Tethering of the Large Subunits of Escherichia coli RNA Polymerase*

The rpoB and rpoC genes of eubacteria and archaea, coding, respectively, for the β and β′-like subunits of DNA-dependent RNA polymerase, are organized in an operon with rpoB always precedingrpoC. Here, we show that in Escherichia colithe two genes can be fused and that the resulting 2751-amino acid β::β′ fusion polypeptide assembles into functional RNA polymerase in vivo and in vitro. The results establish that the C terminus of the β subunit and the N terminus of the β′ subunit are in close proximity to each other on the surface of the assembled RNA polymerase during all phases of the transcription cycle and also suggest that RNA polymerase assembly in vivomay occur co-translationally.

organized in an operon; the gene coding for the ␤-like subunit always precedes that coding for the ␤Ј-like subunit (6). In eubacteria, the two genes are separated by a short, untranslated linker; in archaea they overlap by several codons (7).
In this work we demonstrate that the E. coli rpoB and rpoC genes, coding, respectively, for the 1342 amino acid-long ␤ and the 1407-amino acid-long ␤Ј subunits, can be fused and that the resulting ␤::␤Ј fusion protein assembles into functional RNAP in vivo and in vitro. Furthermore, when a hexahistidine tag (His 6 tag) is inserted at the fusion site between ␤ and ␤Ј, the resulting RNAP can be immobilized on a sorbent containing Ni 2ϩ ions and is transcriptionally active in the immobilized state. On the basis of these data we conclude that the C terminus of ␤ and the N terminus of ␤Ј are within a very short distance of each other on the surface of the RNAP molecule during all phases of transcription.

EXPERIMENTAL PROCEDURES
Plasmid Construction-pRL719, which expresses the ␤::␤Ј fusion protein with a C-terminal His 6 tag from an IPTG-inducible trc promoter, was constructed by a two-step procedure. In the first step, a DNA fragment containing codons 1-941 of rpoC was PCR-amplified from pT7␤Ј (8) using the upstream primer AACTCCGACGGGAGCCTCGAG-GTGAAAGATTTATTAAAG, which places an XhoI site (underlined) encoding Leu-Glu directly before the rpoC GTG initiation codon (in bold). The amplified rpoC fragment was treated with XhoI and SalI (SalI cuts at rpoC codon 877 and is compatible with XhoI) and ligated into the XhoI site of the rpoB expression plasmid pRL706 to yield pKS1000 (Table I). pRL706 contains a modified BamHI (filled in) to SacI fragment from pRL385 (9) between the NcoI (filled in) and SacI sites of the lacI P trc plasmid pTrc99c (10). The modification places a unique XhoI site encoding Leu-Glu directly after the last codon of rpoB and before a His 6 tag, so that insertion of the XhoI-SalI fragment in pKS1000 fused the C-terminal GAG(Glu) codon of rpoB to the Nterminal GTG codon of rpoC (encodes Val in fusion), separated only by the XhoI site encoding Leu-Glu. Plasmids corresponding to pRL706 without the His 6 tag or XhoI site, but containing an SF531(Rif d 18) substitution (pRW225) or an SF531(Rif d 18) substitution and a C-terminally His 6 tagged rpoC gene from pRL663 (11) inserted between XbaI and HindIII downstream of rpoB (pRL717), also were used in our experiments (Table I).
In the second step of pRL719 construction, a BsmI-Sse8387I fragment of pKS1000 was inserted between the corresponding sites in the rpoB(Rif d 18)-rpoC(His 6 ) expression plasmid pRL717, fusing the intact rpoB and rpoC genes as described for pKS1000. pKS1010 (a derivative of pRL719 that expresses the ␤::␤Ј fusion without a His 6 tag) was constructed by replacement of the BsmI-HindIII fragment of pRL719 with the corresponding fragment of pRW308 (11). pKS1020 (which expresses the ␤::␤Ј fusion polypeptide with an internal His 6 tag at the ␤::␤Ј junction site) was constructed by treating pKS1010 with XhoI and ligating with a compatible double-stranded linker containing six histidine codons.
RNAP Purification-Cells overproducing the ␤::␤Ј fusion from plasmids were grown to mid-log phase in 1 liter of LB medium plus 1 mM IPTG and 50 g of ampicillin/ml. After recovery by centrifugation, cells were resuspended in 40 ml of 50 mM Tris-HCl, 300 mM KCl, 10 mM EDTA, pH 7.9, and lysed by passage through a French press. The lysate was cleared by low speed centrifugation, and RNAP was recovered by polyethyleneimine (PEI) precipitation, high salt extraction, and ammonium sulfate precipitation essentially as described by Burgess and Jendrisak (12). The ammonium sulfate pellet containing RNAP was redissolved in 20 ml of 20 mM Tris-HCl, 1 mM EDTA, 5% glycerol, pH 7.9 (TGE buffer), loaded on a 1-ml heparin HiTrap cartridge (Pharmacia Biotech Inc.) equilibrated in the same buffer, washed with TGE ϩ 0.3 M NaCl, and eluted with TGE ϩ 0.6 M NaCl. The eluted protein was precipitated with ammonium sulfate, dissolved in 0.5 ml of TGE ϩ 0.3 M NaCl, and passed through a Superose 6 column (Pharmacia) equilibrated in the same buffer. Fractions containing RNAP (monitored by * This work was supported in part by a Burroughs Wellcome Fund Career Award in the Biomedical Sciences (to K. S.), by grants from the Irma T. Hirschl Trust (to S. A. D.), by the Pew Foundation (Pew Scholar Award in the BIomedical Sciences (to S. A. D.), and by National Institutes of Health Grants GM53759 (to S. A. D.) and GM38660 (to R. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Affinity Labeling and in Vitro Transcription-Lys ␤1065 affinity labeling reactions on T7 A2 promoter containing DNA fragment (13) were performed essentially as described for the T7 A1 promoter-dependent labeling (13) with the following modifications. A derivatized GMP analog, specified by position ϩ1 of the T7 A2 promoter, was used to cross-link Lys ␤1065 at the first stage of the reaction. The synthesis of the GMP derivative was analogous to the AMP derivative synthesis described previously (14). To radioactively label the cross-linked subunits, [␣-32 P]CTP, specified by position ϩ2 of the T7 A2 promoter, was used. Reactions were terminated by the addition of an equal volume of Laemmli loading buffer, and proteins were resolved by SDS-PAGE in 8% precast Tris-glycine gels (Novex, San Diego, CA) followed by autoradiography.
Immobilized transcription on a T7 A1 promoter-driven DNA template was performed as described (15). Elongation complexes stalled at position ϩ20 were prepared in 50-l reactions containing 25 l of Ni 2ϩ -nitrilotriacetic acid-agarose (Qiagen, Inc.), a 20 nM amount of the 302-base pair T7 A1 promoter DNA fragment (template 3 of Nudler et al. (16)), 40 nM RNAP, 0.5 mM ApU, 50 M CTP and GTP, 2.5 M [␣-32 P]ATP (3000 Ci/mmol), 40 mM Tris-HCl, pH 7.9, 40 mM KCl, and 10 mM MgCl 2 . Reactions proceeded for 15 min at 23°C and then were washed three times with 1.5 ml of buffer (40 mM Tris-HCl, pH 7.9, 40 mM KCl, 10 mM MgCl 2 ) as described (15). To desorb transcription complexes from Ni 2ϩ beads, 2 M imidazole, pH 8.0, was added to a final concentration of 100 mM, and reactions were incubated with occasional shaking for 1 h at 4°C. The Ni 2ϩ beads were pelleted by centrifugation, and the supernatant, containing desorbed transcription complexes, was transferred into a fresh Eppendorf tube. Transcription by the ϩ20 elongation complexes was synchronously initiated by adding rNTPs to a final concentration of 10 M or 1 mM. Reactions proceeded for 5 min at 23°C and were then terminated by addition of an equal volume of loading buffer containing 6 M urea. Transcription products were analyzed by urea-polyacrylamide gel electrophoresis (7 M urea, 6% polyacrylamide), followed by autoradiography.

Generation of rpoB::rpoC Fusion Gene and Analysis of Its
Function in Vivo-The plasmid pRL719, which overproduces the ␤::␤Ј fusion protein from the IPTG-inducible trc promoter, was generated as described under "Experimental Procedures." The rpoB portion of the fused gene harbored the dominant rifampicin resistance mutation rif d 18 and allowed us to test for ␤ function of the fused protein in vivo. The growth on plates of rifampicin-sensitive E. coli RL602 cells harboring either pRL719 (p␤(rif d 18)::␤Ј), rif d 18 rpoB expression plasmid pRW225 (p␤(rif d 18)); rpoC expression plasmid pRW308 (p␤Ј), or control plasmid pBR322 was investigated (Fig. 1). After streaking on plates and overnight growth, only cells expressing rpoB::rpoC from pRL719 or rif d 18 rpoB from pRW225 grew in the presence of rifampicin; cells expressing rpoC from plasmid pRW308 as well as cells harboring pBR322 completely failed to grow in the presence of rifampicin. Growth in the presence of rifampicin was dependent on rpoB::rpoC expression, since cells did not grow in the absence of IPTG, which is needed to derepress the trc promoter of the structural gene of pRL719. Since the expression of ␤(rif d 18)::␤Ј confers rifampicin resistance to cells, we conclude that the rpoB::rpoC fusion gene is able to function as a source of the ␤ subunit in vivo.
To test for function of the ␤::␤Ј fusion in place of ␤Ј, E. coli RL602 harboring pRL719 was plated at an elevated (42°C) temperature. RL602 cells harbor an amber mutation in the chromosomal copy of rpoC that is suppressed by a temperaturesensitive suppressor (17). RL602 grows at 30°C, but not at 42°C. The experiment shown in Fig. 1C demonstrates that cells expressing either the ␤::␤Ј fusion or ␤Ј, but not ␤, grew at 42°C. In the absence of IPTG, RL602 cells harboring pRL719 completely failed to form colonies at 42°C. Since the expression of ␤(rif d 18)::␤Ј confers temperature resistance to the cells, we conclude that the rpoB::rpoC fusion gene is able to function as a source of the ␤Ј subunit in vivo. RL602 cells were also able to grow at 42°C and in the presence of rifampicin in an IPTG-dependent manner (Fig. 1D), demonstrating that the rpoB::rpoC fusion gene simultaneously functioned as a source of both ␤ and ␤Ј in vivo.
Purification of RNA Polymerase Containing the ␤::␤Ј Fusion-The results of the in vivo experiments presented above could be explained by the presence in the RL602 cells harboring pRL719 of (i) functional RNAP comprising ␣ 2 ␤::␤Ј, (ii) functional RNAP comprising ␣ 2 (␤::␤Ј) 2 , (iii) ␤ and ␤Ј resulting from proteolytic cleavage of the ␤::␤Ј fusion protein at the subunit junction site followed by assembly of "wild-type" ␤ and ␤Ј subunits into RNAP comprising ␣ 2 ␤␤Ј. To distinguish between these possibilities, RNAP from RL602 cells overproducing ␤::␤Ј was purified using our standard purification procedure (18) involving PEI precipitation and extraction of RNAP from the PEI pellet with 1 M NaCl, followed by heparin affinity chromatography (see "Experimental Procedures"). Heparin affinity chromatography fractions containing RNAP were loaded on a   (Table I). After recovery on LB ampicillin (Amp) plates at 30°C, the transformants were streaked on test plates and incubated. When added, ampicillin was present at 50 g/ml, IPTG at 1 mM, and rifampicin (Rif) at 100 g/ml. A, LB ampicillin at 30°C. B, LB ampicillin at 42°C. C, LB ampicillin IPTG at 42°C. D, LB ampicillin IPTG rifampicin at 42°C. Growth of colonies containing pRL719 (p␤(rif d 18)::␤Ј) in D indicates both halves of ␤::␤Ј fusion must be functional, since no other source of ␤Ј or Rif r -␤ is available in these cells. Superose 6 gel-filtration column attached to the fast protein liquid chromatography system, and the eluted proteins were analyzed by SDS-PAGE (Fig. 2). The ␤::␤Ј fusion protein coeluted exactly with the normal ␤, ␤Ј, and ␣ polypeptides and with RNAP activity (data not shown) during the gel-filtration step. From this experiment we conclude that the fusion protein assembled in an RNAP molecule comprising ␣ 2 ␤::␤Ј.
The C terminus of the ␤Ј portion of the fusion protein encoded by pRL719 is extended with a His 6 tag (see "Experimental Procedures"). As is shown elsewhere (19), the His 6 tag allows facile separation of RNAP containing the plasmid-borne, His 6 -tagged ␤Ј subunit from RNAP containing untagged ␤Ј of chromosomal origin by ion metal affinity chromatography (IMAC). Superose 6 fractions containing RNAP were pooled, fractionated by IMAC, and the flow-through and eluant were analyzed by SDS-PAGE (Fig. 2, lanes E and F). The flowthrough fractions contained wild-type RNAP (␣ 2 ␤␤Ј), while fractions retained on the column and eluted with 100 mM imidazole contained ␤::␤Ј, ␣, and and did not contain the normal ␤ and ␤Ј polypeptides. Thus, the fusion polypeptide assembled into RNAP did not undergo substantial proteolytic degradation into individual ␤ and ␤Ј polypeptides in the cell or during purification.
The rpoB::rpoC Fusion Gene Can Be the Only Source of ␤ and ␤Ј in the Cell-The results presented in Fig. 1D suggest that E. coli RL602 harboring pRL719 can survive even in the absence of functional ␤ and ␤Ј subunits. In our next experiment we tested this possibility directly. E. coli strain MX782 (Table I) harboring pRL719 was inoculated in rich medium containing rifampicin and incubated at 42°C in the absence of IPTG. In these conditions, the structural gene of pRL719 was not expressed, and most cells failed to grow. Cells that did grow must have substituted their rpoB ϩ and rpoC ts alleles with rif d 18 rpoB and rpoC ϩ alleles of pRL719, respectively, by homologous recombination. As a result, the rpoB::rpoC fusion was transferred to the chromosome. The recombinant clones could no longer synthesize normal ␤ and ␤Ј and could be easily selected by screening the whole-cell lysates by SDS-PAGE. On these gels, normal ␤ and ␤Ј, which are among the largest proteins in E. coli, form a characteristic double band (Fig. 3A, lane 1). Cell lysates of some of the clones that continued to grow at elevated temperature and in the presence of rifampicin no longer contained the ␤ and ␤Ј subunits (lane 3).
RNAP purified from these cells using standard procedures contained the ␤::␤Ј fusion but did not contain full-sized ␤ and/or ␤Ј (Fig. 3B, left panel). Affinity labeling with an initiating substrate analog (14) established that RNAP was active and did not contain contaminating wild-type RNAP (Fig. 3B, right  panel). We conclude that the rpoB::rpoC fusion was able to substitute for chromosomal rpoB and rpoC genes in vivo. We failed to observe any growth defect in cells harboring the chromosomal rpoB::rpoC fusion.
The C Terminus of ␤ and the N Terminus of ␤Ј Are Exposed on the Surface of RNA Polymerase-The plasmid pKS1020 encodes the ␤::␤Ј fusion with a His 6 tag positioned at the junction site of the two subunits. RNAP with an internal His 6 tag was purified from cells harboring pKS1020 and tested in an in vitro transcription experiment (Fig. 4). Immobilized RNAP was used to obtain a transcription complex containing radioactively labeled 20-meric RNA on a template containing the T7 A2 promoter and the tR2 terminator (16). One-half of the immobilized transcription reaction was treated with imidazole to desorb transcription complexes from the Ni 2ϩ beads. Unlabeled rNTPs were added to both immobilized and desorbed transcription complexes, and the pattern of transcription products was analyzed by denaturing PAGE. The experiment shown in Fig. 4 demonstrates that RNAP can be immobilized on a solid support through the internal His 6 tag and that the immobilization does not interfere with RNAP function.

DISCUSSION
Structural Implications-The principal conclusion of this work is that the C terminus of ␤ and the N terminus of ␤Ј are positioned close to each other on the surface of E. coli RNAP. Since the ␤::␤Ј fusion polypeptide can support RNAP function in vivo and in vitro with only a two-amino acid linker inserted between the ␤ and ␤Ј sequences, the C terminus of ␤ and the N terminus of ␤Ј must be positioned within about 5 angstroms of each other. In addition, since RNAP harboring the ␤::␤Ј fusion polypeptide appears to be fully functional in vivo and in vitro, the C terminus of ␤ and the N terminus of ␤Ј must remain in close proximity throughout all phases of the transcription cycle. Independent support for this conclusion comes from both genetic and biochemical studies. Yano and Nomura (20) demonstrated that a temperature-sensitive substitution in the Nterminal-most conserved segment A of the ␤Ј homolog of Saccharomyces cerevisiae RNAP I can be suppressed by a substitution near the C terminus of the ␤ homolog, suggesting that  3. A, isolation of recombinants between the plasmid-borne rpoB::rpoC fusion gene and the chromosomal rpoBC locus. The MX782 cells harboring pRL719 were plated at restrictive (42°C) temperature in the absence of IPTG and in the presence of rifampicin. Individual colonies were picked and resuspended in Laemmli loading buffer containing SDS; the proteins were resolved by SDS-PAGE on 4 -15% gradient gels and stained with Coomassie. B, affinity labeling of RNAP purified from cells shown in lanes 1 and 3 of A. Open complexes at the phage T7 A2 promoter were formed using RNAP purified using standard procedures, and the ␤ subunit Lys 1065 was cross-linked to an initiating GMP analog. The cross-linked subunits were then radioactively labeled with [␣-32 P]CTP, specified by position ϩ2 of the promoter. Reactions were loaded onto an 8% Tris-glycine polyacrylamide SDS-gel. The gel was stained with silver (left panel), and the cross-linked RNAP subunits were visualized by autoradiography (Autorad.) (right panel).
these two segments of the largest subunits interact with each other. Nudler et al. (21) investigated protein-DNA contacts at the leading edge of E. coli RNAP elongation complexes and found that the N-terminal conserved segment A of ␤Ј and the C-terminal segment I of ␤ are both cross-linked to a nucleotide derivative positioned 5 nucleotides downstream of the catalytic center in the template strand.
Evolutionary Implications-In eubacteria and archaea, the genes coding for the largest RNAP subunits (␤ and ␤Ј) are organized similarly. The eubacterial rpoB gene, coding for the ␤ subunit, always precedes rpoC, which codes for ␤Ј (6). The two genes are co-transcribed and are separated by a short, untranslated linker. Similarly, archaeal genes coding for the ␤ and ␤Ј homologs are also co-transcribed as part of an operon, and their relative position is the same as in prokaryotes (7). The linker separating the genes is either very short, or the two genes overlap. The functional significance of such an organization of the rpo genes is unknown. In contrast to prokaryotes and archaea, genes coding for the RNAP largest subunits in eukaryotes are located on different chromosomes (1). It is tempting to speculate that the organization of rpoB and rpoC in archaea and prokaryotes is due to transcription/translation coupling that occurs in these organisms in the absence of nuclei and allows RNAP assembly to occur co-translationally. Several lines of evidence support this hypothesis. The pathway of E. coli RNAP assembly in vivo and in vitro is ␣3␣ 2 3␣ 2 ␤3␣ 2 ␤␤Ј (24). Our recent data demonstrate that evolutionarily conserved segments H and the N-terminal half of segment I of the E. coli ␤ subunit are necessary and sufficient for the specific and obligatory interaction with the ␣ subunit (25). In addition, the C-terminal-most half of conserved segment I is necessary for recruitment of ␤Ј into the ␣ 2 ␤ complex (25). Data of Luo et al. (26) suggest that the N-terminal portion of ␤Ј is required for interactions with ␣ 2 ␤. Thus, ordered translation of the rpoBC mRNA is compatible with co-translational assembly of RNAP. Experiments directly testing this hypothesis are currently in progress.
Interestingly, recent genomic sequencing of Helicobacter pylori revealed that in this organism, the rpoB and rpoC genes form a continuous open reading frame (28). In addition, a cytoplasmic killer plasmid of the yeast Klyuveromyces lactis contains an open reading frame corresponding to a fusion of rpoB to the N-terminal portion of rpoC (28). While RNAP has not been purified from these organisms, our results show that such an arrangement is compatible with RNAP function, since the protein product resulting from the E. coli rpoB::rpoC fusion can assemble into functional RNAP. FIG. 4. Transcription by RNAP immobilized on a solid support through an internal His 6 tag at the ␤::␤ junction site. Elongation complexes immobilized on Ni 2ϩ -nitrilotriacetic acid-agarose and stalled at position ϩ20 of the T7 A1 transcription unit were prepared (lanes 1 and 4). One-half of the immobilized transcription reaction was treated with 100 mM imidazole, and desorbed transcription complexes were recovered from the supernatant. Transcription was resumed by adding NTPs to 1 mM. Reactions proceeded for 5 min at 23°C, and the reaction products were analyzed by denaturing 10% PAGE followed by autoradiography.