites on the complementary (the transcribed strand).

DISCUSSION
The findings presented herein suggest that RNAP enhances UvrAB supercoiling as a consequence of the specific binding of RNAP to its promoter. The proposed DNA structure, of an open complex surrounding a transcription start site, provides a po- tential landing or entry site for the UvrAB complex into DNA. The UvrAB complex should be free to translocate along the non-transcribed strand as a consequence of its 5Ј to 3Ј unidirectionality, whereas its binding to the transcribed strand would lead to a collision with the RNAP. Incision of DNA damaged sites on the transcribed strand occurs preferentially as the UvrAB complex translocates along the non-transcribed strand.
There are several explanations for the enhancement of supercoiling. First, negative and positive supercoils generated by the protein tracking along the DNA could spontaneously diffuse along the DNA helical axis. The supercoiling assay was developed to monitor the formation of highly positive supercoils (41), and its sensitivity clearly relies on the action of E. coli DNA topoisomerase I. If the relaxation rate of negative supercoils is slower than the diffusion rate of both supercoils, then positive supercoils may not accumulate. If diffusion of both supercoils slows down, the extent of the resultant supercoiling should increase. Therefore, the binding of RNAP may be one of many factors able to block the diffusion of supercoils. It has been speculated, however, that the frictional torque of the transcription complex, including the nascent mRNA, has to be greater than the DNA supercoiling torque in order to prevent the positively and negatively supercoiled regions from mutual annihilation (39,41). Thus, the RNAP itself may not be sufficient to block mutual annihilation by diffusion. In this experiment, it seems that the binding of RNAP may not contribute to the blocking of rotational diffusion of both supercoils because the supercoiling reactions in this work were performed in the absence of transcription. Second, the binding of RNAP to its promoter may create DNA topoisomerase I-accessible sites. However, topo I acts randomly on negatively supercoiled DNA regions. Third, the extent of supercoiling was saturated when the supercoiling reaction was incubated for 1 h (12). If the UvrAB complex dissociates from highly positively supercoiled DNA and rebinds to the rest of relaxed DNA, the extent of supercoiling would be expected to be concentration-independent. However, the extent of supercoiling was increased as the concentration of the UvrAB complex was increased even when the reactions were incubated for 1 h (12), implying that the final extent of supercoiling may be proportional to the number of plasmid molecules initially bound by the UvrAB complex. RNAP seems, as a consequence, to increase binding sites for the UvrAB complex in this study.
The binding of UvrA to DNA forms complexes that are very short lived because of nucleotide binding and hydrolysis (6,49). In solution, UvrA can interact with UvrB forming a UvrAB complex, resulting in the modulation of the UvrA associated ATPase (5). The helicase activity of the UvrAB complex reflects the active UvrAB complex binding to a single-stranded DNA region with the stimulation of its ATPase. At least a 10-nucleotide-long fragment at the 5Ј single-stranded flanking region is required for helicase activity (50). Strand displacement is not due to just unwinding of the duplex DNA upon the binding of the UvrAB complex to DNA but does need the directional movement of the UvrAB complex (11). The DNA unwinding efficiency of the UvrAB complex upon binding to DNA is not known. Even though DNA binding of the UvrAB complex is necessary for unwinding, each DNA binding event may not induce concomitant unwinding. For instance, using as a helicase substrate, a 20-bp partial duplex consisting of a very long 3Ј tail but no overhang at 5Ј end, there was no helicase-catalyzed displacement reaction (11). If the UvrAB complex can bind to the duplex region of this substrate and unwind one helical turn and translocate along the DNA in a 5Ј to 3Ј direction, a 20-bp region with two helical turns can be easily displaced by the UvrAB complex. It seems that the preferential binding of the UvrAB complex to a single-stranded region or 5Ј-overhang may render the UvrAB complex competent as a substrate for helicase action.
Promoter regions around nucleotide position (Ϫ10) are exposed as a consequence of the binding of the RNAP. For the lac UV5 and tac promoters, cytosines at positions Ϫ6, Ϫ4, Ϫ2, and Ϫ1 are sensitive to methylation, indicating that these cytosines are in unpaired regions (51). Using KMnO 4 , the singlestranded region of lac UV5 promoter (36) and the late promoters of phages (phage , phage 82, phage 21) (48) in the open complex were monitored. The KMnO 4 reactive sites in the P L promoter region are located in the region from Ϫ11 to Ϫ2, which is similar to these other promoters, providing RNAPinduced unpairing of bases is in a specific region of the P L promoter. The KMnO 4 reactivity of thymines in the promoter region is greatly decreased in the presence of the UvrAB complex, suggesting that the UvrAB complex binds to the sites of bound strand in a 5Ј to 3Ј direction. A unique damage recognition domain in the UvrAB complex may dock to the complementary strand. The extent of DNase I protection of the UvrAB complex on the undamaged strand of duplex DNA is greater than that on the UV-damaged strand (19), suggesting asymmetric binding of the UvrAB complex to duplex DNA. The 3-1.6-fold increase (toward the 5Ј-end) of incision on the downstream of the transcribed strand in Fig. 5 (A and B) is in agreement with the increase of DNA repair in terms of in vitro and in vivo DNA repair synthesis of the transcribed strand (22,52). That the M. luteus UV-endonuclease neither unwinds DNA nor has helicase activity (37) shows no preferential strand incision in the presence of RNAP. Further, M. luteus UVendonuclease-catalyzed incision is greatly decreased under transcribing conditions (Fig. 6).
For the following reasons, preferential incision of DNA-damaged sites on the transcribed strand as a consequence of the unique translocation of the UvrAB complex may contribute in part to specific transcription-coupled repair.
First, transcription-coupled repair is responsible for the rapid removal of possible transcription-blocking lesions from the transcribed strand, while the non-transcribed strand is repaired at a slower rate similar to that of the overall genome. TRCF appears to be necessary for transcription-coupled repair in a defined in vitro system. TRCF, encoded by the mfd gene in E. coli, is able to recognize and displace a stalled RNA polymerase leading the UvrABC complex to the lesion presumably via its affinity for the damage recognition subunit of UvrA (26). Parenthetically, however, the stimulation of repair synthesis has not been demonstrated with this coupling factor. In addition, the mfd gene encoding the TRCF null mutant does not render these mutants UV sensitive or sensitive to other damaging agents (26), indicating a minor contribution of transcription-coupled repair to survival of cells after genotoxic treatment. In eukaryotes, the yeast rad26 null mutant is also insensitive to UV when compared to wild-type yeast (53). Repair of the transcribed strand is no more strongly inhibited or incomplete in this mutant. Apparently, yeast has other means to relieve the inhibition of repair caused by a stalled RNA polymerase complex. Although the stability of the stalled RNAP complex has not been determined, the replication machinery can dissociate the blocked RNA polymerase from the DNA (54). Even though the ability of the UvrAB complex to displace the stalled RNAP complex remains to be studied, the data in Fig. 5B indirectly show displacement of the stalled RNAP complex because the incision is increased on the transcribed strand under transcribing conditions. Second, mammalian genes are transcribed more slowly than genes in E. coli. It would seem inefficient to abort nearly completed transcripts of such genes every time RNA polymerase encounters a lesion (55). The transcription elongation factor SII may provide an alternative mechanism; this factor catalyzes nascent transcript cleavage by RNA polymerase II at natural pause sites, enabling the polymerase to back off and try again without aborting the incomplete transcript (56). However, studies of the displacement of stalled RNA polymerase complex at varying sites have not been undertaken in E. coli.
Third, it has been recently discovered that one of the transcription initiation factors, TFIIH (factor b), contains components that are involved in DNA nucleotide excision repair mechanism, thus establishing an important link between transcription and repair in higher organism. TFIIH obviously is able to repair DNA damage in cell free extracts of XP-D, XP-B, or the rad3 mutant (27,29,31). TFIIH contains XP-B (RAD25) and XP-D (RAD3) subunits of nucleotide excision repair and possesses DNA helicase activity (28,30). These experiments suggest that the helicase activity of this factor is necessary in DNA repair, even though its exact role as a helicase is not known. A recent study showed that fast repair rates are seen near the transcription initiation sites, and there is a general gradient of repair efficiency of the transcribed strand with faster repair within the 5Ј-end and diminished repair toward the 3Ј-end of the gene (57). This study may be explained either by increased local concentrations of DNA repair factors that are associated with TFIIH functioning in transcription initiation (57) or by the loading of TFIIH onto promoter region during transcription initiation providing preferential targeting of repair proteins to actively transcribed genes (32).
Results of this work lead us to propose that the RNA polymerase provides the UvrAB complex with preferred binding sites in the promoter region (Fig. 7). The physical interaction of RNAP with the UvrAB complex, 3 in a promoter-dependent manner, may recruit the UvrAB complex to transcription bubble regions. Some transcription activators can bind to a DNA site adjacent or within the promoter of the RNAP-promoter complex (58). Once the UvrAB complex binds to the bubble region, a productive nucleoprotein complex may be formed, the UvrAB complex then translocates unidirectionally along the non-transcribed strand (5Ј to 3Ј) in an ATP hydrolysis-dependent reaction. Because of the helicase unidirectionality, translocation along the transcribed strand would be blocked by collision with the RNAP. During translocation along the nontranscribed strand, the DNA damage recognition domains (helix-turn-helix and polyhinge) of the UvrA subunit (15,16) of the UvrAB complex sense damaged sites on the complementary strand. Once the UvrAB complex encounters a damaged site, the polyhinge region of the UvrA subunit of the UvrAB complex can form a stable nucleoprotein complex. The helicase action of the UvrAB complex may then be turned off (11). This stalled Uvr complex may further influence the dissociation of the UvrA subunit from the UvrAB-DNA complex, resulting in an UvrB-DNA complex (17).