The Largest Subunits of RNA Polymerase from Gastric Helicobacters Are Tethered*

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. The genome sequence of the gastric pathogenHelicobacter pylori (strain 26695) revealed homologs of two genes in one continuous open reading frame that potentially could encode one 2890-amino acid-long β-β′ fusion protein. Here, we show that this open reading frame does in fact encode a fused β-β′ polypeptide. In addition, we establish by DNA sequencing thatrpoB and rpoC are also fused in each of four other unrelated strains of H. pylori, as well as inHelicobacter felis, another member of the same genus. In contrast, the rpoB and rpoC genes are separate in two members of the related genus Campylobacter(Campylobacter jejuni and Campylobacter fetus) and encode separate RNA polymerase subunits. TheCampylobacter genes are also unusual in overlapping one another rather than being separated by a spacer as in other Gram-negative bacteria. We propose that the unique organization ofrpoB and rpoC in H. pylori may contribute to its ability to colonize the human gastric mucosa.

The rpoB and rpoC genes of eubacteria and archaea, coding respectively for the ␤and ␤-like subunits of DNA-dependent RNA polymerase, are organized in an operon with rpoB always preceding rpoC. The genome sequence of the gastric pathogen Helicobacter pylori (strain 26695) revealed homologs of two genes in one continuous open reading frame that potentially could encode one 2890-amino acid-long ␤-␤ fusion protein.
Here, we show that this open reading frame does in fact encode a fused ␤-␤ polypeptide. In addition, we establish by DNA sequencing that rpoB and rpoC are also fused in each of four other unrelated strains of H. pylori, as well as in Helicobacter felis, another member of the same genus. In contrast, the rpoB and rpoC genes are separate in two members of the related genus Campylobacter (Campylobacter jejuni and Campylobacter fetus) and encode separate RNA polymerase subunits. The Campylobacter genes are also unusual in overlapping one another rather than being separated by a spacer as in other Gramnegative bacteria. We propose that the unique organization of rpoB and rpoC in H. pylori may contribute to its ability to colonize the human gastric mucosa.
DNA-dependent RNA polymerase (RNAP) 1 is the central enzyme of gene expression and a major target for regulation. RNAPs are large, multisubunit protein complexes. The best studied RNAP, from Escherichia coli (Ͼ400 kDa), contains four core polypeptides: ␤Ј (155 kDa), ␤ (150 kDa), a dimer of ␣ (37 kDa), and one of several possible (specificity) subunits. RNAPs from other bacteria have similar subunit composition and exhibit striking and co-linear sequence similarities with the E. coli enzyme (1). The two largest RNAP core subunits comprise 60% of the RNAP mass and appear to be responsible for most of the functions of the enzyme.
The synthesis of RNAP subunits is coordinately regulated (2), but the exact mechanisms at play are unknown. In most eubacteria and archaea, genes encoding the ␤and ␤Ј-like subunits are organized in an operon with the gene for the ␤-like subunit (rpoB) always preceding that for the ␤Ј-like subunit (rpoC) (3,4). The two genes are separated by a short, untranslated linker (3,5,6), 2 whereas in archaea they overlap by several codons (4).
The genome sequence of the gastric pathogen Helicobacter pylori (strain 26695) revealed one continuous open reading frame containing the homologs of rpoB and rpoC, potentially encoding one fused 2890-amino acid-long ␤-␤Ј polypeptide (8). Our previous analysis using E. coli RNAP showed that such a ␤-␤Ј fusion is compatible with RNAP function: (i) the product of artificially fused rpoB and rpoC genes of E. coli could assemble into a functional RNAP in vivo and in vitro and (ii) an E. coli strain containing the fused rpoBC gene as its only source for RNAP was viable and contained RNAP of the expected (␤-␤Ј)␣ 2 subunit composition (9). This tethering of E. coli ␤ and ␤Ј increased the efficiency of RNAP assembly in vitro and suppressed an rpoC ts assembly mutation in vivo. 3 It thus seemed that natural tethering could be advantageous for an organism like H. pylori that needs to colonize the stomach, an intrinsically acid-rich and putatively hostile environment (10). However, RNAP had never been purified from H. pylori, and therefore post-translational proteolysis of fused ␤-␤Ј protein followed by assembly into "normal" RNAP with a ␤␤Ј␣ 2 subunit composition could not be ruled out a priori.
H. pylori belongs to the ⑀ group of proteobacteria (10). With the exception of H. pylori, no rpoBC gene sequences from this group of bacteria were known. Thus, it seemed possible that rpoBC fusion could be (i) characteristic of ⑀ proteobacteria in general; (ii) a specific feature of the Helicobacter genus; (iii) an accidental feature of the H. pylori species; or (iv) an accidental feature of the particular H. pylori strain that was sequenced. To assess these possibilities, we sequenced the rpoB-rpoC junction in four different strains of H. pylori, in an isolate of Helicobacter felis (11), and in two species of the related genus Campylobacter. In addition, we purifed and characterized the product of rpoBC gene from H. pylori 26695.
Our results establish that in H. pylori 26695 the rpoB-rpoC gene does in fact encode a fused ␤-␤Ј polypeptide. In addition, we find that translational fusion of rpoBC genes is characteristic of two gastric Helicobacters but not of Campylobacter jejuni and Campylobacter fetus, members of a related genus that colonize nongastric sites. We suggest that the ␤-␤Ј tethering in Helicobacter might (i) be an accident of evolution because of a frameshift mutation in an ancestor that, like current Campylobacters, contained overlapping but separate rpoB and rpoC genes or (ii) help gastric organisms cope with their acidand urea-rich niche.

EXPERIMENTAL PROCEDURES
Bacterial Growth and DNA Preparation-H. pylori were grown under microaerobic conditions (5% O 2 , 10% CO 2 , 85% N 2 ) on Brucella agar medium supplemented with 5% horse blood, 1% Isovitalex, amphotericin B (8 mg/liter), trimethoprim (5 mg/liter), vancomycin (6 mg/liter), essentially as in Ref. 12. For biochemical purification of RNAP H. pylori 26695 liquid cultures were grown in Brucella broth with 10% fetal calf serum in 500-ml screw capped flasks; the medium was equilibrated with 7% O 2 , 5% CO 2 in the microaerobic incubator for 1 h prior to inoculation, and then the bacteria were added, and the flasks were sealed and placed on a rotary shaker at 150 rpm. The bacteria were harvested in late log phase (OD 660 ϭ ϳ0.8), and the cell pellets were stored at Ϫ70°C. C. jejuni strain H840 and C. fetus strain have been previously described (13) and were grown at 37°C on Brucella agar plates in a microaerobic incubator maintained at 7% O 2 , 5% CO 2 .
Helicobacter genomic DNA was extracted from confluent cultures using the Qiamp tissue kit (Qiagen). Genomic DNA was purifed from 50 -100 mg of Campylobacter cell paste by resuspending cells in 200 l of buffer containing 25 mM Tris-HCl, 1 mM EDTA, pH 7.9, and 0.1 mg/ml RNase A. Cell suspension was extracted three times with equal volume of phenol, followed by chloroform extraction and ethanol pre- Protein Purification and Sequencing-4 g of H. pylori (strain 26695) cells were resuspended in 15 ml of lysis buffer (50 mM Tris-HCl, 100 mM NaCl, 10 mM EDTA, pH 7.9, 1 mM ␤-mercaptoethanol) and lysed by passage through an Emulsiflex C-5 homogenizer (Avestin). The lysate was cleared by low speed centrifugation, and PEI was added to a final concentration of 0.8%. The PEI pellet was collected by low speed centrifugation, washed by 20 ml of lysis buffer, and extracted with 20 ml of lysis buffer containing 1 M NaCl. Proteins in 1 M NaCl extract were precipitated with ammonium sulfate (0.7 g/ml extract), and the pellet was recovered by centrifugation, dissolved in 20 ml of lysis buffer, and loaded on a 1-ml heparin HiTrap cartridge (Amersham Pharmacia Biotech) equilibrated in the same buffer and attached to a Waters 650 chromatographer. The column was washed with the buffer ϩ 0.3 M NaCl and eluted with buffer ϩ 0.6 M NaCl. Fractions containing 300-kDa band (monitored by SDS-PAGE) were pooled concentrated on a C-100 concentrator (Amicon) to ϳ1 mg/ml, diluted 2-fold with glycerol, and stored at Ϫ20°C.
For protein sequencing, heparin affinity chromatography fraction containing ϳ5 g of 300-kDa band was blotted on polyvinylidene difluoride membrane after SDS-PAGE and submitted to the Rockefeller University Protein-DNA Techonology Center.

RESULTS
rpoBC Genes Are Fused in Helicobacter but Not in Campylobacter-Two primers that target the rpoB-rpoC junction and that are complimentary to highly conserved sequences in the 3Ј end of the rpoB portion and the 5Ј end of the rpoC portion of the H. pylori 26695 rpoBC gene were used for PCR amplification with genomic DNA from the following organisms: H. pylori strains Hp1 (14), J-166 (15), SS1 (16), and NCTC11638 (17); an isolate of Helicobacter felis; and isolates of two different Campylobacter species: C. jejuni, strain H840, and C. fetus (13). In all cases, a single major PCR fragment ϳ500 bp in length was amplified. The fragment was cloned, and its sequence was determined. Alignment of sequences at and around the rpoB-rpoC junction site is shown on Fig. 1. Each H. pylori strain differed from the 26695 sequence in ϳ10 of 474 positions (italicized in Fig. 1). This relatively high (ϳ2%) level of DNA polymorphism between different H. pylori strains is consistent with published data on the extent of polymorphism within H. pylori (18,19). Most of these differences involved third codon positions, and none resulted in changes in the deduced amino acid sequence of the protein. Thus, the rpoB-rpoC fusion is maintained in all four strains of H. pylori.
Equivalent DNA sequencing of 3Ј and 5Ј ends of the rpoB and rpoC homologs from H. felis (11), another gastric Helicobacter, showed that it also encoded a ␤-␤Ј fusion protein, although the deduced amino acid sequence of this portion of the H. felis protein differed from that of H. pylori in 20 of 158 positions (shaded in Fig. 1). Hence we conclude that such organization is not unique to H. pylori but may be a common feature of gastric Helicobacters.
C. jejuni and C. fetus are members of a genus closely related to Helicobacter, but they colonize the small intestine not gastric tissue (20). Equivalent DNA sequence analysis showed that rpoB and rpoC are separate genes in these two species (Fig. 1) in contrast to H. pylori and H. felis. In most Gram-negative bacteria, rpoB and rpoC are separated by a short, untranslated linker of 50 -100 bp with potential for extensive intrastrand pairing (3,5,6). 2 The Campylobacter genes are unusual in this context, because they overlap by two codons (no linker). We infer that the ATG codon shown in bold type in Fig. 1 probably encodes the first methionine of Campylobacter ␤Ј because: (i) it is near an appropriately spaced A/G-rich sequence that could serve as a ribosome-binding site (underlined in Fig. 1) and (ii) the next methionine residue that might possibly serve as initiator is at position 80, in the middle of the universally conserved segment A (1).
Helicobacter rpoBC Encodes a Fused ␤-␤Ј RNAP Subunit-The results of DNA sequencing experiments presented above do not test directly whether Helicobacter actually produces a fused RNAP subunit. To critically test this issue, we purified RNAP from H. pylori strain 26695.
RNAP ␤ and ␤Ј are among the largest proteins in bacterial cells and can be detected in whole cell extracts by SDS-PAGE.  1 and 2) contained a characteristic double band that comigrated with the ␤ and ␤Ј subunits of purified E. coli RNAP (lane 4). In contrast, H. pylori lysates (lane 3) contained no such ␤ and ␤Ј bands. Rather, a single band with an apparent mobility of ϳ300 kDa was observed. No such 300-kDa bands were detected in E. coli or C. jejuni lysates.
To establish whether this 300-kDa band in H. pylori lysates is indeed fused ␤-␤Ј, we began H. pylori RNAP purification using a standard procedure that involved PEI precipitation and extraction of RNAP from the PEI pellet with 1 M NaCl, followed by heparin affinity chromatography (see "Experimental Procedures"). The 300-kDa band was quantitatively precipitated by PEI at low (200 mM) NaCl concentration (Fig. 3, lane 3) and was extracted from the PEI pellet with buffer containing 1 M NaCl (Fig. 3, lane 4). Heparin affinity purification allowed further purification of the 300-kDa band (Fig. 3, lane 5). When heparin affinity chromatography fractions containing 300-kDa band were mixed with pure E. coli RNAP and loaded on a Superose-6 gel filtration column attached to fast protein liquid chromatography, the 300-kDa protein coeluted with the E. coli ␤, ␤Ј, and ␣ polypeptides (data not shown). Thus, the chromatographic behavior of the 300-kDa protein is consistent with its being part of RNAP of the ␤-␤Ј␣ 2 subunit composition.
When fractions containing the 300-kDa band were incubated in a standard E. coli transcription buffer in the presence of poly(dA-dT) and NTPs, no RNAP activity was observed. Control fractions containing E. coli RNAP of similar purity were highly active in this assay (data not shown). More importantly, 1 M extracts of C. jejuni PEI pellets also exhibited significant poly(A-U) synthesizing activity (data not shown). We were also unable to detect any transcription in crude H. pylori lysates with either poly(dA-dT) or strong E. coli promoters in the presence of exogenously added E. coli 70 subunit. Control extracts from E. coli containing the same amount of total protein were highly active in this assay (data not shown) Because no transcriptional activity was found associated with the 300-kDa band, we identified it by protein sequencing. The protein appeared to be N-terminally blocked, and therefore internal sequencing was performed after Lys-C protease digestion and high pressure liquid chromatography purification. The sequence obtained, IQQQYDQGLLTDQER, matches exactly to positions 2040 -2054 of the predicted fused product of H. pylori rpoBC gene product. We conclude that the 300-kDa band is the fused ␤-␤Ј RNAP subunit. DISCUSSION The principal conclusion of this work is that a single fused RNAP rpoB-rpoC gene is a regular feature of at least two species of gastric helicobacters, H. pylori and H. felis, and that the gene product, a 300-kDa ␤-␤Ј fusion protein is the predominant or only form of this gene product in vivo. Because no proteins in H. pylori lysates that would comigrate with separate RNAP ␤ and ␤Ј subunits were detected, the fused subunit is probably not extensively proteolyzed; it is likely to be the only source of RNAP ␤ and ␤Ј in the cell, as is also the case for an E. coli strain with a ␤-␤Ј fusion RNAP that we had engineered to study holoenzyme topology and assembly (9).
No transcriptional activity was found in fractions of H. pylori extract containing the ␤-␤Ј fused protein under standard conditions that had been optimized for E. coli RNAP transcription in vitro, although similarly prepared RNAP-containing fractions from extracts of E. coli and also of C. jejuni were active under our assay conditions. It is known that RNAPs from different eubacterial species can have markedly different requirements for efficient transcription in vitro (21)(22)(23), and hence, the inactivity of H. pylori RNAP under our present conditions may reflect an unusual buffer or salt requirement and conditions in the gastric environment in which it grows. Because H. pylori is a fastidious microbe and rather difficult to grow in large quantities, we have not yet purified the large amounts of H. pylori RNAP that should facilitate establishing a system for transcription by this RNAP in vitro.
Although H. pylori is extremely diverse as a species, our DNA sequencing results establish that the rpoBC fusion is not just a peculiar feature of the one H. pylori strain that was chosen for the genome project (26695), but rather it is a common feature of H. pylori in general and indeed of at least one other gastric helicobacter. In contrast, rpoB and rpoC are separate genes in the closely related genus, Campylobacter. The Campylobacter genus is also unusual among Gram-negative bacteria, however, because its rpoB and rpoC genes overlap by two codons. A frameshift mutation at or shortly before the overlap area might have created the continuous open reading frame found in present day helicobacters. Alternatively, a frameshift mutation in a Helicobacter-like ancestral rpoBC gene might have created separate but overlapping genes of present day campylobacters. A number of nongastric helicobacters and also members of closely related genera that colonize gastric and nongastric sites have been discovered recently (10,24), and DNA sequence analyses similar to those carried out here should help us learn how these unusual arrangements of RNAP subunit genes have evolved.
The functional significance of the rpoBC fusion is not known. In organisms with transcriptional-translational coupling, the rpoB and rpoC genes are always found in the same operon, suggesting that RNAP assembly in the cell may occur contranslationally. Fusion of the two genes may further increase RNAP assembly efficiency. In E. coli, the ␤-␤Ј fusion appears to stabilize RNAP in vitro and in vivo. 3 Thus the fusion could be advantageous for gastric helicobacters, which must grow in the putatively hostile, acid-rich stomach environment. On the other hand, many acidophilic archaea have separate rpoB-and rpoClike subunits and in fact contain natural splits in their ␤and ␤Ј-like subunits (7). Experiments aimed at directly testing the importance of the rpoBC fusion in H. pylori are in progress.