Interaction between RNA Polymerase and RapA, a Bacterial Homolog of the SWI/SNF Protein Family*

Recently, we identified a novel Escherichia coli RNA polymerase (RNAP)-associated protein, an ATPase, called RapA (Sukhodolets, M. V., and Jin, D. J. (1998) J. Biol. Chem. 273, 7018–7023). RapA is a bacterial homolog of SWI2/SNF2. We showed that RapA forms a stable complex with RNAP holoenzyme and that binding to RNAP holoenzyme stimulates the ATPase activity of RapA. We have further analyzed the interactions between purified RapA and the two forms of RNAP: core RNAP and RNAP holoenzyme. We found that RapA interacts with either form of RNAP. However, RapA exhibits higher affinity for core RNAP than for RNAP holoenzyme. Chemical cross-linking of the RNAP-RapA complex indicated that the RapA-binding sites are located at the interface between the α and β′ subunits of RNAP. Contrary to previously reported results (Muzzin, O., Campbell, E., A., Xia, L., Severinova, E., Darst, S. A., and Severinov, K. (1998) J. Biol. Chem. 273, 15157–15161), our in vivo analysis of a rapAnull mutant suggested that RapA is not likely to be directly involved in DNA repair.

In Escherichia coli, RNA polymerase (RNAP) 1 exists in two forms: core RNAP and RNAP holoenzyme. The basic transcription machinery, core RNAP, consisting of subunits ␣ 2 ␤␤Ј, is capable of transcription elongation and termination, but it is unable to initiate transcription; whereas RNAP holoenzyme, consisting of ␣ 2 ␤␤Ј, is capable of initiating transcription from promoters on a DNA template (1,2). A number of proteins that associate with core RNAP and/or holoenzyme and participate in different aspects of transcription have been identified, such as NusA (3,4), GreA, and GreB (5)(6)(7)(8)(9).
Recently, we described a new RNAP-associated protein named RapA with a molecular mass of 110 kDa (10). Interestingly, the RapA protein is a member of the SWI/SNF protein family (10 -16), which has been implicated in eukaryotic nucleosome remodeling and DNA repair (for reviews, see Refs. 16 -18). We found that the RapA protein co-purifies with RNAP holoenzyme exclusively and that after purification to homogeneity, RapA is capable of forming a stable complex with RNAP holoenzyme in vitro (10). In addition, the RapA protein is an ATPase, and its ATPase activity is stimulated upon binding to RNAP holoenzyme, indicating that RapA interacts with RNAP holoenzyme both physically and functionally. Independently, Muzzin et al. (19) also reported that the same 110-kDa protein (but named HepA) is associated with RNAP in E. coli. However, they found that this new protein appeared to be associated only with core RNAP but not with RNAP holoenzyme (19). Thus, it is important to address whether and how RapA interacts with the two forms of RNAP. This is not only an unsolved issue but could also shed some light on the function of RapA.
In this study, we further characterized the interaction between RNAP and the RapA protein biochemically. Specifically, we analyzed the conditions that affect the interactions between RapA and either core RNAP or RNAP holoenzyme and measured the dissociation constants (K d ) of RapA-RNAP complexes. We also identified the subunits of RNAP that are in close contact with RapA by chemical cross-linking. Furthermore, in an attempt to address the cellular function of RapA, we constructed a rapA null mutation and studied the effect of this mutation in vivo.

EXPERIMENTAL PROCEDURES
Materials and Chemical Reagents-The RapA protein and RNAP were purified from E. coli K12 cells (MG1655) as described previously (10). The protein concentrations were determined using the Bradford assay (20) with bovine serum albumin as a standard. RNAP concentrations were also determined by UV absorbance using the molar extinction coefficient data of Lowe et al. (21). The RapA-specific and -70specific polyclonal antibodies were previously described (10). The monoclonal antibodies specific for the ␣ (4RA1), ␤ (NT63), or ␤Ј (NT73) subunits of RNAP were kindly provided by Dr. Richard Burgess (University of Wisconsin, Madison, WI), and the polyclonal antibodies specific for the subunit of RNAP were kindly provided by Dr. Dan Gentry (SmithKline Beecham). The bifunctional cross-linker ethylene glycol bis(succinic acid N-hydroxysuccinimide ester) (EGS-NHS) was purchased from Sigma.
Reconstitution of RNAP-RapA Complex in Vitro-Stability of the complex of RapA with either core RNAP or RNAP holoenzyme was studied by gel filtration using a Superose 6 HR 10/30 column (Amersham Pharmacia Biotech) as described previously (10). All runs were performed in TGED buffer (0.01 M Tris-HCl, pH 7.9, 5% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol) with the salt concentrations indicated in the legends for Figs. 1 and 2.
Glycerol Gradient Ultracentrifugation Experiments-Purified enzymes were premixed in 100 l of glycerol gradient centrifugation buffer (10 mM Tris, pH 7.8, 10 mM MgCl 2 , 0.1 mM EDTA, 0.1 mM dithiothreitol, 200 mM NaCl) and layered on top of 4 ml of 15-30% (top to bottom) linear gradients of glycerol in the above buffer. The samples were then spun in a SW-60 rotor for 21 h at 37,500 rpm (6°C). Each 4-ml tube was fractionated into 13 fractions using the Beckman Fraction Recovery system. Equal volumes of 2ϫ Laemmli sample buffer were then added to each fraction, and the samples were analyzed on SDS 10% polyacrylamide gels. The gels were stained with Coomassie Brilliant Blue R-250 or silver.
ATPase Assays-The ATPase activities were determined by measuring the amount of [␣-32 P]ADP released from [␣-32 P]ATP. The reaction conditions and separation of samples by chromatography using poly-(ethyleneimine)-cellulose plates (J.T. Baker Inc.) were as described (10). Plates were autoradiographed using Kodak Bio-Max MR film and scanned on a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA) to quantitate the amount of [␣-32 P]ATP hydrolyzed. The ATPase activity was expressed as pmol of ATP hydrolyzed/min/g of RapA.
* 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 10; this work). Sequential dilutions of purified RapA alone or purified RapA premixed with either purified core RNAP or RNAP holoenzyme (2 mol of RapA per 1 mol of RNAP) in binding buffer (40 mM Tris-HCl, pH 7.4, 4 mM MgCl 2 , 1 mM dithiothreitol, 50 g/ml bovine serum albumin, 50 mM NaCl) were made. It was found necessary to use siliconized tubes and maintain bovine serum albumin in the buffer to eliminate nonspecific binding of proteins to the tube walls, particularly at low protein concentrations. ATPase activity of each dilution was determined and calculated as pmol of ATP hydrolyzed/min/g of RapA (A obs ). The difference between the ATPase activities of RNAP/RapA mixtures and RapA alone at each RapA concentration was calculated as ⌬A obs at that particular RapA concentration. The value of ⌬A max (maximal activation) was calculated from the double-reciprocal plot (1/⌬A obs versus 1/[RapA]). The ratio of ⌬A obs and ⌬A max was calculated and plotted as a function of RapA concentration. The apparent K d values were obtained directly from these adsorption isotherms as the RapA concentrations corresponding to half-maximal ⌬A obs /⌬A max. Each titration was repeated at least twice with similar results.
Chemical Cross-linking-Purified RNAP or a 1:1 RNAP-RapA complex, typically at a final concentration of 2.6 M, was incubated with 1 mM EGS-NHS in 50 mM HEPES, pH 7.8, in a final volume of 100 l. Reactions were incubated as specified in the legends for Figs. 6 and 7 and quenched by the addition of 100 l of 2ϫ Laemmli sample buffer. The cross-linked protein products were analyzed by electrophoresis on SDS-5% polyacrylamide gels and subsequently silver-stained or transferred onto Immobilon-P membranes (Millipore Corp.), followed by immunostaining with antibodies specific for RapA or various subunits of RNAP. Anti-rabbit and anti-mouse peroxidase-conjugated secondary antibodies (Calbiochem) were used to visualize the membrane-bound antibody-antigen complexes.
Cloning of the rapA Gene-The rapA gene was cloned into the expression vector pBAD24 (22) as follows. 1) A DNA fragment containing the rapA gene was amplified by polymerase chain reaction using the primers DJ142 (5Ј-TAG CAG GAG GAA TTC ACC ATG CCT TTT ACA CTT GGT CAA CGC TGG), which contains the sequence upstream of the EcoRI site of pBAD24 and the beginning of the coding sequence for rapA, and DJ143A (5Ј-ACA CTT ATC AAG CTT TAT GGT CAT CCT GAT ACA GGA TAA CCA ACC A), which covers sequences about 80 base pairs downstream of the rapA gene and introduces a HindIII site.
2) The amplified DNA fragment was purified, digested with the restriction enzymes EcoRI and HindIII, and ligated with the pBAD24 vector DNA, which had been digested with EcoRI and HindIII. The resulting plasmid pDJ61 (Amp r ) was confirmed by restriction mapping and by arabinose induction of rapA.
Construction of the rapA Null Mutation-In pDJ61, there are only three PvuII sites, all of which are located in the middle of the rapA gene. After the pDJ61 plasmid DNA had been digested with the restriction enzyme PvuII, the largest DNA fragment (6.5 kilobase pairs), which contained some vector sequence and parts of the rapA gene (including the coding sequence from base pairs 1-752 and base pairs 2042-2906 but deleted most of the conserved regions in the SNF2 family), was purified. This 6.5-kilobase pair DNA fragment was then ligated with a 1.0-kilobase pair DNA fragment containing a cat gene conferring chloramphenicol resistance (Cm r ), which was purified from pCAT19 (23) after it had been digested with the restriction enzyme SmaI. The resulting plasmid pDJ62 was linearized by digestion with the restriction enzyme EcoO109I (a unique site in vector), followed by transformation with the strain JC7623, which contains the recBC and sbcBC mutations (24). Recombinants between the linear DNA and the chromosome were selected on LB ϩ Cm plates. The true recombinants were Cm-resistant but Amp-sensitive. The rapA::cat null mutation (both the deletion and insertion mutation) was confirmed by the following criteria: 1) the absence of detectable RapA protein in the rapA null mutant as determined by two-dimensional electrophoresis and immunoblotting with antibodies raised against RapA and 2) the demonstration of linkages between rapA::cat and leu and between rapA::cat and carA by P1 transductions.
Bacterial Techniques-Bacterial media and techniques were as described (25). Bacterial growth was followed by measuring the optical density of cultures at 600 nm with a spectrophotometer. The rapA::cat null mutation was transferred into different strain backgrounds by P1 phage-mediated transduction, and the rapA null mutant was selected on LB ϩ Cm plates. The mfd mutant (26 -28) with a linked Tn10 marker was obtained from Dr. Aziz Sancar (University of North Carolina). The mfd mutation was transferred into other strains by P1 phage-mediated transduction. The linkage between mfd and Tn10 is Ͼ50%, and the mfd phenotype was screened by UV sensitivity at a high UV irradiation dosage (28). Spontaneous or UV-induced rifampicinresistant (Rif r ) mutations were isolated as described (29). For UV mutagenesis (25), about 5-10% of cells survived UV irradiation. Eight cultures of different strains were used for each set of experiments, and the cultures were plated on LB ϩ rifampicin (50 g/ml) plates.
UV Sensitivity Assays-Cultures of different strains were grown in LB or M63 medium supplemented with glucose, amino acids, and vitamins. Either overnight or mid-log phase cultures were diluted 10-fold with plain M63 medium, and 1-ml aliquots of the diluted cultures were placed into 24-well cell culture plates (Costar, Corning Inc.). Cultures of different strains were tested simultaneously in different rows on each plate. To avoid differences that might be caused by their positions in the wells, cells from different cultures were randomly placed in different rows in different sets of experiments. The 24-well plate was placed either underneath a UV lamp(s) (254 nm) that was fixed on a stand or in the chamber of an UV Stratalinker 1800 (Stratagene, CA) with five 254-nm 8-watt UV lamps. Cells were irradiated with different doses of UV light by placing aluminum foil on different wells for different lengths of time. The numbers of surviving cells for each UV dose were counted after plating an appropriate dilution of the cells on LB plates and incubating them overnight at 37°C. Alternatively, a series of culture dilutions (10 Ϫ3 to 10 Ϫ6 ) were microdropped (10 l) on LB plates. Cells from different strains on one plate were exposed to a given dosage of UV irradiation as described above. Different sets of plates were exposed to different dosages of UV irradiation. The numbers of surviving cells were counted after incubation overnight at 37°C. To minimize light-induced reactivation of repair pathways, the experiments were performed with minimal background light.

RapA Forms Complexes with either Core RNAP or RNAP
Holoenzyme-Previously, we showed that RapA forms a stable complex with RNAP holoenzyme, as if it were a subunit of RNAP (10). Using the same conditions (in the presence of 0.1 M NaCl), we determined whether RapA also interacts with core RNAP. We found that a stable core RNAP-RapA complex can be reconstituted with highly purified RapA and core RNAP in vitro (Fig. 1). When the 110-kDa RapA protein and core RNAP were mixed and passed through a Superose 6 HR gel filtration column (Amersham Pharmacia Biotech), they coeluted as a complex (Fig. 1A), whereas each of the two proteins eluted in different fractions when they were run separately ( Fig. 1, C and  D). These results showed that RapA also forms a stable complex with core RNAP, just as it forms a stable complex with RNAP holoenzyme (Fig. 1B).
Because we found that RapA co-purified only with RNAP holoenzyme and not with core RNAP during our purification procedure and that the fractions containing both RapA and RNAP holoenzyme were eluted at about 0.4 M NaCl in the final step of Mono-Q chromatography (10), we again studied the interaction of RapA with core or holoenzyme by gel filtration at 0.4 M NaCl (Fig. 2). When RapA and the core RNAP were mixed and passed through the gel filtration column at 0.4 M NaCl, a significant fraction of RapA coeluted with RNAP as an RNAP-RapA complex ( Fig. 2A). The core RNAP-RapA complex was less stable at 0.4 M NaCl than at 0.1 M NaCl, as shown by the increasing dissociation of RapA from the complex, resulting in more free, unbound RapA at 0.4 M NaCl (compare Fig. 2A with Fig. 1A). However, the fraction of RapA bound to core RNAP at 0.4 M NaCl was still much greater that that bound to RNAP holoenzyme (Fig. 2B). By scanning the Coomassie Brilliant Blue R-250-stained gels, we estimated that the amount of RapA complexed with core RNAP was approximately 10 times greater than that complexed with RNAP holoenzyme.
Apparently, the RNAP-RapA complexes behaved differently in gel filtration than in Mono-Q chromatography. To further demonstrate that RapA can interact with either core or holoenzyme, we also analyzed the interaction between RapA and the two forms of RNAP by glycerol gradient ultracentrifugation ( Fig. 3). Consistent with the gel filtration results, we detected complex formation between RapA and either core or holoenzyme. Again, we found that the interaction between RapA and core RNAP was stronger than that between RapA and RNAP holoenzyme. At 0.7 M RNAP and 0.7 M RapA, while all RapA remained bound to core RNAP during the course of ultracentrifugation (Fig. 3A), the less stable RNAP holoenzyme-RapA complex showed a trace of RapA dissociating from the complex (Fig. 3B). The differences in RapA binding to core or holoenzyme were even more apparent at 0.06 M RNAP and 0.12 M RapA. At this lower concentration of proteins, while a fraction of RapA was still associated with core RNAP (Fig. 3D), almost no RapA was associated with holoenzyme ( Fig. 3E).
To determine whether binding to core RNAP can also stimulate the ATPase activity of RapA, we compared the ATPase activities of free RapA and core RNAP-RapA complexes (Fig. 4). The ATPase activity of core RNAP-RapA complex was nearly 4-fold higher than that of RapA alone (compare lane 2 with lane 5), very similar to that of RNAP holoenzyme-RapA complex (compare lane 2 with lane 6). The stimulation of RapA ATPase by binding to either core RNAP or RNAP holoenzyme further confirms that RapA interacts with both forms of RNAP. Moreover, it also provides a basis for determining the dissociation constants of RNAP-RapA complexes.
Determination of the Affinity of RapA to RNAP Holoenzyme or Core RNAP-We also determined the relative affinities of holoenzyme and core RNAP-RapA complexes by taking advantage of the stimulatory effect of RNAP on the RapA ATPase activity (Fig. 4). We used a fixed ratio (2:1) of Rap and RNAP in one set of experiments and RapA alone in a parallel set of experiments. We measured ATPase activity (A obs ) in sequentially diluted reaction mixtures. We calculated the difference (⌬A obs ) between the ATPase activities of RNAP-RapA mixtures and RapA alone at each RapA concentration and determined ⌬A max (maximal activation) as described under "Experimental Procedures." The ratio of ⌬A obs and ⌬A max was calculated and plotted as a function of RapA concentration (Fig. 5). We found that at high concentrations of proteins, the activation of RapA ATPase was relatively constant (plateau in Fig. 5). Upon sequential dilution, the activation of RapA ATPase decreased, as the RNAP-RapA complexes dissociated. Although this method cannot be used to obtain an explicit K d for the complex, we assumed that at half-maximal activation of RapA ATPase the free and bound RNAP concentrations were equal. Therefore, the K d can be approximated as the concentration of RapA at that point. The apparent K d values, calculated from this method, were 5-10 nM for RapA-core RNAP and 50 -80 nM for RapA-RNAP holoenzyme complex. These values show that the RapA protein exhibits higher affinity to core RNAP than to RNAP holoenzyme, in agreement with the gel filtration and glycerol gradient ultracentrifugation studies described above.
RapA Cross-links to the ␤Ј and ␣ Subunits of RNAP-We probed the interface between RapA and RNAP using various cross-linking agents. Cross-linking experiments with EGS- NHS, a 12-carbon atom linker-bifunctional cross-linking reagent capable of modifying amino groups, yielded positive results, although we failed to obtain cross-linked RNAP-RapA complexes using several bifunctional reagents with the linker arm of fewer than 10 carbon atoms.
To simplify the identification of new cross-linked species of the RNAP-RapA complex, we compared the patterns of the cross-linked protein products in the reactions containing RapA and RNAP with those in parallel reactions containing RNAP alone. It has been shown that RNAP subunits are capable of intramolecular cross-linking forming complex patterns (30). After cross-linking, the reaction mixtures were separated on SDS-5% polyacrylamide gels, and the gels were either silverstained or transferred onto Immnobilon P membranes for immunostaining with antibodies specific for RapA or various subunits of RNAP (Fig. 6). Fig. 6A shows the kinetics of a representative cross-linking reaction. Core RNAP treated with EGS-NHS showed a complex pattern of high molecular weight bands as a result of intramolecular cross-linking of its subunits (Fig. 6A, lanes 2-6). However, new cross-linked species appeared in the reaction containing core RNAP and RapA (Fig. 6A, lanes 8 -12). One new cross-linked product with an apparent molecular mass of 260 -FIG. 5. Determination of the apparent K d of RNAP-RapA complex. The dissociation constants of the complexes of RapA with core RNAP or RNAP holoenzyme were determined based on the activation of RapA ATPase by RNAP. Sequential dilutions of RapA alone, the RNAP holoenzyme-RapA complex (molar ratio 1:2, respectively; closed rectangles), and core RNAP-RapA complex (molar ratio 1:2, respectively; open rectangles) were made, and the dilutions were assayed for ATPase activity. The ratio of ⌬A obs and ⌬A max was calculated as described under "Experimental Procedures" and plotted as a function of RapA concentration. The apparent K d values were obtained directly from these adsorption isotherms as the RapA concentrations corresponding to halfmaximal ⌬A obs /⌬A max . Inset, the RapA amount (ng) is plotted versus pmol of ATP hydrolyzed in the sample (37°C, 45-min reactions). The specific activity of the RapA ATPase calculated from this plot was about 30 pmol of ATP hydrolyzed/min/g of RapA. 270 kDa that reacted with both ␤Ј-specific monoclonal antibodies (Fig. 6B, lane 6, arrow) and RapA-specific polyclonal antibodies (Fig. 6B, lane 2, arrow) is the RapA-␤Ј complex. Another new cross-linked product with an apparent molecular mass of about 150 kDa that reacted with both ␣-specific monoclonal antibodies and RapA-specific polyclonal antibodies (Fig. 6C,  lanes 8 and 4, respectively) is therefore the RapA-␣ complex. Because this cross-linked product co-migrated with the ␤ and ␤Ј subunits, it was not apparent by silver staining (Fig. 6A).
In addition, a few new cross-linked complexes were found in the presence of RapA. One example is the product with a molecular mass of about 125 kDa that was apparent even by silver staining (Fig. 6A, marked as RapA CL1 ). It reacted with RapA-specific polyclonal antibodies (Fig. 6B), suggesting that the cross-linked product might have an additional mass of about 15 kDa. Since our RNAP preparation contained the subunit of RNAP, which has a molecular mass of 10.1 kDa, we investigated whether this 125-kDa product was the RapAcomplex. However, this cross-linked product did not react with -specific polyclonal antibodies (data not shown), confirming that it is not the RapA-complex. Since no other small proteins (or small RNAs) were noted in RNAP or RapA preparations (judging from silver-stained SDS-12% polyacrylamide gels), it seems likely that the unidentified cross-linked species are the result of intramolecular cross-linking of RapA itself (Fig. 6A, RapA CL1 and RapA CL2 ). The same intramolecular cross-linking of RapA could explain the formation of another new RapAcontaining cross-linked complex in Fig. 6C (marked as RapA CL1 -␣).
Similarly, we treated either purified RNAP holoenzyme or the mixture containing RNAP holoenzyme and RapA with EGS-NHS in parallel cross-linking experiments (Fig. 7). The patterns of the cross-linked complexes of RNAP holoenzyme were significantly more complicated than that of core RNAP due to the presence of the -70 subunit, which is known to cross-link to multiple subunits of RNAP (30). We detected no apparent cross-linked species that reacted with both RapAspecific and -70-specific antibodies in the reactions containing RNAP holoenzyme and RapA (data not shown). However, it is interesting to note that in the presence of RapA the amount of the cross-linked ␣--70 complex (which has a molecular mass of about 130 kDa) and the amount of the cross-linked ␣-␣ complex were significantly reduced (Fig. 7, lanes 2 and 4 and lanes 6 and  8). This apparent RapA effect on the cross-linking efficiency indicates that RapA changes the configuration of holoenzyme upon its binding to RNAP.
The rapA Null Mutants Exhibited No Significant UV Sensitivity-Previously, we analyzed the effects of RapA in in vitro transcription and detected only a marginal activation of the RNAP transcriptional activity (10). To study the function of the RapA protein in the cell, we constructed a rapA null mutation by combining an internal deletion and an insertion of a cat (Cm r ) cassette in the gene.
The rapA gene appeared to be nonessential for the bacterial cell, based on the following criteria. 1) The rapA null mutation could be introduced into other cells that harbored either a plasmid expressing the wild-type rapA gene or only the vector with similar efficiency by phage P1-mediated transduction. 2) The rapA null mutant exhibited no detectable difference in growth compared with wild-type cells under a variety of conditions, such as different incubation temperatures and growth media (data not shown). Similar results were also reported in Ref. 19.
Because some eukaryotic SWI/SNF family members have been implicated in DNA repair, we explored the possibility that RapA could be important for DNA damage recovery. Originally, our preliminary results suggested that the rapA null mutant was hypersensitive to UV irradiation and the antibiotic mitomycin C (a DNA-damaging agent). 2 However, further studies determined that these phenotypes were caused by the presence of a cryptic phage with -80 immunity. Thus, the originally constructed rapA mutant was a lysogen for the phage; upon UV irradiation or mitomycin C treatment, the repressor of the cryptic phage was inactivated and the lysogenic phage was induced, and it entered the lytic phase of its cycle, resulting in increased cell death. After we constructed a phage-free rapA null mutant, we observed no UV-hypersensitive or mitomycin C-hypersensitive phenotypes compared with the wild-type isogenic strain (data not shown).
Because it was reported recently by Muzzin et al. (19) that disruption of the same E. coli gene (also called hepA) caused UV sensitivity, we carefully analyzed this phenotype further (see "Experimental Procedures"). To assure that we could detect even marginal increases in UV sensitivity in our assays, we included the mfd mutant in every UV sensitivity assay as a control. The mfd gene encodes the transcription-repair coupling factor (26). The mfd mutation has been reported to confer either no (27) or only mild increase in UV sensitivity (28) when compared with the isogenic wild-type strains. Initially, we performed the UV sensitivity assays in the JC7623 strain background that contains the recBC and sbcBC mutations and was used for UV sensitivity assays by Muzzin et al. (19). The rapA null mutant behaved somewhat similar to the isogenic wildtype strain in response to UV irradiation, whereas the mfd mutant exhibited mild UV sensitivity compared with the wildtype strain (Fig. 8). We also performed the same experiments in the MG1655 strain background that is a prototype of the wildtype K12 strain and observed a pattern similar to that seen in JC7623 (Fig. 8). At low UV doses, there was a very small difference among the wild type and mfd and rapA mutants. At high UV doses, while the mfd mutant became noticeably UV- Note that the level of RapA is higher in the exponentially growing cells than that in the overnight cultures. sensitive, the rapA mutant was only slightly more UV-sensitive than wild type strain. We also determined the UV sensitivity of the mfd rapA double mutant and found that the double mutant was only slightly more UV-sensitive than the mfd mutant (data not shown). The rapA null mutation in each strain background was confirmed, and we detected no RapA in the mutants (Fig.  8B). We also performed UV sensitivity assays using a microdrop method (see "Experimental Procedures") and detected no significant differences in UV sensitivity between the rapA null mutation and the isogenic wild type strain, whereas the mfd mutation conferred an increase in UV sensitivity compared with the wild type strain (data not shown).
To determine whether the rapA mutation affects mutation rate, we measured the occurrence of mutations conferring rifampicin resistance (Rif r ) from either wild type cells or the rapA null mutant (Fig. 9). We observed very similar spontaneous mutation rates of Rif r mutations from the two strains. With UV mutagenesis, the Rif r mutation rates were increased for the two strains. It appeared that the UV-induced Rif r mutation rate was about 2-fold lower in the rapA mutant when compared with the wild type strain. DISCUSSION We have demonstrated that RapA interacts with both core RNAP and RNAP holoenzyme in a purified system using several different methods. By binding to either form of RNAP, the RapA ATPase activity is stimulated. Under the experimental conditions we used, RapA formed complexes with either core RNAP or RNAP holoenzyme in the nanomolar range, suggesting that the interaction is highly specific. However, the affinity of RapA to core RNAP is about 5-16-fold higher than to RNAP holoenzyme. If core RNAP is indeed the main RapA-binding form of RNAP in vivo, it indicates that the function of RapA is more likely related to transcription elongation/termination than to initiation. We are currently exploring this possibility.
Previously, we found that almost all of the RapA protein molecules in the cell co-purified with RNAP holoenzyme (10). We do not know why RapA was not associated with core RNAP during the Mono Q step of the RNAP purification procedure. Several possibilities could contribute to this apparent difference: 1) some special conditions during the RNAP purification procedure; 2) special chromatographic conditions; and 3) some factors missing in a purified system. On the other hand, Muz-zin et al. (19) reported that the same protein interacted only with core RNAP but not with holoenzyme. It is possible that differences in the nature of the protein and in the conditions for the binding assays contributed to these apparently different results.
We have shown that RapA cross-links to the ␣ and ␤Ј subunits of RNAP, suggesting that it lies in the interface of these two subunits. Furthermore, it appeared that upon binding to RapA the RNAP holoenzyme alters its configuration, because the cross-linking efficiency of the ␣--70 and ␣-␣ complexes was greatly reduced in the presence of RapA, although we detected no cross-linking between RapA and the -70 subunit or ␣--70 complex. Because RapA is able to bind to RNAP holoenzyme, forming a 1:1 complex (10), it is clear that the binding sites for RapA and the -70 are distinct. The E. coli RNAP structure has been determined from a combination of x-ray crystallography and electron microscopy (31). The binding site for GreB on RNAP has been proposed (32). The binding sites for RapA on RNAP await future determination.
The rapA null mutant behaved very similar to the isogenic wild-type strain under various laboratory conditions. In the course of extended examination using two different genetic backgrounds, our experiments consistently showed that the rapA null mutant exhibited only a marginal increase in UV sensitivity compared with the wild-type strains (Fig. 8). The observed UV sensitivity of the rapA mutant was significantly less than that of the mfd mutant, which by itself showed only a mild increase in UV sensitivity compared with the wild type strain. Thus, our results were quantitatively different from the results reported by Muzzin et al. (19). This apparent difference could be due to the differences in strain backgrounds or experimental conditions used between the two laboratories. In that report, the authors also claimed that the expression level of the RapA (HepA) protein was substantially higher in the recBC sbcBC (JC7623) strain and speculated that the possible disruption of RecBCD function that is important for DNA repair pathways may be compensated by increased expression of RapA (HepA). However, we detected no substantial difference in the amount of RapA protein between the JC7623 strain and the wild-type MG1655 strain by immunoassays using antibodies against RapA (Fig. 8B). Furthermore, the RapA expression level was not increased upon DNA damage induced by mitomycin C (data not shown). We also detected little or no effect of rapA on spontaneous or UV-induced mutational rates. Based on our data, it seems unlikely that RapA plays a major role in DNA repair, at least for damage induced by UV irradiation or mitomycin C. Currently, we are investigating the regulation of rapA in the hope that it will lead to understanding of the function of this RNAP-associated protein.