Transcription Termination Factor Rho Contains Three Noncatalytic Nucleotide Binding Sites*

The active form of transcription termination factor rho from Escherichia coli is a homohexamer, but several studies suggest that the six subunits of the hexamer are not functionally identical. Rho has three tight and three weak ATP binding sites. Based on our findings, we propose that the tight nucleotide binding sites are noncatalytic and the weak sites are catalytic. In the presence of RNA, the rho-catalyzed ATPase rate is fast, close to 30 s−1. However, under these conditions the three tightly bound nucleotides dissociate from the rho hexamer at a slow rate of 0.02 s−1, indicating that the three tight nucleotide binding sites of rho do not participate in the fast ATPase turnover. These slowly exchanging nucleotide binding sites of rho are capable of hydrolyzing ATP, but the resulting products (ADP and Pi) bind tightly and dissociate from rho about 1500 times slower than the fast ATPase turnover. Both RNA and excess ATP in solution are necessary for stabilizing nucleotide binding at these sites. In the absence of RNA, or when solution ATP is hydrolyzed to ADP, a faster dissociation of nucleotides was observed. Based on these results, we propose that the rho hexamer is similar to the F1-ATPase and T7 DNA helicase-containing noncatalytic sites that do not participate in the fast ATPase turnover. We propose that the three tight sites on rho are the noncatalytic sites and the three weak sites are the catalytic sites.

The transcription termination factor rho is an essential protein in Escherichia coli that is required for the release of certain nascent RNAs from the transcription complex (1)(2)(3)(4). The rho protein contains an ATPase-dependent helicase activity that can separate the strands of an RNA-DNA duplex. Transcription termination by rho starts after the rho protein binds to the nascent RNA transcript at a cytosine-rich entry site (5). The RNA binding event activates the ATPase activity of rho, which is believed to fuel a directional 5Ј 3 3Ј movement of the protein along the nascent RNA chain. Both the ATPase and the helicase activities are necessary for the rho-dependent termination event (6). Thus, it is believed that translocation and RNA-DNA duplex unwinding activities cause disruption of the elongating complex, which results in transcription termination (7)(8)(9).
Several studies of the rho protein have shown that it assem-bles into a planar hexagonal structure that contains a central hole (10,11). The self-association of rho subunits into a hexamer is reversible in solution, and stable hexamer is formed in the presence of ATP and/or RNA (12,13). Each subunit of rho hexamer contains two functionally distinct domains, an ATPand an RNA-binding domain (1). The ATP-binding domain of rho shows greater than 50% amino acid sequence homology to the ␤ subunit of F 1 -ATPase (14). The tertiary structure of the RNA-binding domain of rho is also very similar to the Nterminal domain of F 1 -ATPase, despite little amino acid sequence homology between these domains of the two proteins (15). A recent report has also shown similarities among the ATPase mechanism of rho, the F 1 -ATPase, and T7 DNA helicase (23,24). These observations support the proposal that the structure of the rho hexamer is likely to be similar to that of the F 1 -ATPase (14 -16). Several studies in the literature indicate that the rho hexamer is not symmetrical. Geiselmann et al. (17) suggested a D 3 symmetry for the rho hexamer, and recently Horiguchi et al. (18), using extensive chemical cross-linking experiments, have suggested a C 3 or a C 3/6 symmetry for the rho quaternary geometry. Equilibrium nucleotide binding experiments have shown that the rho hexamer binds only three ATP molecules with a high affinity (19). Others have shown that in addition to these high affinity sites, there are three low affinity ATP binding sites on rho (20,21). Thus, rho protein falls into the class of hexameric helicases, such as the E. coli DnaB (22) and T7 DNA helicase (23) that were shown to contain two classes of nucleotide binding sites. The three tight nucleotide binding sites on T7 DNA helicase hexamer were shown to be noncatalytic, with properties similar to such sites of the F 1 -ATPase protein. Because rho protein is structurally similar to the F 1 -ATPase protein, we have investigated the possibility of noncatalytic sites on the rho hexamer.
Our results show that the two classes of nucleotide binding sites on rho catalyze ATPase turnover at different rates. The ATPase turnover rate at one class of nucleotide binding sites was about 1500 times slower than the turnover rate at the other sites. These studies allow a better understanding of the nature of the two classes of nucleotide binding sites on the rho hexamer. We propose that the three tight nucleotide binding sites on the rho hexamer are noncatalytic, serving a role similar to such sites of the F 1 -ATPase and T7 DNA helicase, and that the weak sites are the catalytic sites.

MATERIALS AND METHODS
Rho Protein-Rho protein was overexpressed in E. coli strain HB101 carrying the rho overexpression plasmid pKS26 (25) and purified according to Finger and Richardson (26) with slight modifications. Briefly, protein extracts from heat-induced cells were subjected to fractionation with Polymin P, ammonium sulfate precipitation, and successive chromatography on SP-Sepharose FF and heparin-Sepharose (Amersham Pharmacia Biotech). The final preparation was stored at Ϫ20°C in rho storage buffer containing 50% glycerol, 20 mM Tris-HCl (pH 7.9), 100 * This work was supported by National Institutes of Health Grant GM55310. 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.
Nucleotides, RNA Homopolymers, and Other Reagents-ATP and RNA homopolymers (poly(C) and poly(U)) were purchased from Amersham Pharmacia Biotech. Poly(C) RNA had a reported s 20,w value of 7.1 in 0.015 M NaCl, 0.0015 M sodium citrate buffer, pH 7.0, with an average length of 420 bases. Poly(U) had an s 20,w value of 6.0, with an average length of 290 bases. Poly(C) RNA concentration was determined by UV absorption at 269 nm, using an extinction coefficient of 6200 M Ϫ1 cm Ϫ1 for the cytosine base, and poly(U) RNA concentration at 260 nm, with an extinction coefficient of 9350 M Ϫ1 cm Ϫ1 for uracil base. These RNAs were dissolved in TE buffer (40 mM Tris-HCl, pH 7.0, 0.5 mM EDTA) and used without further purification. [␥-32 P]ATP and [␣-32 P]ATP were purchased from Amersham Pharmacia Biotech, and their purity was assessed by polyethyleneimine (PEI) 1 -cellulose TLC and corrected for in all experiments. The ATP-regenerating reagents, phosphocreatine and creatine kinase, were purchased from Roche Molecular Biochemicals.
Equilibrium Binding of ATP to Rho Protein-Equilibrium binding of ATP to rho was measured using the nitrocellulose membrane binding assay. The nitrocellulose membrane circles (25 mm) were treated with 0.5 N NaOH, rinsed extensively with water, and equilibrated in the membrane wash buffer (40 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 50 mM KCl, and 0.1 mM dithiothreitol) before use. The reaction (15 l) contained 0.5-500 M radiolabeled ATP ([␣-32 P]ATP) in binding buffer (40 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 50 mM KCl, 0.1 mM dithiothreitol, and 10% (v/v) glycerol). Rho protein was added to a final concentration of 1.0 M (hexamer) and the reactions were incubated for 30 s at 18°C. 12 l of samples were then filtered through the nitrocellulose membrane assembly. The membranes were washed with 0.5 ml of wash buffer before and after filtration. 1-l aliquots of each sample were spotted on a separate nitrocellulose membrane to quantitate total ATP. The radioactivity on the membrane was quantitated on a PhosphorImager (Molecular Dynamics). The molar amount of ATP bound to each rho hexamer was determined and plotted versus ATP concentration. The resulting titration data were fit using a hyperbolic equation to obtain the apparent K d , the equilibrium dissociation constant of rho ATP complex.
ATPase Activity-The ATPase activity of rho protein was measured at 18°C. The rho protein was incubated in the ATPase reaction buffer (40 mM Tris-HCl, pH 7.8, 100 mM KCl, 10 mM MgCl 2 , 0.1 mM dithiothreitol and 10% (v/v) glycerol) in the presence of the indicated amount of poly(C) or poly(U) RNA. Trace amounts of a Ͼ90% pure [␥-32 P]ATP (3000 Ci/mmol), together with nonradioactive ATP (1.0 -3.0 mM), were added to start the reaction (30 -50-l reaction volumes). At various times, aliquots (5 l) were withdrawn and quenched with 10 l of 1 M HCl and 20 l of chloroform. Quenched reactions were neutralized by adding 1 M NaOH, 0.25 M Tris base (about 9 l), and 1 l of the quenched reactions was spotted on a PEI-cellulose thin layer chromatography plate (Whatman). The radioactive ATP and P i were separated using 0.3 M potassium phosphate, pH 3.4, as the chromatography running buffer. The ATP and P i spots were quantitated using the Phos-phorImager. The initial rates of P i formation were determined from plots of molar amounts of P i versus time of reaction.
Kinetics of Nucleotide Exchange and Identification of the Tightly Bound Nucleotides-Either [␥-32 P]ATP or [␣-32 P]ATP was used in this experiment. All the nucleotide exchange experiments were carried out in an 18°C room. Rho protein (0.276 mg/ml, 1.0 M rho hexamer) was pre-incubated in the ATPase reaction buffer with a mixture of nonradioactive ATP (1.0 mM) and radioactive ATP for less than 10 s, either in the absence or in the presence of RNA. The reaction mixture was subsequently chased with nonradioactive ATP (10 mM in the same buffer as above). At defined intervals after addition of the chase ATP, 10 -15-l aliquots were removed and filtered through pre-wetted nitrocellulose membrane filters. 1 l of each sample was spotted on a separate nitrocellulose membrane to quantify the total ATP. Nonspecific binding of ATP to the nitrocellulose filters was measured by replacing rho protein with buffer and corrected in all experiments. The resulting nucleotide binding/exchange data were fit using nonlinear regression analysis (SigmaPlot) to obtain the dissociation rate of rho-bound nucleotides. The chemical form of the nucleotides bound to rho was identified by extracting the nitrocellulose membrane filters with 0.5 M perchloric acid, neutralizing with 0.25 M Tris, 1 M NaOH, and analyzing the extracted nucleotides by PEI-cellulose TLC.

RESULTS
Interaction of Rho with ATP-To investigate the interaction of rho protein with ATP and to determine the number of ATPs bound to rho at equilibrium, we have used nitrocellulose filter binding assay. A constant amount of rho protein was titrated with increasing amounts of [␣-32 P]ATP, and the rho-bound ATP was separated from free ATP by filtration through nitrocellulose membranes. As shown in Fig. 1, three ATP molecules were bound to rho at high ATP concentrations. The apparent K d of rho-ATP complex was determined to be 17 M. These results are consistent with studies that show that although rho has six identical ATP binding sites, rho binds only three ATPs at equilibrium (19 -21). The ATPs bound to the weak sites are not easily detectable by this membrane filtration method, possibly because these ATPs dissociate during the filtration and washing steps.
We then explored the possibility of noncatalytic sites on rho. We designed experiments to determine whether the ATPs bound at the tight sites were catalytically competent. The experiment consisted of measuring the dissociation rate of nucleotides bound to the tight sites (k off ) and comparing this value to the ATPase turnover rate, k cat . If the tight sites are catalytically competent, then the k off should be equal to or faster than the steady state k cat . If the k off from these sites is slower than the k cat , then the ATPs bound to the tight sites cannot be catalytically competent and are likely to be noncatalytic.
The ATPase Turnover Rates in the Presence of Poly(C) and Poly(U) RNAs-The rho-catalyzed ATPase turnover rate was measured at 18°C in the presence of poly(C) or poly(U) RNA. The concentrations of RNA and ATP were kept at saturating values; thus the measured velocity would be the V max value, which, divided by the rho hexamer concentration, provided the k cat . Table I lists the k cat values obtained in the presence of poly(C) and poly(U) RNAs. In the presence of poly(C) RNA, the ATPase k cat was close to 30 mol of ATP hydrolyzed per s per rho hexamer. Under the same conditions, the poly(U)-stimulated 1 The abbreviation used is: PEI, polyethyleneimine. ATPase k cat was only 7.2 mol of ATP hydrolyzed per s per rho hexamer. This rate appears to be lower because of the weaker affinity of rho for the poly(U) RNA. Accordingly, the ATPase k cat increased to 11.4 mol of ATP hydrolyzed per s per rho hexamer when the poly(U) RNA was increased to 0.33 g/l.
The Kinetics of Nucleotide Dissociation from the Rho Hexamer-The following experiment was designed to determine the dissociation rate of the three nucleotides bound to the rho protein at equilibrium. As shown in Scheme I, the experiment consisted of preincubating a complex of rho and RNA with radioactive ATP and then chasing the complex with an excess of nonradioactive ATP. The radioactive nucleotide bound to rho after various chase times were quantitated by the nitrocellulose filter binding assay. The results of such an experiment with poly(C) RNA and [␣-32 P]ATP are shown in Fig. 2A. Before chase was added, there were about 3 nucleotides bound per rho hexamer ( Fig. 2A). After chase was added, all of the ␣-32 Plabeled nucleotides dissociated from the rho protein with a single exponential rate of 0.02 Ϯ 0.002 s Ϫ1 ( Fig. 2A). A similar nucleotide binding/exchange experiment, in the presence of poly(U) RNA at 0.2 g/l, provided a 10-fold faster nucleotide dissociation rate constant of 0.15 Ϯ 0.02 s Ϫ1 (Fig. 2B). When the poly(U) concentration was increased about 16-fold, the dissociation rate decreased to the same level as observed with poly(C) RNA (0.033 Ϯ 0.005 s Ϫ1 at 3.3 g/l of poly(U) RNA). When the experiment was carried out in the absence of RNA (Fig. 2C), 3-4 nucleotides were bound to rho at equilibrium, in the absence of the chase. After chase was added, the ␣-32 Plabeled nucleotides dissociated from the rho hexamer at a rate that was too fast to measure manually. Table II compares the k off values of rho-bound nucleotides and the rho-catalyzed ATPase turnover rates. In the presence of poly(C) RNA, the k off of the three tight nucleotides was about 1500 times slower than the ATPase turnover rate. In the presence of poly(U) RNA, the k off of the nucleotides was 50 -340 times slower than the ATPase turnover rate. Thus the k off of the tightly bound nucleotides was consistently slower than the RNA-stimulated ATPase turnover rate. These results indicate that the nucleotides bound at the three tight sites on the rho hexamer do not participate in the fast ATPase turnover.
Identification of the Slowly Exchanging Nucleotides Bound at the Tight Sites of Rho-The above experiments indicated that the ATPs bound at the tight sites exchanged very slowly with solution ATP. The next question was whether the tightly bound nucleotides were ATP or ADP. In other words, does ATP get hydrolyzed at these tight sites? Two experiments were designed to determine the identity (ATP versus ADP) of the slowly exchanging nucleotides. These experiments were carried out in the presence of poly(C) RNA. Essentially the same experiment described above and shown in Fig. 2A was performed with [␣-32 P]ATP. After various chase times, the nitrocellulose membrane that contained the rho-bound radioactive nucleotide was extracted, and the nucleotides were analyzed by PEIcellulose TLC. As shown in Fig. 3, this experiment showed that most of the nucleotides bound to rho both before and after addition of the chase were ADP. The total amount of rho-bound ADP decreased with increasing chase time, as expected. A control experiment (Fig. 3, lane 9) showed that about one-third of the radioactive ATP was hydrolyzed with rho and RNA during the Ͻ10-s preincubation period. Therefore, the ATP substrate was not exhausted during the preincubation period, and the rho-bound radioactive ADP did not result from rebinding of radioactive ADP from solution. Similarly, we also determined that the acid extraction procedure resulted in less than SCHEME I FIG. 2. Dissociation kinetics of rho-bound ␣-32 P-labeled nucleotides. The dissociation rate of radioactive nucleotides bound to rho was measured in the absence and in the presence of poly(C) and poly(U) RNAs at 18°C as shown in Scheme I. A, rho protein (1.0 M, hexamer) and poly(C) RNA (0.2 g/l) were preincubated and mixed with [␣-32 P]ATP (1.0 mM) for about 7 s before nonradioactive ATP (10 mM) was added as a chase. After varying chase times (10 -400 s), rho-bound radioactive nucleotide was quantitated by nitrocellulose membrane binding assay. In the presence of poly(C) RNA, a total of 3.1 Ϯ 0.48 nucleotides (error calculated from five independent measurements) were bound to rho before chase was added (shown as q at zero chase time). After ATP chase was added, the rho-bound nucleotides dissociated with an exponential rate constant of 0.02 Ϯ 0.002 s Ϫ1 (E). B shows the dissociation kinetics of rho-bound ␣-32 P-labeled nucleotides in the presence of two different poly(U) RNA concentrations (0.2 g/l, shown as Ⅺ, and 3.3 g/l, shown as E). About 3-4 nucleotides were bound to the rho hexamer at both RNA concentrations before ATP chase was added (shown as f, q). The three rho-bound nucleotides dissociated with a rate constant of 0.15 Ϯ 0.02 s Ϫ1 (Ⅺ) in the presence of 0.2 g/l RNA and 0.033 Ϯ 0.005 s Ϫ1 (E) in the presence of 3.3 g/l poly(U) RNA. C, in the absence of RNA, a total of 4.2 Ϯ 0.54 nucleotides were bound per rho hexamer before chase was added (shown as f). After ATP chase was added, the rho-bound nucleotides dissociated at a rate that was too fast to manually measure (Ⅺ). 7% hydrolysis of ATP to ADP (Fig. 3, lane 11). The above results indicate that the tight sites are capable of hydrolyzing ATP, but the hydrolysis product dissociates from the tight sites at a slow rate.
To confirm these results, a second nucleotide binding/exchange experiment was carried out with [␥-32 P]ATP instead of [␣-32 P]ATP. If the ATPs are hydrolyzed at the tight nucleotide binding sites of rho and if the radioactive P i does not remain tightly bound, then we should observe no detectable radioactivity on the nitrocellulose membranes in the nucleotide binding/chase experiment with [␥-32 P]ATP. As shown in Fig. 4, we observed radioactivity bound to rho which corresponded to about one site per hexamer, before chase was added. This radioactivity was lost after addition of the chase with the same slow rate constant of 0.02 s Ϫ1 as observed in the [␣-32 P]ATP experiment. Because the radiolabel on ATP is in the ␥ position, the slowly exchanging species must be ATP or P i . Moreover, because the nucleotide identification experiment had shown very little rho-bound ATP at the tight sites, we conclude that about 30% P i is tightly bound to rho, either as free P i or in a complex with ADP, such as ADP⅐P i .
Cooperativity Between the Catalytic and the Noncatalytic Sites of Rho-Results thus far suggested that the ATPase turnover at the noncatalytic sites was about 1500 times slower than the ATPase turnover at the catalytic sites. We wanted to investigate whether there was cooperativity between the catalytic and noncatalytic ATPase sites of rho. To this end, we measured the ATPase turnover at the noncatalytic sites by measuring the k off of nucleotides from the tight sites under two conditions. In one experiment, we measured the k off under reaction conditions where active hydrolysis of ATP was not occurring at the catalytic sites. In a second experiment, we measured the k off under reaction conditions where hydrolysis was occurring at the catalytic sites. To satisfy the first condition, a mixture of rho protein and RNA was incubated with a lower concentration of [␣-32 P]ATP (0.3 mM). Because of the lower ATP concentration, most of the ATP was hydrolyzed with RNA in the 2-min period before nonradioactive ATP chase was added, as shown in Fig. 5 (inset). Under these conditions, there were less than three radioactive nucleotides bound to rho at the tight sites, and these nucleotides dissociated at a rate that was too fast to manually measure. To satisfy the second condition, the same nucleotide binding/exchange experiment was carried out in the presence of an ATP regeneration system. Fig. 5, inset, shows that in the presence of the ATP regeneration system, the reaction mixture contained mostly ATP at the time chase was added. The three radioactive nucleotides that were bound prior to addition of the chase dissociated at a slow rate of 0.015 s Ϫ1 (Fig. 5). These results indicate that the three tight sites of rho retain their nucleotides at the active site as long as there is excess ATP in solution. Rho undergoes active ATP binding/hydrolysis events at the catalytic sites when ATP is present in solution. This active ATP binding and hydrolysis event at the catalytic sites appears to be necessary for retaining the ADP and P i at the tight sites. These results suggest cooperativity among the two types of nucleotide binding sites of rho. DISCUSSION Rho protein is known to contain two classes of nucleotide binding sites. About 3-4 ATPs bind to rho protein tightly and 2-3 bind weakly at equilibrium (20). Very little is known about the significance and the role of these two classes of ATP binding sites. It has been assumed that the tight sites are the catalytic sites. The results reported in this paper clarify the role of the tight ATP binding sites on the rho hexamer. The findings from this study indicate that the tight nucleotide binding sites on the rho hexamer are noncatalytic. These sites are capable of hydrolyzing ATP, but the ATPase turnover at the noncatalytic sites is slow. The hydrolysis products (ADP and P i ) remain bound tightly at these sites and dissociate at a slow rate, relative to the ATPase turnover rate at the catalytic sites. Thus, the three tight ATP binding sites on rho do not participate in the fast ATPase turnover. The chemical nature of rho-bound nucleotides in the presence of poly(C) RNA was determined. The nitrocellulose membranes from the experiment described in Fig. 2A were acid-extracted and the eluted nucleotides were analyzed by PEI-cellulose TLC (see "Materials and Methods"). The PhosphorImager scan of the TLC plate is shown. Lane 1 shows the filter-extracted rho-bound nucleotides before chase was added: ATP, 4% and ADP, 96%. Lanes 2-7 show the filter-extracted nucleotides after 10, 50, 90, 150, 250, and 400 s of chase addition. In all the lanes, less than 10% of the total nucleotide was ATP, and the nucleotide ADP was the major rho-bound species (Ն90%). Lane 8 shows the nucleotides bound nonspecifically to the nitrocellulose membrane. No nucleotides were detected. Lane 9 shows the composition of nucleotides in the nucleotide binding/exchange reaction before nonradioactive ATP chase was added. This control shows that before chase was added the reaction contained 63% ATP and 37% ADP. Lane 10 shows the composition of nucleotides in a reaction in which ATP was allowed to completely hydrolyze. Most of the nucleotide is ADP. Lane 11 shows a control in which radioactive ATP without rho was spotted on the membrane, which was acid-extracted. This procedure resulted in the isolation of 89% ATP. Because the commercial radioactive ATP contained about 4 -6% ADP, very little ATP was hydrolyzed by the acid extraction procedure. was then added, and after varying chase times (10 -400 s), aliquots were removed and rho-bound radioactivity was determined by the nitrocellulose membrane binding assay. A total of 1.37 Ϯ 0.21 sites of rho hexamer had radioactivity bound in the presence of poly(C) RNA (q) before chase was added. After ATP chase was added, the radioactivity bound to rho decreased with a rate constant of 0.026 Ϯ 0.0015 s Ϫ1 (E). In the absence of RNA, 4.28 Ϯ 0.44 sites of rho hexamer had radioactivity bound (f) before chase was added. After addition of the chase, the radioactivity decreased at a rate that was too fast to measure (Ⅺ). Asterisk indicates 32 P-labeled nucleotide.
The possibility of rho protein containing noncatalytic sites was considered previously by Stitt (19), but no evidence for such sites was obtained. In these previously reported experiments, rho protein was mixed with radioactive ATP, and the rho-ATP complex was chased with nonradioactive ATP and RNA. Three radioactive ATPs were bound to rho before chase was added, and all of them were hydrolyzed upon addition of RNA. Based on these results, it was concluded that the three nucleotides were bound at the catalytic sites. Our results are consistent with the results of Stitt (19). We have also observed hydrolysis of ATP at these tight sites. However, when we measured the dissociation rate of the tightly bound nucleotides, we observed a slow dissociation of the ADP and P i species bound to these sites. Therefore, the three tight sites on rho cannot participate in the fast ATPase turnover and thus cannot be the catalytic sites.
The exchange of nucleotides at the noncatalytic sites is slow only in the presence of RNA and when excess ATP is present in solution. In the absence of RNA or when ATP solution was hydrolyzed to ADP and P i , the nucleotide exchange rate was faster. As long as the catalytic sites are actively binding/hydrolyzing ATP, the noncatalytic sites show stable binding of ADP and P i . This suggests cooperation between the noncatalytic and the catalytic sites of rho. The rate of ATPase turnover at the noncatalytic sites also depends on the identity and the concentration of the RNA cofactor. The nucleotide dissociation rate constant was smallest from the rho-poly(C) complex, intermediate from the rho-poly(U) RNA complex, and fastest from rho uncomplexed with the RNA. The exact reason for this behavior is not known. If the nucleotides are bound at the interface of the rho subunits, as found in the F 1 -ATPase protein, then one might be able to provide a rationale for the above behavior, in terms of rho hexamer stability. A slower rate of nucleotide dissociation may occur from a more stable rho hexamer. It is known that poly(C) RNA binding to rho results in a stable rho hexamer (12,27). In a stable hexamer, breathing or disruption of the subunit interfaces may be slow or occur less frequently, and thus nucleotide dissociation will be slower. This is consistent with the observation that nucleotides dissociate more slowly from poly(C) RNA-bound hexamer.
If the three tight sites on rho are noncatalytic, then the weak sites that are not detectable by filter binding or ultrafiltration methods must be the catalytic sites. By performing ATP binding experiments at higher concentrations of ATP, Geiselmann et al. (20) were able to detect 2-3 additional ATP binding sites. Similarly, we have observed 5-6 ADP⅐AlF 4 Ϫ species bound to rho in the presence of RNA (data not shown). Geiselmann et al. noted that the equilibrium dissociation constant of the three weak ATP binding sites (K d ϳ 10 M) was the same as the enzymatic K m of the ATPase activity (about 10 M). Furthermore, the three tight sites had a 10-fold tighter K d value relative to the K m value (20). These findings support the hypothesis that the weak ATP binding sites, those sites that hydrolyze and turn over ATP at a rapid rate, are the catalytic sites. A recent report has shown that binding and hydrolysis of ATP at the catalytic sites is sequential, and only one ATP is hydrolyzed at a fast rate (24). This implies that at any given time there is only one catalytic site that has a tightly bound nucleotide, and ATP binding at the other two catalytic sites is weak, because ATP exchange at those sites is rapid.
Noncatalytic nucleotide binding sites appear to be general to the hexameric helicases, because such sites are now found in at least two hexameric helicases, rho and bacteriophage T7 DNA helicase. Experiments similar to the ones reported in this paper were used to show that the three tight nucleotide binding sites of T7 DNA helicase hexamer are noncatalytic (23). In T7 DNA helicase, these sites contained dTTP and some dTDP, and these nucleotides exchanged with a rate constant that was much slower than the dTTPase turnover rate. Such sites are well established in the hexameric F 1 -ATPase protein, which has a high degree of structural and amino acid sequence homology to rho. The noncatalytic sites of the F 1 -ATPase protein bind ATP, and these ATPs are not hydrolyzed under the conditions of the experiments (28,29). In the rho protein, the ATPs at the noncatalytic sites are hydrolyzed to ADP and P i , but the hydrolysis products remain tightly bound and their release into solution occurs very slowly. Thus, the noncatalytic sites on the hexameric helicases do show differences from such sites in the F 1 -ATPase protein. This is understandable because the structures of these proteins are different. Rho and T7 DNA helicases are homohexamers and F 1 -ATPase is an ␣ 3 ␤ 3 heterohexamer. The ␣ subunits of F 1 -ATPase are designed to be noncatalytic. That is, the ATP binding site on the ␣ subunit is lined with amino acids that facilitate tight ATP binding, but some critical amino acids that promote efficient hydrolysis of ATP are missing.
In rho, the different rates of ATPase turnovers at the noncatalytic versus the catalytic sites indicate different conformations of these subunits. These conformational changes do not have to be large to be significant. Even small changes in the FIG. 5. Dissociation kinetics of rho-bound ␣-32 P-labeled nucleotides in the presence and absence of the ATP regeneration system. Rho (1 M, hexamer) was mixed with [␣-32 P]ATP (0.3 mM) at 18°C for 10 s. Subsequently, poly(C) was added and the mixture was further incubated for 120 s in the presence or in the absence of the ATP regeneration system containing phosphocreatine (10 mM) and creatine kinase (0.8 mg/ml). Nonradioactive ATP (5 mM) chase was added, and after varying chase times (10 -300 s), aliquots were removed and assayed for rho-bound nucleotides. In the presence of the ATP regeneration system, a total of 2.45 nucleotides were bound per rho hexamer. After ATP chase was added, these rho-bound nucleotides dissociated with a rate constant of 0.015 Ϯ 0.003 s Ϫ1 (q). In the absence of the ATP regeneration system, a total of 1.6 nucleotides per rho hexamer were bound. After ATP chase was added, all the rho-bound nucleotides dissociated at a fast rate (E). The inset shows the molar amount of ADP present in the reaction mixture after various periods of incubation with rho and RNA in the presence (q) and in the absence (E) of the ATP regeneration system. Asterisk indicates 32 P-labeled nucleotide. position of the amino acids at the active sites can be enough to change the microenvironment of the ATP binding sites and to affect binding and hydrolysis of ATP. Because rho is a homohexamer, the asymmetry may be induced upon oligomerization or upon ATP binding/hydrolysis. Based on the similarity of rho with the F 1 -ATPase protein, we propose that rho hexamer may adopt a C 3 or a pseudo-C 6 symmetry. Such an intrinsic symmetry (C 3 or C 3/6 ) has been proposed by Horiguchi et al. (18) for the rho protein using chemical cross-linking experiments. In the F 1 -ATPase protein, the two classes of sites, catalytic and noncatalytic, are located on alternating subunits or interfaces between the subunits. This may be the case with the hexameric helicases as well. The exact role of the noncatalytic sites in the hexameric helicases or the F 1 -ATPase is not clear at present. In helicases, nucleotide binding at these sites may be necessary for the stabilization of the quaternary structure of the helicase, or these sites may play a regulatory function. Alternatively, these subunits may be involved in nucleic acid binding. There are many questions regarding the role of these sites that will need further investigation.