Characterization of the Rhodobacter capsulatus Housekeeping RNA Polymerase

To begin to characterize biochemically the transcriptional activation systems in photosynthetic bacteria, theRhodobacter capsulatus RNA polymerase (RNAP) that contains the ς70 factor (R. capsulatusRNAP/ς70) was purified and characterized using two classical ς70 type promoters, the bacteriophage T7A1 and the RNA I promoters. Transcription from these promoters was sensitive to rifampicin, RNase, and monoclonal antibody 2G10 (directed against the Escherichia coli ς70 subunit). Specific transcripts were detected in vitro for R. capsulatus cytochrome c 2(cycA) and fructose-inducible (fruB) promoters and genes induced in photosynthesis (puf andpuc) and bacteriochlorophyll biosynthesis (bchC). Alignment of these natural promoters activated byR. capsulatus RNAP/ς70 indicated a preference for the sequence TTGAC at the −35 region for strong in vitro transcription. To test the −35 recognition pattern, theR. capsulatus nifA1 promoter, which exhibits only three of the five consensus nucleotides at the −35 region, was mutated to four and five of the consensus nucleotides. Although the nifA1wild type promoter showed no transcription, the double mutated promoter exhibited high levels of in vitro transcription by the purified R. capsulatus RNAP/ς70 enzyme. Similarities and differences between the RNAPs and the promoters ofR. capsulatus and E. coli are discussed.

As elucidated by many in vitro studies during the last 30 years, bacterial transcription requires a core RNA polymerase (RNAP) 1 enzyme and factors that recognize specific promoter elements. RNAP has been purified from a variety of bacterial species but has been characterized most thoroughly in Escherichia coli (for review, see Ref. 1). RNAP from the photosynthetic bacterium Rhodobacter sphaeroides has been purified and shown to have a subunit composition similar to that of other Gram-negative bacteria (2). A report of a partially purified preparation of Rhodobacter capsulatus RNAP was published a decade ago (3), but from that study it was not possible to determine definitively which, if any, was the housekeeping 70 subunit nor to evaluate promoter recognition determinants. In vitro transcription of R. sphaeroides RNAP has been demonstrated from templates that contain E. coli 32 and 70 type promoters and from an R. sphaeroides rrn promoter (4). In the present report we describe in vitro studies on the transcription apparatus of R. capsulatus and promoters that are recognized by this system.
Most of the studies to date on gene regulation in R. capsulatus have concerned genetic characterization of signal transduction pathways or the in vivo analysis of mRNAs. From these investigations unique activators and repressors are theorized to regulate 70 -dependent transcription at a variety of promoters in photosynthetic bacteria. In R. capsulatus, operons involved in photosynthesis are regulated by light and oxygen (for review, see Ref. 5). For example, puc, puf, and puh encode the structural polypeptides for the photosynthetic complexes, and bch encodes bacteriochlorophyll biosynthetic enzymes. The promoters of some of these genes have been studied by deletion and in vivo primer extension analysis (e.g. 6 -8). Although bch and puc are thought to be transcribed by the R. capsulatus RNA polymerase/ 70 holoenzyme, based on promoter sequences, it is unclear what factor(s) recognize the puf and puh genes. Site-directed mutational analysis of the bchC promoter demonstrated that nucleotides typical of 70 promoters in the Ϫ10 and Ϫ35 hexamers upstream of the transcriptional start site are important for in vivo transcription (9). Light-dependent stimulation of transcription from the puf and puh operons requires the hvrA gene (10). The oxygen-regulated puf, puh, and puc operons require the regA/regB-encoded two-component system, although it is unknown whether these promoters are directly activated by such proteins (11,12). The recently discovered CrtJ is required for aerobic repression of the puf, puc, puh, and bch operons by an unknown mechanism (13). In some cases cis-DNA elements upstream of these promoters have been proposed to mediate light and oxygen regulation by binding of the regulatory proteins (e.g. 9,11). In the present study it is shown that the cytochrome c 2 gene (cycA) and the fructose-inducible gene fruB (14) may also be activated by the R. capsulatus RNAP/ 70 . Other signal transduction pathways in R. capsulatus elucidated mainly by genetic studies include nitrogen sensing (for review, see Refs. 15 and 16) and the FnrL regulon. 2 As an important step toward in vitro reconstitution of these signal transduction pathways, the R. capsulatus housekeeping RNAP holoenzyme was characterized. 70 -Dependent tran-scription from a variety of promoters using linear or supercoiled templates was demonstrated. A consensus for optimal 70 type promoters in this bacterium was tested further by mutating the nitrogen-regulated nifA1 promoter (18) toward the consensus and engineering each mutant promoter upstream of a transcriptional terminator on a supercoiled plasmid. Results of in vitro transcription studies on these mutant promoters confirmed the importance of specific Ϫ35 recognition elements for the holoenzyme.

EXPERIMENTAL PROCEDURES
Strains and rpoB Plasmids-The bacterial strains and plasmids used in this study are shown in Table I. pUC:SBrif and pUC:B10rif contain the sequences that encode the rifampicin binding domain of rpoB strains SB1003 (Rif r ) and B10 (Rif s ), respectively. The plasmids were constructed by polymerase chain reaction (PCR) of chromosomal DNA of the two strains using primers described in Table II. The R. capsulatus rpoB sequence for design of primers was provided by Dr. Robert Haselkorn (University of Chicago). The PCR products were digested with BamHI and PstI and cloned into pUC118. The rpoB RNAP ␤ subunit fragments were sequenced using the Sequenase enzyme according to company protocols (Amersham Corp.).
Plasmids for Supercoiled Templates-Linear templates for transcription reactions were created by PCR using the primers indicated in Table  II. For supercoiled templates, all DNA fragments that contained putative promoters were cloned into pUC118 directly upstream of a strong transcriptional terminator. The fruB promoter was cloned by PCR of SB1003 chromosomal DNA using primers described in Table II, based  on Ref. 14. The 300-bp fruB promoter fragment was digested with KpnI and EcoRI and cloned into pUC118 to create pUC:fruB. Plasmid pUC:T7 that contains 150 bp of the T7A1 promoter was made by excision of the T7A1 promoter fragment with BamHI and EcoRI from plasmid pCL185 (19) and ligation into pUC118. pUCT that contains 150 bps of the T4 bacteriophage gene 32 -independent transcriptional terminator was made by PCR of plasmid pHP45 that contained the terminator (20) using the primers described in Table II. The PCR product was digested with PstI and HindIII and cloned into pUC118. Templates that contained 70 -dependent promoters (pucC, pufQ, puhA, bchC) were created by PCR of each promoter fragment using a 5Ј-(upstream) primer that contained a PstI site and a 3Ј-(downstream) primer that contained a BamHI restriction site (see Table II). The PCR products were digested with BamHI and PstI and cloned into pUC118 to create pUC:pucC, pufQ, puhA, bchC, nifA1.
The transcriptional terminator was cloned downstream of each of the promoters by excision of the 125-bp terminator from pUCT with PstI and HindIII and into pUC:pucC, pufQ, bchC, to create the supercoiled templates pUCT:pucC, pufA, bchC. The pUCT:fruB was made by excision of the fruB promoter region from pUC:fruB with EcoRI and PstI and ligation into pUCT. The pUCT:nifA1 template was created by excision of the nifA1 promoter from pA1-P16 with SalI and PstI followed by ligation into pUCT. Templates were confirmed by restriction and sequence analysis and purified in CsCl gradients for all reactions.
The nifA1 promoter mutants A1Mut1, A1Mut2, and A1Mut3 were generated by PCR of plasmid pPA1-P16 using the primers described in Table II. The PCR products were digested with MluI and PstI and cloned into pUCT:nifA1 to create pA1M1, pA1M2, pA1M3. The mutations were confirmed by sequence analysis, and the plasmids were purified in CsCl gradients for in vitro transcription reactions.
RNA Polymerase Purification, Properties, and Stability-R. capsulatus RNAP was purified from 12 liters of R. capsulatus cells (strain SB1003) grown aerobically to mid exponential phase (ϳ17 h, A 600 2.0) in RCV medium (21) at 34°C. These conditions, although aerobic, still allowed the synthesis of some photosynthetic pigments, albeit at lower levels than fully anaerobic, light-grown cells. The rifampicin-sensitive strain R. capsulatus B10 was used for some experiments, where noted. Otherwise, R. capsulatus RNAP refers to the enzyme from SB1003. Cells were harvested by centrifugation and stored as a cell pellet at Ϫ80°C. Purification was based on procedures described previously (22,23) with modifications. Cell lysis was performed on ice, and the purification was at 4°C unless otherwise noted. Cells (ϳ200 g, wet weight) were lysed by resuspension in 108 ml of sucrose solution (10 mM Tris-HCl, pH 8, 25% sucrose, 100 mM NaCl) for 15 min, followed by the addition of 24 ml of lysozyme solution (300 mM Tris-HCl, pH 8, 100 mM EDTA, and 4 mg/ml lysozyme) for 5 min and addition of 135 ml of lysis solution (1 M NaCl, 20 mM EDTA, 110 mg of sodium deoxycholic acid). Lysis was allowed to proceed for 10 min at 10°C. Following cell lysis, the RNA polymerase-nucleic acid complexes were precipitated by the addition of 380 ml of PEG solution (17% polyethylene glycol 8000 (PEG, Sigma), 157 mM NaCl, 1 mM DTT), followed by centrifugation at 7,000 rpm for 10 min in a Sorvall centrifuge. The supernatant was removed, and proteins were eluted from the PEG pellet by the addition of 60 ml of high salt solution (10 mM Tris-HCl, pH 8, 5% PEG, 2 M NaCl, 1 mM DTT).
The PEG supernatant containing the R. capsulatus RNAP was diluted in 430 ml of column buffer (10 mM Tris-HCl, pH 8, 10 mM MgCl 2 , 1 mM EDTA, 1 mM DTT, and 7.5% glycerol) to 300 mM NaCl and loaded onto 5 ϫ 4-ml heparin-agarose (Sigma) columns that were run at ϳ0.5 ml/min using a peristaltic pump. The columns were washed with 20 column volumes (320 ml total) of column buffer that contained 300 mM NaCl, and RNA polymerase was eluted with 48 ml of column buffer in 450 mM NaCl. Fractions that contained peak RNA polymerase activity (ϳ5 ml; see below) were diluted to 135 ml in column buffer to 150 mM NaCl and loaded onto 3 ϫ 4-ml DEAE-Sepharose (Sigma) columns run at 0.6 ml/min. The columns were washed with 20 column volumes (320 ml total) of column buffer in 150 mM NaCl, and the RNA polymerase was eluted off of the column in 80 ml of column buffer in 300 mM NaCl.
Fractions that contained the peak RNAP activity (ϳ9 mls) were ammonium sulfate precipitated and stored at 2 mg/ml in 30% glycerol at Ϫ80°C. The E. coli RNAP was purified using the above procedure from 2 liters of cells (ϳ10 g wet weight; strain TB1) which were grown aerobically for 17 h in Luria broth (24) at 37°C. RNA polymerase from the Rif s R. capsulatus strain B10 (2 liters grown to mid exponential phase at 34°C) was purified by PEG precipitation and heparin-agarose chromatography as described above. Proteins were quantitated by BCA assays (Pierce) and SDS-PAGE analysis using serial dilutions of 2 mg/ml bovine serum albumin as a control. The E. coli RNAP HPLCpurified RNAP/70 holoenzyme was provided by Dr. Robert Landick (University of Wisconsin, Madison). In vitro transcription assays using the T7A1 promoter were used to determine the fractions that contained RNAP/ 70 holoenzyme of the highest specific activity, which were used for subsequent experiments. Under these conditions, R. capsulatus RNAP was stable for in vitro transcription assays for at least 1 year with less than a 3-fold loss of activity. Dialysis was not performed as it resulted in a dramatic loss of holoenzyme activity and caused precipitation of R. capsulatus RNAP when at high concentrations (2 mg/ml). Separation of the core R. capsulatus RNAP from the R. capsulatus RNAP/ 70 was attempted with a Bio-Rex 70 column (Bio-Rad), but this caused dissociation of the R. capsulatus RNAP subunits, whereas the E. coli RNAP core was stable under these same conditions. We do not understand the basis for this difference. For R. capsulatus RNAP/ 70 , initial elution fractions from the heparin-agarose column were enriched for 70 activity and later fractions for core RNAP, as assayed by in vitro transcription reactions and by SDS-PAGE. Sonication of R. capsulatus cells followed by the above PEG and heparin-agarose purification also resulted in purification of core-enriched R. capsulatus RNAP and a loss of the 70 subunit, as assayed by SDS-PAGE, Western analysis, and in vitro transcription assays. The heparin-agarose-purified E. coli RNAP contained a contaminating nuclease that was not present in R. capsulatus RNAP preparations, and therefore the DEAE-purified E. coli RNAP was always used, whereas the R. capsulatus RNAP heparin-agarose-purified fraction could be used for specific experiments, when noted.
Nonspecific Transcription Assays-Samples that contained RNA polymerase were assayed for activity in reactions (20 l) which contained 25 mM Tris-HCl, pH 8; 10 mM MgCl 2 ; 1 mM EDTA; 0.25 mM ATP, CTP, GTP, and UTP, (fast protein liquid chromatography grade, Pharmacia Biotech Inc.); 0.8 M [ 3 H]UTP (NEN Life Science Products); 1 mM DTT; 150 mM NaCl; 250 g/ml salmon sperm DNA; 250 g/ml bovine serum albumin; 3% glycerol; and 1 mM K 2 HPO 4 (to inhibit polynucleotide phosphorylase). For nonspecific templates, salmon sperm DNA was transcribed approximately 2-fold more efficiently than poly(dA-dT) DNA and so was used for these assays. Samples that contained RNAP (2.5 l) were added to initiate the reactions, which were incubated at 30°C for 15 min and terminated by spotting onto DE81 DEAE-cellulose filters (Whatman). The filters were washed five times in 5% sodium phosphate buffer, washed twice in water, and once in 95% ethanol to remove unincorporated label. Filters were counted in a Beckman liquid scintillation counter. One unit of RNA polymerase activity corresponds to 1 nmol of UTP incorporation in 15 min at 30°C. The cell lysis solution, PEG precipitation fractions, column flow-through, wash, and elution fractions for the heparin-agarose and DEAE-Sepharose columns were assayed for RNAP activity to determine the most active fractions.
Specific in Vitro Transcription Assays-In vitro transcription reactions in 16 l total (e.g. Ref. 25) were performed in transcription buffer (50 mM Tris-HCl, pH 8, 100 mM potassium acetate, 8 mM MgCl 2 , 1 mM DTT, 3.5% PEG, and 1 l of RNasin (Promega). The RNAP (40 nM) and DNA (approximately 40 nM for supercoiled templates and either 600 nM for linear templates or 3 nM for the T7A1 template) were incubated for 10 min at 24°C in transcription buffer before the simultaneous addition of heparin (50 g/ml) and nucleotide triphosphates (0.4 mM ATP, GTP, and UTP; 0.01 mM CTP; 0.5 l of 25 Ci [␣-32 P]CTP, NEN Life Science Products). The reactions were incubated for 30 min at 24°C, which was determined to be the optimal temperature for R. capsulatus RNAP. The reactions were terminated by the addition of 7 l of stop solution (80% urea, 270 mM Tris-HCl, pH 8, 270 mM boric acid, 6 mM EDTA, and 0.1% bromphenol blue), heated to 70°C for 5 min, and loaded onto an 8% acrylamide sequencing gel. Radiolabeled DNA size markers were prepared as described (18).
Other Methods-Western analysis was performed using peroxidase detection reagents from Pierce. Monoclonal antibody 2G10 was kindly provided by Dr. Richard Burgess and Nancy Thompson (University of Wisconsin, Madison). Antibodies to the R. capsulatus RNAP were generated by immunization of New Zealand White rabbits with the holoenzyme of greater than 95% purity. In vivo primer extension analysis was performed as described (26). The primer 5Ј-TGTTGAATTCTTTTTCGC-CCTTCGCGGCGT-3Ј was used for the reverse transcription reaction for the cycA gene.

RESULTS AND DISCUSSION
Purification of the R. capsulatus RNA Polymerase/ 70 Holoenzyme-The R. capsulatus RNAP was purified to characterize 70 type promoters and to study activator-and repressorregulated transcription in R. capsulatus. The R. capsulatus RNAP protein was purified by PEG precipitation followed by heparin-agarose and DEAE-Sepharose chromatography to at least 95% homogeneity by SDS-PAGE analysis (Fig. 1). The purification was assayed by nonspecific transcription assays (see below) and SDS-PAGE. Analysis of the R. capsulatus RNAP by SDS-PAGE showed subunits that by molecular weight correspond to an ␣, ␤, ␤Ј, and , as well as the 90-kDa major (housekeeping) 70 (Fig. 1, lane 6). The E. coli RNA polymerase (E. coli RNAP, strain TB1) was purified using the same procedure (Fig. 1, see lanes 1-3), which was compared with the HPLC-purified E. coli RNAP/ 70 holoenzyme (Fig. 1,  lane 7). A doublet at approximately 150 kDa was confirmed by SDS-PAGE analysis of underloaded DEAE-Sepharose pure R. capsulatus RNAP fraction; moreover, a Bio-Rex 70 column was also able to separate these two 150-kDa bands (data not shown). The ratio of intensity of the R. capsulatus RNAP ␣ polypeptides to the ␤ and ␤Ј bands correspond to 2:1:1 stoichiometry, similar to the E. coli RNAP (Fig. 1, compare lane 3 with  lane 6). An subunit occasionally purifies with the core E. coli RNAP (27) but has no known function in vivo (28). The polypeptide observed by SDS-PAGE at approximately 15 kDa may be the R. capsulatus subunit homolog, but this has not been investigated further. The approximately 90-kDa polypeptide observed by SDS-PAGE in the R. capsulatus RNAP DEAE-Sepharose fraction (Fig. 1, lane 6) was shown to be the major (housekeeping) R. capsulatus 70 factor. Monoclonal antibody 2G10 is specific for a 15-amino acid epitope in region 3.1 of E. coli 70 (29). 2G10 also cross-reacts with the major factor from a variety of bacterial species, including R. sphaeroides (4). Western analysis demonstrated that 2G10 cross-reacts with the 90-kDa subunit observed by SDS-PAGE analysis of the R. capsulatus RNAP purification fractions (not shown).
Initially, transcriptional activities of R. capsulatus RNAP and E. coli RNAP enzymes at sequential purification steps were measured by a nonspecific transcription assay that determines the accumulation of RNA product using salmon sperm DNA as template and radiolabeled UTP (Table III). The R. capsulatus RNAP purification resulted in a 140-fold enrichment (40% yield) of R. capsulatus RNAP and a total of 1.8 mg of greater than 95% pure protein from 12 liters of cells. The E. coli RNAP purification resulted in a similar enrichment (150fold) and a higher yield (75%), which was comparable to results published previously (e.g. 26). Nonspecific transcription activity of R. capsulatus RNAP (as well as the E. coli RNAP) was dependent upon MgCl 2 , DNA template, and all four nucleotide triphosphates (not shown).
Characterization of Rifampicin-sensitive and -resistant R. capsulatus RNAPs-The antibiotic rifampicin specifically inhibits transcription of bacterial RNA polymerase by binding to the ␤ subunit (30, 31) of RNAP and preventing elongation of the nascent RNA chain past a few nucleotides. The traditional R. capsulatus strain that is often used for genetic studies, called SB1003, is a rifampicin-resistant (Rif r ) derivative of strain B10 (32). To confirm that the nonspecific transcription activity was caused by transcription by R. capsulatus RNAP, we character-ized the RNAP from the Rif s R. capsulatus strain B10 and Rif r SB1003. The RNAP from R. capsulatus strain B10 (B10RNAP) was purified by PEG precipitation and heparin-agarose chromatography. RNAP from strain B10 and E. coli had the same MgCl 2 , nucleotide, and DNA template requirements as that from SB1003, but nucleotide incorporation was not observed in the presence of rifampicin for B10RNAP or E. coli RNAP (not shown). Greater than 95% inhibition was observed at 0.5 g/ml rifampicin for B10RNAP, and partial inhibition was observed at 0.05 g/ml, similar to observed values and previously reported values for E. coli RNAP (30). The R. capsulatus RNAP (from SB1003) was resistant to a high level of rifampicin (50 g/ml) similar to RNAP from a particular class of Rif r E. coli mutants (30).
Rif r is typically conferred by changes in the ␤ subunit between amino acids Ala-501 and Arg-687 (30), with the notable exception of Val-146 changes in the NH 2 terminus of the ␤ subunit (31). To understand further the differences between the enzymes from R. capsulatus strains SB1003 and B10, we cloned and sequenced the rpoB ␤ subunit DNA that encodes the rifampicin binding domain from these strains. A single nucleotide change (A 3 T) in the SB1003 sequence was discovered which corresponds to a glutamine to leucine (Q532L) amino acid change. The glutamine residue is completely conserved in all known bacterial RNA polymerases, and the Q532L amino acid substitution also confers Rif r to RNAP in E. coli at the homologous residue Q513L (30).

R. capsulatus RNAP/ 70 Holoenzyme Activates Transcription of Strong E. coli 70 Promoters in Vitro-
To characterize the biochemical activity and specificity of the R. capsulatus RNAP/ 70 holoenzyme we used the bacteriophage T7A1 promoter in an in vitro transcription assay (Fig. 2). The T7A1 promoter is a 70 -dependent promoter that contains sequences in the Ϫ35 and Ϫ10 hexamers of the DNA which are characteristic of efficient 70 promoters in E. coli (33). RNAP/ 70 from a variety of bacterial species utilizes this promoter efficiently (e.g. Ref. 33). A linear template containing the T7A1 promoter was used in an in vitro transcription reaction with the E. coli RNAP/ 70 holoenzyme, and as expected, a 141-nucleotide product was observed (e.g. see Fig. 2A). The purified R. capsulatus RNAP/ 70 also produced a 141-nucleotide product from this template. The 141-nucleotide band was observed only in the presence of all four nucleotide triphosphates and MgCl 2 ; 10 mM EDTA completely inhibited formation of the product (data not shown). Potassium acetate was required at concentrations of at least 100 mM, and 100 mM KCl could not replace the potassium acetate requirement. The 141-nucleotide transcript was degraded by RNase but not DNase (data not shown). Rifampicin completely inhibited production of the 141-nucleotide transcript when the Rif s R. capsulatus RNAP from strain B10 was used or the E. coli RNAP (Fig. 2A). However, the Rif r R. capsulatus RNAP from strain SB1003 produced the transcript in  the presence of rifampicin ( Fig. 2A). Monoclonal antibody 2G10 has been shown to inhibit transcription of the E. coli RNAP/ 70 holoenzyme in vitro (34). The 2G10 antibody also inhibited R. capsulatus RNAP production of the 141-nucleotide product although a 230-nucleotide readthrough transcript was unaffected by the antibodies (e.g. Fig. 2B, lane 5; see below). Occasionally, larger transcripts of varying intensity were observed using the T7A1 linear template and other templates (Fig. 2B). In particular, a 230-nucleotide transcript, of approximately the same length as the template, appeared to be relatively more intense when using R. capsulatus RNAP preparations that have a higher core/ 70 ratio. ("Experimental Procedures" describes the purification of such core-enriched preparations from R. capsulatus). To understand more fully the basis for these products we investigated their synthesis. Synthesis of the 230-nucleotide transcript (designated readthrough in Fig. 2B) was sensitive to rifampicin and RNase (data not shown) but not to the monoclonal antibody 2G10 (Fig. 2B, lane  5) or low levels of potassium acetate (Fig. 2B, lane 3). We conclude that this transcript is the product of end-to-end readthrough of the T7A1 230-nucleotide template by the core R. capsulatus RNAP (which was also observed for E. coli RNAP; Fig. 2B, lanes 7-9). The 371-, 601-, and 831-nucleotide transcripts (designated transposition in Fig. 2B) were sensitive to low levels of potassium acetate, rifampicin, RNase, and to the monoclonal antibody 2G10 (Fig. 2B, lanes 3 and 5, respectively) and thus are probably the result of initiation from the T7A1 promoter and transposition to a neighboring template, as has been characterized with the T7 RNAP (35) and E. coli RNAP (e.g. Ref. 36).
The R. capsulatus RNAP/ 70 holoenzyme was characterized using a supercoiled template that contained a 70 promoter. The RNA I promoter (37) from the ColEI plasmids pBR322 and pUC derivatives, also called the replication inhibitor promoter, is a strong 70 -dependent promoter that utilizes the E. coli RNAP/ 70 in vitro and in vivo, producing a characteristic 108nucleotide transcript that inhibits plasmid replication in vivo (38). The R. capsulatus RNAP/ 70 holoenzyme also produces a 108-nucleotide product in vitro from pUC118 and other pBR322-based plasmids (e.g. Fig. 3A). This transcription is sensitive to the 70 monoclonal antibodies 2G10, rifampicin, and is optimal at 100 mM potassium acetate, although the salt requirement was not as stringent as for the T7A1 promoter (not shown). This promoter serves as an internal control for all of the R. capsulatus promoter regions that have been engineered into pUC118.
R. capsulatus RNAP/ 70 Utilizes R. capsulatus Promoters in Vitro-A number of regulated promoters in R. capsulatus have been analyzed with respect to their in vivo transcription start sites. The fruB gene has been shown by lacZ fusion studies to be induced by the addition of fructose in R. capsulatus (14). Unfortunately, delineation of the in vivo transcription start site for fruB has not been reported. We have attempted to determine the fruB start site by primer extension analysis, but at least five products of similar intensity result. Thus, the fruB start site and promoter shown in Fig. 4 are based on the in vitro site determined in the present study (see below) and must be considered tentative. All other promoters shown in Fig. 4, including that for cycA, are based on the in vivo transcription start sites and correlate with the transcripts observed in vitro, as shown below. (We have determined the in vivo cycA transcription start site, as depicted in Fig. 4. 3

)
To characterize R. capsulatus promoters using the R. capsulatus RNAP/ 70 holoenzyme we initially used PCR to create linear templates that contain various R. capsulatus promoters predicted either to use 70 or unknown factors. Using the purified R. capsulatus RNAP and linear templates that contained the photosynthetic promoter, pucC (Fig. 3B, lane 1) and the cytochrome c 2 cycA (lane 3) promoter, runoff transcripts were produced in vitro. A 102-nucleotide product was observed in reactions that contained the pucC promoter template, in addition to a probable readthrough transcript at 180 nucleotides (Fig. 3B, lane 1). A 165-nucleotide band was observed in the presence of the cycA promoter template (lane 3). Interestingly, neither the pucC (Fig. 3B, lane 2) nor the cycA promoter (lane 4) was transcribed by E. coli RNAP/ 70 , although a readthrough transcript from the pucC template was observed.
Specific transcription products from templates containing other promoters (fruB, bchC, and pufQ) were not detected using linear templates when analyzed with E. coli RNAP or R. capsulatus RNAP. Thus, most promoters were analyzed using supercoiled templates that contained the T4 bacteriophage gene 32 transcriptional terminator engineered downstream. The exact site of termination in vitro from the gene 32 terminator is known for E. coli RNAP, and thus the sizes of RNA 3 P. J. Cullen, C. K. Kaufman, W. C. Bowman, and R. G. Kranz, unpublished data. products initiating from a predicted promoter could be calculated. Templates that contained the pufQ, fruB, and bchC promoters gave major specific transcripts of approximately 290, 280, and 145 nucleotides, respectively (Fig. 3A). Each of the transcripts was similar to the level of intensity of the RNA I transcript (108 nucleotides) and was not present in control plasmids that did not contain the promoter insertions (Fig. 3A). The E. coli RNAP also produced a transcript of the same size as R. capsulatus RNAP from each of the supercoiled templates in vitro (data not shown). Synthesis of the pufQ, fruB, and bchC products was sensitive to rifampicin and to the 70 monoclonal antibodies 2G10 (e.g. Fig. 3C for fruB promoter). However, variability for each of these promoters was observed in the potassium acetate requirement. Lower levels of potassium acetate (10 mM) resulted in increased levels of the pufQ product, did not affect the bchC promoter, and abolished transcription from the fruB promoter (e.g. Fig. 3C, lane 7). Minimal transcription was observed when potassium acetate was replaced with KCl (at 100 mM, see Fig. 3C, lane 8).
The promoters activated by R. capsulatus RNAP/ 70 in vitro were aligned and compared with the consensus 70 recognition sequence of E. coli (Fig. 4). With one significant exception (i.e. pufQ) the comparison suggests that homology in the Ϫ35 region (specifically TTGAC) is important for optimal recognition of the promoter by the R. capsulatus 70 subunit. The region 4.2 of E. coli 70 is responsible for recognition of the Ϫ35 hexamer, and specific amino acids Arg-584 and Arg-588 make basespecific contacts with the nucleotides within the Ϫ35 region (39,40). The rpoD gene encoding the R. capsulatus 70 protein has recently been sequenced by Haselkorn and colleagues (GenBank accession number U28162). Sequence comparison between the E. coli and R. capsulatus 70 shows that Arg-584 and Arg-588 residues are conserved as are the 5 upstream and 10 downstream amino acids. Nucleotides within the conserved Ϫ10 hexamer show less homology among strong R. capsulatus 70 promoters (Fig. 4A). The region 2.4 of E. coli 70 makes base-specific contacts at Thr-440 and Gln-437 with the Ϫ10 hexamer (41), and this region is also completely conserved with the R. capsulatus 70 . The A-T-rich properties of the Ϫ10 region probably aid in open complex formation at this region, but conclusions about exact contacts with 70 cannot be made.
It is surprising that the pufQ promoter appears to be utilized by the 70 holoenzyme in vitro since this promoter is atypical in the Ϫ35 region. Based on this poor Ϫ35 homology it has been proposed that a factor other than the housekeeping subunit may be required for transcription of pufQ (42). However, no genetic evidence has yet been generated to suggest such an alternative factor, as discussed previously (42). We could not find other sequences near the puf promoter which are similar to the Ϫ35 consensus which might yield this in vitro transcript, suggesting that an aberrant start site is not the reason for the observed product. It may be important that all other promoters were optimal at 100 mM potassium acetate, whereas preliminary studies indicate that the puf promoter appears to function at lower salt concentrations. Nevertheless, we cannot rule out the possibility that another promoter in the pufQ fragment is responsible for this transcription product. On the other hand, the consensus Ϫ35 hexamer may not always reflect the most important determinant for recognition/transcription.  . cycA and pucC). To determine more carefully the optimal Ϫ35 recognition elements for R. capsulatus RNAP, the R. capsulatus nifA1 promoter was used. The natural wild type nifA1 promoter has only 3 of the TTGAC nucleotides in the Ϫ35 region (Fig. 4B). Two distinct mutations were generated to improve this to 4 consensus nucleotide matches, called the nifA1mut1 and nifA1mut2 promoters. A third, called the nifA1mut3 promoter, contained both mutations and has the complete Ϫ35 consensus TTGAC sequence. Each of these promoters was cloned in front of the transcription terminator, and supercoiled templates were used for in vitro transcription reactions with the R. capsulatus RNAP/ 70 holoenzyme (Fig. 5). Although the pUC118 alone (lane 1) or the wild type nifA1 (lane 2) template yields no transcripts other than RNA I (108 nucleotides), the nifA1mut3 template yields an intense 97-nucleotide transcript (lane 5). Levels of this transcript are at least 100-fold higher than could be detected with the other templates, it is of the exact size predicted for this promoter and terminator, and it is more intense than the RNA I transcript. We have also observed a modest level of the 97nucleotide transcript for mut1 (lane 3) but none for mut2 (lane 4). These results support the contention that a major recognition determinant for R. capsulatus RNAP/ 70 is TTGAC at the Ϫ35 region.

CONCLUSIONS
The R. capsulatus RNA polymerase and associated 70 were purified to homogeneity and used to characterize 70 -dependent promoters. The R. capsulatus RNAP exhibits structural similarity to the E. coli RNAP with regard to subunit sizes and immunological criteria, including the 70 subunit. The molecular basis for rifampicin resistance in R. capsulatus strain SB1003 was determined, and the R. capsulatus RNAP enzymes with and without the ␤ subunit Rif r change were characterized. The purified R. capsulatus RNAP/ 70 enzyme transcribed from the T7A1, RNA I, photosynthetic (puc, puf), fructose-inducible (fru), bacteriochlorophyll (bchC), and cytochrome c (cycA) promoters. The bchC, puf, and fru promoters were only recognized in the context of supercoiled templates, and it is clear that each promoter is affected distinctly by salt concentrations. Interestingly, the pucC and cycA promoters were recognized on linear templates by R. capsulatus RNAP/ 70 but not by the E. coli RNAP/ 70 . In this context the R. sphaeroides rrnB promoter was also recognized only by the R. sphaeroides RNAP and not E. coli RNAP (4). These results suggest that differences do exist in promoter recognition and/or transcription initiation between photosynthetic and enteric holoenzymes. The molecular basis for these differences, with some possibilities discussed by Karls et al. (4) remains to be determined. It may not be surprising that organisms like Rhodobacter, with a genomic composition  of 65% GC, have evolved some recognition/melting differences from organisms with considerably higher AT content. On the other hand, strong E. coli promoters (e.g. T7A1, RNA I) function almost as efficiently using the R. capsulatus RNAP, indicating clearly analogous ideal promoter elements. An ideal promoter in the Ϫ35 region was determined for the R. capsulatus RNAP/ 70 holoenzyme which agrees with the Ϫ35 sequence determined in vivo using site-directed mutagenesis of the R. capsulatus bchC promoter (9). In the present study, the importance of the Ϫ35 recognition element, TTGAC, was confirmed by site-directed mutagenesis of the nifA1 promoter; a dramatic increase of in vitro transcription was observed with only two changes to the consensus TTGAC pattern. The investigation of activator-and repressor-dependent transcription in vitro in R. capsulatus will be an important next step to a more complete understanding of signal transduction and gene expression in photosynthetic bacteria.