Transcription factor Rho does not require a free end to act as an RNA-DNA helicase on an RNA.

Escherichia coli Rho factor is a ring-shaped, homohexameric protein that terminates synthesis of RNA through interactions with the nascent RNA transcript. Because its mechanism of action may involve translocation of the RNA transcript through the hole in its ring structure, its action could depend on the availability of a free 5' terminus. To determine whether Rho's activity is 5'-end-dependent, its ability to bind to and function on a circular derivative of lambda cro mRNA was investigated. The circular derivative was made in vitro by action of RNA ligase on a derivative of lambda cro RNA containing an extra 10-nucleotide sequence near the 5'-end that was complementary to a sequence located near the 3'-end. Rho bound nearly as tightly to the circular derivative RNA as to the standard cro transcript. Rho was also able to readily dissociate a DNA oligonucleotide from its helical complex with the circular RNA in an ATP-dependent reaction. Thus, the action of Rho on a transcript does not depend on the availability of a free 5' terminus.

Transcription termination factor Rho is an RNA-binding protein that couples the energy derived from ATP hydrolysis to actions that dissociate a RNA transcript from its biosynthetic complex with DNA and RNA polymerase (1,2). Rho also can dissociate a RNA molecule bound to DNA through a short hybrid helix in a reaction that is dependent on ATP hydrolysis and the presence of an attachment site for Rho on the RNA on the 5Ј side of the hybrid helix (3)(4)(5). This RNA-DNA helicase activity may mimic the process of removal of the nascent RNA strand from the transcription elongation complex.
Rho factor is believed to function in vivo as a hexamer consisting of six polypeptide subunits arranged in a ring-shaped structure (6,7). Burgess and Richardson (8) have recently presented a model for the Rho-RNA complex in which a segment of RNA, called a rut site (Rho utilization site), binds to a cleft comprised of the N-terminal RNA-binding domains of the six individual subunits located at one end of the hexameric structure (the crown) (9 -11). RNA sequence 3Ј of the rut site then passes into the hole located at the center of the hexamer. Evidence for passage of a single-stranded nucleic acid substrate through a ring-shaped hexameric structure has been observed for other hexameric helicases, such as DnaB and the T7 gene 4 product (12,13).
The process by which Rho binds to and captures the 3Ј-end of an RNA in the center of the hexamer is not known. Gan and Richardson (14) have recently presented data indicating that Rho, in vitro, could form hexamers by partial assembly of subunits on an RNA. Another mechanism could have the RNA enter into the center of the hexameric structure through an opening or notch in the ring (7). A third mechanism may involve Rho threading onto the nascent transcript from the free 5Ј-end of the RNA. To test the model in which Rho requires a free end of RNA to thread onto in order to function as a terminator, we investigated Rho's ability to utilize its ATP-dependent helicase activity on an RNA lacking a free end. This can be done by using a circular RNA to test Rho's ability to disrupt RNA-DNA hybrid helices. These results indicate that Rho is indeed able to function as a helicase on an RNA lacking a free end.

EXPERIMENTAL PROCEDURES
Materials-All restriction enzymes, T4 RNA ligase, T4 polynucleotide kinase, and calf intestinal alkaline phosphatase were purchased from New England Biolabs, Inc. The oligonucleotides were purchased from Integrated DNA Technologies. All of the ribonucleotides were purchased from Roche Molecular Biochemicals. Radioactive nucleotides were purchased from ICN Radiochemicals. RNasin was purchased from Promega. T7 RNA polymerase and wild-type Rho were provided by Lislott Richardson (Indiana University).
pBcCro-pBcCro is a derivative of pIF2 (15) that contains a 10-base pair insertion sequence at the 5Ј-end of the transcription unit for the cro RNA. This derivative was prepared by polymerase chain reaction amplification of pIF2 with the mutagenic primers: primer 1 (5Ј-CGAC-TCACTATAGGGATCGTAGAGCCATTACTAAGGAGGTTG-3Ј) and primer 2 (5Ј-CAACCTCCTTAGTAATGGCTCTACGATCCCTATAGTG-AGTCG-3Ј). The resulting plasmid was sequenced to confirm the presence of the desired insert in the cro gene.
T7 Transcription of Linearized pBcCro-The pBcCro plasmid was prepared for T7 transcription by linearization with the restriction enzyme TaqI. Subsequent purification of the DNA, followed by transcription and purification of the RNA product were done as reported by Burgess and Richardson (8).
Circularization of the Linear pBcCro RNA Transcript-The 5Ј-terminal triphosphate of the transcript was removed by treatment of 20 g of RNA with 2 units of calf intestinal alkaline phosphatase in 50 l of a solution containing 20 mM Tris-HCl, pH 8.0, 50 mM KCl, 1 mM dithiothreitol, and 40 units of RNasin for 2 h at 23°C (16). The reaction mixture was then treated with an equal volume of phenol, followed by an equal volume of chloroform/isoamyl alcohol (24:1). The RNA was precipitated with ethanol, washed with 70% ethanol, dried, and dissolved in water.
The RNA was phosphorylated at its 5Ј-end by treatment with 20 units of T4 polynucleotide kinase in a 50-l reaction volume containing 10 mM Tris acetate, pH 7.5, 10 mM magnesium acetate, 50 mM potassium acetate, 8.0 M [␥-32 P]ATP (ϳ1500 Ci/mmol), and 40 units of RNasin for 2 h at 23°C. The RNA was again purified by treatment with phenol, chloroform/isoamyl alcohol (24:1), and ethanol precipitation.
The 5Ј-end labeled RNA was circularized by treatment with 25 units of T4 RNA ligase in a 50-l solution containing 25 mM Tris-HCl, pH 7.6, 10 mM dithiothreitol, 5 mM MgCl 2 , 5 mM ATP, and 0.25 mg/ml acetylated bovine serum albumin, and 40 units RNasin for 15 h at 17°C. The RNA was precipitated with ethanol; washed with 70% ethanol; dried; dissolved in 16 l of a solution containing 98% formamide, 2 mM EDTA, * This work was supported by National Institutes of Health Grant GM 56095. 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.
‡ To whom correspondence should be addressed: Dept. of Chemistry, 0.03% bromphenol blue, and 0.03% xylene cynanol; and separated by electrophoresis on a 6%, 7 M urea polyacrylamide gel (acrylamide/ bisacrylamide, 19:1) containing 45 mM Tris borate and 1 mM EDTA. The ligated, circularized RNA ran with a slower mobility than the unligated, linear form on this denaturing gel. Approximately 30% of the RNA in the ligation reaction was circularized. The RNA was identified by UV shadowing and excised using a sterile razor blade. The RNA was recovered as described by Burgess and Richardson (8). Concentration of the RNA was determined by its optical absorbance at 260 nm.
Filter Binding Experiments-Filter binding experiments involving both the unligated and circular derivatives of the cro RNA with wild-type Rho were performed and analyzed according to Gan and Richardson (14). All binding experiments were performed in a solution containing 40 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 50 mM KCl, 0.1 mM dithiothreitol, 0.1 mM EDTA, 1 mM ATP, and 0.25 mg/ml acetylated bovine serum albumin.
Annealing of DNA Oligonucleotides to the RNA-DNA oligonucleotides complementary to nucleotides 28 -46 (5Ј oligonucleotide), 326 -348 (rut oligonucleotide), and 327-348 (3Ј oligonucleotide) of the standard cro RNA was chosen for these helicase experiments (see Fig. 1 for sequences) The annealing reaction contained 50 nM 32 P-labeled RNA, 500 nM DNA oligonucleotide, 20 mM Hepes-KOH buffer, pH 7.5, 150 mM KCl, and 0.1 mM EDTA. The annealing mixture was placed in a thermocycler (Minicycler; MJ Research), incubated at 90°C for 2 min, followed by a cooling phase of 2°C/min to 25°C. The annealed RNA-DNA hybrids were stored at Ϫ20°C. Under these conditions, ϳ85% of the RNA was in the hybrid form.
Rho RNA-DNA Helicase Reaction-The helicase reactions contained one-tenth volume of the annealing reaction (50 nM RNA-500 nM DNA oligomer), which resulted in ϳ4.5 nM RNA-DNA hybrid. The reactions also contained 32 mM Hepes-KOH buffer, pH 8.0, 50 mM KCl, 1 mM MgCl 2 , 0.1 mM EDTA, 1 mM dithiothreitol, 40 units of RNasin, 0.2 mg/ml acetylated bovine serum albumin, and 5 nM wild-type Rho hexamer. The reaction was heated to 37°C for 1 min, and the reaction was initiated by the addition of 1 mM ATP. Samples (6 l) were taken at the indicated times and mixed with 3 l of a buffer containing 150 mM Tris-HCl, pH 6.8, 4% SDS, 0.2% bromphenol blue, and 20% glycerol. The helicase reactions were analyzed on a discontinuous Laemmli polyacrylamide gel (17); stacking gel contained 5% polyacrylamide (acrylamide/bisacrylamide, 29:1), 125 mM Tris-HCl, pH 6.8, and 0.1% SDS; resolving gel contained 380 mM Tris-HCl, pH 8.8, and 0.1% SDS. The gel was run in a buffer containing 25 mM Tris, 250 mM glycine, and 0.1% SDS. These gels were run at low current (11 mA) and temperature (4°C) to prevent disruption of the RNA-DNA hybrid during electrophoresis. Following exposure of the gel to a PhosphorImager plate, the data were analyzed using ImageQuant software (Molecular Dynamics, Inc., Sunnyvale, CA).
The data points relating the fraction of RNA-DNA hybrid remaining versus time for each sample were fit to the following two-phase decay equation (4), where F is the fraction RNA-DNA hybrid remaining after time t (t ϭ 0 s was normalized to 1), A 1 and k 1 are the amplitude and rate constant (units s Ϫ1 ), respectively, for the first fast burst (exponential) decay phase, and A 2 and k 2 are the intercept and slope, respectively, of the second slower linear-decay phase.

RESULTS AND DISCUSSION
To determine whether Rho could bind to and act on a transcript lacking a free 5Ј-end, a circular derivative of cro RNA, a transcript that is terminated at the well characterized Rhodependent terminator tR1 located between genes cro and cII of bacteriophage , was prepared. The method chosen to make a circular RNA (16) involved preparing a 391-nucleotide derivative of the cro RNA, which contained an inserted sequence of 10 nucleotides (between nucleotides 4 and 5 in the wild-type cro RNA sequence) that was complementary to a 10-nucleotide sequence near the 3Ј-end of the RNA (nucleotides 366 -375). With this extra sequence, the RNA can form a stable stem helix, which would bring the 5Ј-and 3Ј-ends into close proximity. After converting the 5Ј-triphosphate of the primary transcript to a monophosphate, treatment of this RNA with T4 RNA ligase converted ϳ30% of the RNA to a circular form. The ligated, circular RNA was separated from the unligated RNA because of its slower mobility during electrophoresis on a 7 M urea polyacrylamide gel.
Rho Binds to Circular RNA-To determine whether Rho can bind the circular cro derivative RNA, we used the nitrocellulose filter-binding technique to isolate complexes of Rho with radiolabeled RNA (15,18). With saturating levels of Rho, ϳ60% of the circular cro RNA was retained on the filter. This efficiency of retention of the complex was nearly identical to that found with the standard cro RNA under the same binding conditions (14). The affinity of Rho for the RNAs was determined by measuring the fraction of RNA retained on the filter at varying concentrations of Rho and finding the best fit of these data to the equation for formation of a simple bimolecular complex between the RNA and hexameric Rho. As was shown for the standard cro RNA, the assumption that Rho bound RNA as a hexamer is a good approximation when ATP was present in the binding mixture (14). The K d values determined by this procedure were: 0.03 Ϯ 0.01, 0.07 Ϯ 0.01, and 0.14 Ϯ 0.03 nM for the complex with the standard cro RNA, the ligated (circular) cro derivative RNA, and unligated cro derivative RNA, respectively. Thus, Rho does not bind as tightly to the derivative cro RNAs as to the standard cro RNA, but it does bind with slightly higher affinity to the circular RNA than to its unligated precursor.
One possible reason why Rho binds with a somewhat weaker affinity to the derivative RNAs than to the standard cro RNA is that the formation of the stem-helix between the 3Ј-and 5Јends in the derivative RNAs has changed their overall tertiary structure. These structural changes may have made the binding site on the derivative RNAs less accessible to Rho than on the standard cro RNA. In any case, these data indicate that Rho can bind to the circular RNA nearly as well as it binds to the standard cro RNA.
Rho Exhibits ATP-dependent RNA-DNA Helicase Activity on a Circular RNA Substrate-Rho can utilize the energy derived from ATP hydrolysis to translocate along a RNA in a 3Ј direction and to disrupt RNA-DNA hybrids that are 3Ј of the primary Rho attachment site (3,4). To determine if Rho can perform this ATP-driven reaction on a circular RNA, a DNA oligonucleotide complementary to a sequence on the 3Ј side of the rut site (Fig. 1, 3Ј oligo) was annealed to the RNA. When separated on a polyacrylamide gel containing SDS (no urea), the circular RNA-DNA oligonucleotide complex migrated more slowly than did the free RNA (Fig. 2). Incubation of the complex with a stoichiometric amount of Rho and ATP caused a timedependent displacement of 3Ј oligonucleotide from the circular RNA. 86% of the oligonucleotide was displaced within 6 min. In the absence of either Rho or ATP, less than 7% of 3Ј oligonucleotide was displaced within 6 min. These results demonstrate that Rho acts as a helicase on a circular RNA.
To rule out the possibility that the displacement occurred because there was sufficient contaminating nuclease present to nick a majority of the circular RNA, samples of the reaction mixture were also analyzed on a polyacrylamide gel containing 7 M urea. The results revealed that Ն95% of the circular RNA retained its characteristic slower mobility on the 7 M urea polyacrylamide gel. Hence, nearly all of the circular RNA remained covalently sealed throughout the course of the helicase reaction.
Kinetics of the RNA-DNA Helicase Reaction-To determine whether the absence of a free end alters the rapidity by which Rho can separate RNA from a hybrid complex with a DNA oligonucleotide, the time courses for the displacement reaction for the various RNAs was determined. With the standard cro RNA, the displacement reaction had two phases. About half of the complexes were dissociated rapidly, while the rest were dissociated at a much slower rate (Fig. 3). With the two cro derivatives, the initial displacement was less rapid but involved a larger fraction of the complexes. The two-phase reaction observed for these RNAs was very similar to results obtained by Walstrom et al. (5) performing similar experiments using DNA oligonucleotides complexed with trptЈ RNA as the substrate in their helicase reaction. These reactions can be described in terms of a two-phase rate expression composed of a rapid, first-order burst phase followed by a slower linear phase (see Equation 1). The curves in Fig. 3 represent the fit of the data to the equation. The corresponding rate parameters are presented in Table I. For the hybrid complex with the standard cro RNA, about 38% of the hybrid complexes were disrupted in the initial burst phase, with a first-order rate constant (k 1 ) of 0.28 s Ϫ1 . The remaining complexes were dissociated with a steady-state rate constant (k 2 ) of 8 ϫ 10 Ϫ4 s Ϫ1 . In contrast, for both of the cro derivatives, about 80% of the complexes were dissociated in the burst phase; however, the rates (ϳ0.03 s Ϫ1 ) were about 10-fold slower than the rate determined for the standard cro RNA. The small fraction of complexes remaining for both of the derivative cro RNAs were dissociated with a k 2 of Ͻ5 ϫ 10 Ϫ4 s Ϫ1 . These results indicate that the RNA-DNA hybrid complexes with the derivative cro RNAs were poorer substrates for helix displacement by Rho in the initial burst phase reaction than the complexes with the standard cro RNA.
One possible explanation for differing rates of the initial burst phase between the standard cro RNA and the derivative RNAs is that formation of the stem structure between the 3Јand 5Ј-ends in the derivative RNAs, ligated or not, makes the RNA less accessible to Rho for immediate action. Another possibility is that the overall changes in the tertiary imposed on the two derivative RNAs due to the formation of the stem structure alters Rho's ability to efficiently displace downstream hybrid complexes. The second explanation may be more likely, since the RNA-DNA complexes were preincubated with Rho for two min at 37°C prior to the addition of ATP, allowing Rho ample time to bind to the complexes. Whatever the reason, the ligation of the 3Ј-and 5Ј-ends did not make the circular derivative any less reactive than its unligated counterpart.
Investigation of Rho's Tracking along an RNA Molecule-The helicase action of Rho has been shown to be directional. Rho will disrupt RNA-DNA hybrids located 3Ј of the rut site but not hybrids located at the 5Ј-end of the transcript (3,4). Another way to demonstrate Rho's ability to act on a circular RNA is to show that it can disrupt a hybrid with a DNA oligonucleotide (called 5Ј oligonucleotide) that is complementary to a segment of RNA near the 5Ј-end of the original transcript that was used to make the circular derivative (Fig. 1). After a 6-min incubation of the hybrid complex with ATP and a stoichiometric amount of Rho, 44% of the 5Ј oligonucleotide was removed from the circular RNA as compared with 71% of the 3Ј oligonucleotide from its complex (Table II). In contrast with the unligated cro RNA, only 12% of the 5Ј oligonucleotide was displaced, whereas again 71% of the 3Ј oligonucleotide was displaced. On the ligated, circular cro RNA, the 5Ј oligonucleotide was located further downstream (54 nucleotides) from the rut site than the 3Ј oligonucleotide was in its complex. In addition, the 5Ј oligonucleotide hybrid was downstream of the stem structure that brought the 3Ј-and 5Ј-ends together for ligation. Both of these conditions are likely reasons why the 5Ј oligonucleotide was removed less efficiently than the 3Ј oligonucleotide from the circular cro RNA. On the other hand, the much lower displacement of the 5Ј oligonucleotide from the unligated derivative RNA containing the same stem structure demonstrates a dependence on the covalent continuity of the RNA across this segment of RNA for the improved efficiency of displacement. This is thus further evidence that Rho can function on a circular RNA.
This result with the unligated cro derivative is also consistent with other observations that Rho will not displace a hybrid that is on the 5Ј side of the rut sequence. This latter observation is important because of the unexpected finding that Rho was able to displace as much as 30% of the 5Ј oligonucleotide within 6 min (Table II) from a complex with this oligonucleotide hybridized to the standard cro RNA. The reason for this partial but significant level of displacement is unknown. However, the displacement was dependent on the presence of both Rho and ATP, since less than 2% of this hybrid was displaced from the standard cro RNA or either of the cro derivatives when incubated with either ATP or Rho alone for 6 min (data not shown). In all of the previous cases in which there was little or no displacement of an upstream DNA hybrid, the 5Ј-end of the RNA transcript was in a double helical structure. In experiments with trptЈ RNA (3,4), the upstream hybrid was at the 5Ј-end and in our experiment with the unligated cro derivative RNA, the 5Ј-end was in a helical stem with the 3Ј-end of the transcript. In contrast to those RNAs, the standard cro RNA had a 5Ј RNA segment of 28 nucleotides that could provide Rho access to an upstream hybrid from the 5Ј side. Because the amount of Rho was stoichiometric with the amount of RNA under the conditions used, most, if not all, of the Rho molecules would be bound to the rut segment of the cro RNA (19). In this case, an interaction of the 5Ј-end of an RNA with the hybrid dissociation site in Rho could occur by looping of the same RNA into the site or by action on a second, separate RNA molecule.
To investigate the importance of the rut site to Rho action, the rut oligonucleotide was used to block the beginning of the rut site for all three RNAs. Faus and Richardson (15) showed that blocking this portion of the rut site with an oligonucleotide inhibited Rho's ability to bind to cro RNA. Also, Chen et al. (20) showed that blocking the rut site with a similar DNA oligonucleotide inhibited Rho's in vitro termination activity at the tR1 terminator of the cro gene. As expected from these prior observations, Rho displaced less than 7% of the rut oligonucleotide from the complexes with any of the three cro RNAs (Table  II). These results also provide further evidence that the helicase activity is a faithful representation of Rho's action in transcription termination.
Implications for Rho's Function in Vivo-According to a recent model of Rho-RNA interaction presented by Burgess and Richardson (8), Rho is believed to initially bind to a loading site located on the RNA transcript through interactions in a cleft located around the crown of the hexameric structure, composed of the N-terminal RNA-binding domains of the individual subunits of Rho. Following this primary RNA binding event, RNA is then passed into the center of the ring. The mechanism for this transit is unknown. From these results, a mechanism that requires Rho to thread onto a free 5Ј-end of the RNA for function is unlikely. Other possible mechanisms include either passage of the RNA through an opening of a notched form of the hexamer (as seen in electron micrographic images of Rho) (6,7) or through an opening caused by partial disassembly of the hexamer on the RNA (14).
In its normal context at the end of an operon, a Rho-dependent terminator would encode a rut segment that would be located downstream of the translated segment of the RNA. In order for Rho to load onto this RNA, it must be able to bind an  RNA from which the 5Ј-end would be blocked by co-translating ribosomes and the 3Ј-end would be blocked by the RNA polymerase. Thus, Rho would have to bind by a mechanism that is not dependent on the availability of a free end. A similar constraint would apply to the loading of Rho onto an RNA segment downstream of a ribosome stalled on a nascent RNA during amino acid starvation, as happens for the functioning of latent intragenic terminators under starvation conditions (21). Thus, these results with the circular RNA are consistent with the expected mechanism of Rho action in the cell.