Enzymatic Mechanism of RNA Translocation in Double-stranded RNA Bacteriophages*

Many complex viruses acquire their genome by active packaging into a viral precursor particle called a procapsid. Packaging is performed by a viral portal complex, which couples ATP hydrolysis to translocation of nucleic acid into the procapsid. The packaging process has been studied for a variety of viruses, but the mechanism of the associated ATPase remains elusive. In this study, the mechanism of RNA translocation in double-stranded RNA bacteriophages is characterized using rapid kinetic analyses. The portal complex of bacteriophage (cid:1) 8 is a hexamer of protein P4, which exhibits nucleotide triphosphatase activity. The kinetics of ATP binding reveals a two-step process: an initial, fast, sec-ond-order association, followed by a slower, first-order phase. The slower phase exhibits a high activation energy and has been assigned to a conformational change. ATP binding becomes cooperative in the presence of RNA. Steady-state kinetics of ATP hydrolysis, which proceeds only in the presence of RNA, also exhibits cooperativity. On the other hand, ADP release is fast and RNA-independent. The steady-state rate of hydrolysis increases with the length of the RNA substrate indicating processive translocation. Raman spectroscopy reveals that RNA binds to P4 via the phosphate backbone. The ATP-induced conformational change affects the backbone of the bound RNA but leaves the protein secondary structure unchanged.

Genome encapsidation in many viruses is realized through energy-dependent packaging into a preformed capsid (procapsid). This requires molecular motors (portal complexes) converting chemical energy (ATP hydrolysis) into mechanical work (translocation). A wealth of structural and functional data has been obtained for portal complexes of dsDNA 1 bacteriophages (e.g. 29, , SPP1, P22, T4, and T-odd phages (1)(2)(3)(4)(5)(6), but the molecular basis of the mechanochemical coupling is not understood. This is partly because of the complexity of portal complexes, in which the packaging ATPase is a multifunctional enzyme and associates transiently with the packaging machinery (i.e. the ATPase is a non-structural protein) (4). In particular, very little is known about the enzymatic action of these ATPases during packaging (7)(8)(9).
Cystoviruses have a tripartite dsRNA genome. The procapsid-associated P4 sequentially translocates the three ssRNA genomic precursors into the particle interior (18 -20). The packaged RNA precursors are then replicated inside the viral particle into dsRNA genome segments (21). Although the procapsid-associated packaging is highly specific for the viral RNA, an unspecific, in vitro translocase activity was recently demonstrated for isolated P4 hexamers from bacteriophages 8 and 13 (13). The translocation is unidirectional (5Ј to 3Ј) and leads to efficient helicase activity (13).
Given the hexameric ring morphology, common ␣/␤-fold, and helicase activity, it was concluded that P4 proteins share structural and functional similarity to hexameric helicases (13). In particular, P4 proteins share limited sequence similarity with the bacteriophage T7 gene 4 product, which is a hexameric helicase of known three-dimensional structure (22). Based on the structure, a model for sequential ATP hydrolysis was proposed (22): two opposite binding sites are occupied by ATP molecules, two binding sites contain ADPs and the two remaining sites are empty. Subsequently, the ATP molecules are hydrolyzed, ADP is released, and the empty sites become occupied by ATP. Binding, hydrolysis, and release of nucleotides induces conformational changes and modulates subunit affinity for DNA. However, the x-ray structure was obtained for a 2-fold symmetric form of the hexamer without a bound ssDNA substrate. Because ssDNA lacks such symmetry, the P4 conformation and nucleotide stoichiometry in the presence of ssDNA may differ from the solved x-ray structure (23)(24)(25). In accordance with this view, a DNA substrate has been shown to induce cooperativity in a replicative hexameric helicase, DnaB, from Escherichia coli (26). Although considerable structural and biochemical evidence has been gathered for hexameric helicases (26 -28), the role of the nucleic acid substrate in coordinating ATP hydrolysis has not been clarified. It remains to be seen whether the apparent similarity between hexameric helicases and packaging ATPases of dsRNA bacteriophages extends also to enzymatic and mechanochemical functions.
In this study, we exploit the in vitro RNA translocation system of bacteriophage 8 as a model and present the first detailed enzymatic characterization of an RNA translocation reaction. The 8 P4 system offers two advantages: 1) P4 exhibits tight coupling between ATP hydrolysis and RNA translocation; 2) P4 hexamer is stable in the presence of magnesium alone without the need for nucleotide di-or triphosphates (13). We characterize the steps and corresponding rates of ATP binding, hydrolysis, and product release. Conformational changes during the enzymatic cycle and upon RNA binding are also characterized. We discover that RNA induces cooperativity and undergoes conformational changes upon ATP binding to P4. A model of sequential ATP hydrolysis and ssRNA translocation consistent with the experimental data is presented and compared with mechanisms proposed for the hexameric helicases.

Protein Purification
8 P4 protein was expressed in E. coli and purified as previously described (17). P4 concentrations were determined by absorption at 280 nm using an extinction coefficient of 11,920 M Ϫ1 cm Ϫ1 that was calculated based on amino acid composition.
Rapid mixing of fluorescence nucleotide analogs with P4 protein was achieved in a SFM 20 stopped-flow apparatus (Bio-Logic) equipped with an F-15 flow cell (dead time 6 ms). Changes in TNP-nucleotide fluorescence intensity were detected by an MOS 250 fluorometer (Bio-Logic) using exc ϭ 408 nm (20 nm slit) and det ϭ 550 nm (20 nm slit) and concurrently by an external photomultiplier through a 515-nm edge filter.

Data Analysis
The kinetic traces of fluorescence intensity, f, were fitted by a single exponential term, where a is the amplitude of fluorescence change, b is the saturating fluorescence value, k is the apparent first-order rate, and t is time.
In the case of a simple second-order reaction, the linear dependence of the apparent rate on substrate concentration [S] was described by, k on and k off are the fitted second-order rate constants.
In the case of two-step binding the hyperbolic dependence of the apparent rate on substrate concentration was fitted by, K s is the equilibrium constant for the fast second-order binding, which is followed by a slower, first-order phase with the corresponding forward and reverse rate constants being k f and k r .

Equilibrium Measurements of Nucleotide Binding
The nucleotide dissociation constants were estimated using amplitudes of fluorescence intensity increase, ⌬F, upon TNP-ATP and TNP-ADP binding. The fluorescence spectrum of a nucleotide analog at a given concentration was measured first ( exc ϭ 408 nm, 20 nm slit, 10 nm detection slit). Then a small volume of P4 was added (final concentration 6 M) and another fluorescence spectrum was measured. The difference between these two spectra yielded ⌬F. Measurements with TNP-ATP in the presence of poly(C) were done as fast as possible (within 15 s) to minimize the effect of TNP-ATP hydrolysis. All spectra were corrected for dilution and the inner filter effect as previously described (29), where F and F obs are the corrected and the measured fluorescence intensities in arbitrary units; the term V i /V 0 accounts for the dilution; the inner filter effect because of TNP-nucleotide absorption was corrected according to Beer-Lambert's law. ⑀ is the molar extinction coefficient of TNP-nucleotide at the excitation wavelength, c is the TNPnucleotide concentration, and l is the optical path (1.5 mm).
In the case of stoichiometric binding of nucleotide analogs the fluorescence enhancement ⌬F was fitted to, where [P] is the protein concentration, [S] is the substrate concentration, and K d is the dissociation constant. The constant C describes the maximum increase in quantum yield upon binding.

Phosphate Release Measurements
Steady-state Kinetics-Phosphate release assays were done using the EnzChek Phosphate Assay Kit (Molecular Probes, Inc.) in the standard reaction buffer (0.1 M Tris-HCl, pH 7.5, 2 mM MgCl 2 , containing 0.2 mM sodium azide). The concentration of inorganic phosphate (P i ) was proportional to the absorbance at 360 nm and was calibrated by standard KH 2 PO 4 solutions. AMP was used as a background control for all experiments. The detection limit of the assay was 7 Ϯ 1 M s Ϫ1 (room temperature, 25°C). At the typical enzyme concentration of 0.1 M the detection limit was 70 s Ϫ1 , i.e. 1 order of magnitude faster than the highest k cat measured. Steady-state kinetics of phosphate release were measured using a Victor 2 plate reader (Wallac-PerkinElmer, time resolution 20 s) and a Jasco V-560 spectrophotometer (time resolution, 0.2 s). ATP and ADP were purchased from Amersham Biosciences, poly(rC) was from Sigma (concentration is expressed as moles of bases).
The rate of NTP hydrolysis (measured by phosphate release), v, is in the simplest case described by Michaelis-Menten kinetics, The activation energy, E A , for an enzymatic reaction was estimated from the temperature dependence of the rate constant, , using the Arrhenius equation, where A is a frequency factor and R is the universal gas constant.

RNA Length Dependence
SsRNA strands of lengths of 62, 700, 1410, and 2948 (full-length viral s segment) bases were prepared using T7 RNA polymerase (Promega), plasmid pLM659 containing a cDNA clone of the 6 genome s segment, and restriction enzymes BanI, BamHI, and EcoRV (New England Biolabs, Inc.) as previously described (30). The steady-state rates of ATP hydrolysis were measured as a function of RNA length according to the procedure described above. In the case of processive translocation, the rate of ATP hydrolysis, k, increases with the length of RNA substrate, L. ATP hydrolysis during translocation (with rate k tr ) is interrupted by periods of no hydrolysis between protein dissociation from the 3Ј end of one ssRNA molecule and association with another substrate. The frequency of these interruptions is proportional to 1/L. P4 may also hydrolyze ATP (with rate k end ) while it remains associated with the 3Ј end. In the limiting situation of a very long RNA substrate, the contributions of end effects and dissociation are negligible, and ATP will be hydrolyzed with an apparent rate equal to k tr . Conversely, in the case of a short RNA substrate the translocation will be negligible and all ATP will be hydrolyzed with rate k end . Thus, the dependence of the ATPase rate on the RNA substrate length can be described as a combination of two terms corresponding to the two limiting cases (where c is a param-eter corresponding to the RNA length for which end effects become negligible).

Raman Spectroscopy
Protein and RNA (poly(rC)) solutions were concentrated (to ϳ10 mg/ml each), sealed in sterile capillaries (Kimax number 34504), and thermostated at 5°C. Raman spectra were recorded on a Spex 500M spectrograph equipped with a notch filter and charge-coupled device detector (SpectrumOne, Instruments S.A., Edison, NJ) using a spectral slit width of 4 cm Ϫ1 . A solid state, diode-pumped, frequency doubled Nd:YVO 4 laser (excitation wavelength 532 nm, model Verdi, Coherent, Santa Clara, CA) was used to generate 150 milliwatts at the sample. Each spectrum represents the accumulation of 15 exposures of 2 min duration on each of two independently prepared protein solutions. Spectra were corrected for buffer and NTP contributions, and normalized for difference computations as described previously (31). The positions of strong bands are reported with Ϯ1 cm Ϫ1 accuracy, whereas those of weak and broad bands are reported with Ϯ2 cm Ϫ1 accuracy. An intensity change within a difference spectrum is deemed significant if its magnitude exceeds two times the level of noise.

Catalytic Cycle of P4 ATPase
ATP Binding Triggers Conformational Change in the Hexamer-ATP binding was investigated using fluorescence ATP analogs producing spectral changes upon binding. Several analogs were tested: TNP-ATP, MANT-ATP, and ⑀-ATP. All analogs were hydrolyzed by P4 but only TNP-ATP produced a measurable change in fluorescence intensity (20% increase). After rapid mixing of TNP-ATP and P4 protein in a stoppedflow apparatus a monoexponential increase in fluorescence intensity was observed (Fig. 1A). Such a simple kinetics profile suggested that only one type of ATP binding site could be kinetically resolved. Dependence of the apparent first-order rate, k, on TNP-ATP concentration (Fig. 1B) in the absence of RNA was hyperbolic indicating a two-step binding process, where a fast, second-order association is followed by a slower, first-order phase. The slower, first-order phase could represent a conformational change within the hexamer. This is further studied by Raman spectroscopy (see below). The binding parameters are summarized in Table I. RNA Binding Induces ATP Binding Cooperativity-The dependence of the apparent rate constant, k, on TNP-ATP concentration (Fig. 1B) was sigmoidal in the presence of poly(C). The time course of fluorescence increase remained monoexponential. Thus, the plateau regions of the concentration dependence were assigned to first-order processes, whereas the linear part of the sigmoidal curve (Fig. 1B) was interpreted as a second-order binding. The initial plateau is followed by a linear  rise and a second plateau indicating sequential, cooperative, binding of two or more ATP molecules. Upon assumption that the cooperativity results from sequential binding of two ATP molecules the observed kinetics can be qualitatively explained. At low concentrations, the kinetics are first-order indicating that the binding rate (k f (1) ϩ k r (1) ϭ 24 Ϯ 1 s Ϫ1 ) of the first ATP is limited by a conformational change. At higher concentrations binding becomes second order, suggesting that the first conformational change facilitates binding of the second ATP. Ultimately, the binding of the second ATP is limited by another conformational change as it becomes first-order (k f (2) ϩ k r (2) ϭ 81 Ϯ 3 s Ϫ1 ) at concentrations above 60 M (TNP-ATP, Fig. 1B).
The first-order step exhibited strong temperature dependence (Fig. 1B), whereas the fast second-order binding remained unaffected. Fig. 1C shows the temperature dependence of the rate of fluorescence increase upon mixing the 3 M P4 protein and 1 mM poly(C) with 100 M TNP-ATP (saturating concentration, e.g. the apparent rate corresponded to k f ). The activation energy of the first-order phase 63 Ϯ 9 kJ/mol was estimated from an Arrhenius plot (Fig. 1C, inset).
Equilibrium Nucleotide Binding-Equilibrium dissociation constants were estimated using the stationary amplitudes of the fluorescence intensity increase, ⌬F, upon nucleotide binding (Fig. 2). No cooperativity was observed for TNP-ATP binding in the presence of poly(C) in the TNP-ATP concentration region 0 to 50 M, in which the kinetic and catalytic (see below) cooperativity was apparent.
The dissociation constants were obtained from fitting the data to Equation 5, which was derived with an assumption of stoichiometric occupancy, i.e. each subunit binds one nucleotide. In principle, only one or a few subunits per hexamer may be occupied by the same nucleotide. In that case [P] should be replaced by apparent concentration of the binding sites [A] ϭ r[P]/6, where r is the number of nucleotides per hexamer. We assumed r ϭ 6 for all cases to calculate the apparent dissociation constant of TNP-ATP and TNP-ADP (Table II). Comparison of dissociation constants suggests lower affinity of P4 for ADP than for ATP.
Cooperativity of ATP Hydrolysis-The steady-state rate of ATP hydrolysis at room temperature was measured as the rate of phosphate release after addition of ATP. No ATP hydrolysis was detected in the absence of nucleic acid or in the presence of DNA (Fig. 3A). ATP hydrolysis was detected only in the presence of ssRNA (poly(C) or 20-mer RNA oligonucleotide) (Fig.  3A). The steady-state ATPase rate had sigmoidal concentration dependence and kinetic parameters were extracted by fitting the data to Equation 7 (Table III). The fit indicates cooperativity of ATP hydrolysis involving at least two subunits (n Ͼ 1). This suggests that at least two ATP molecules have to be bound before one of them is hydrolyzed.
Because P4 hexamer is stable (13) the other possible source of cooperativity would be stacking of two or more hexamers into a larger oligomer as observed in negative stain electron micrographs for 6 P4 (15). If the cooperativity resulted from such an oligomer then the Hill coefficient should be dependent on P4 concentration. This is not the case (data not shown), i.e. only one P4 hexamer participates in hydrolysis and the cooperativity results from interactions within the same ring.
P4 hydrolyzes also the fluorescence ATP analog, TNP-ATP, although with lower k cat . The apparent K M constant was lower, suggesting tighter binding of TNP-ATP than ATP but cooperativity and tight coupling between hydrolysis and RNA binding was conserved. Thus, TNP-ATP constitutes a good model for kinetic analysis.
The rate of ATP hydrolysis increases with temperature ( Fig.  3B). Apparent activation energies were estimated to be 36 Ϯ 3 kJ/mol for ATP and 58 Ϯ 8 kJ/mol for TNP-ATP hydrolysis, respectively.
Processive Translocation-The maximum rate of ATP hydrolysis depended on the length of the ssRNA (Figs. 3A and 4) indicating processive translocation, for which the rate-limiting step was the P4-RNA association and/or dissociation (28). The rate of ATP hydrolysis for translocation along the poly(C) substrate (an average length 3000 bases, estimated by gel electrophoresis) was higher than in the case of the full-length 6 genomic segment s (length 2948 bases) most likely because the secondary structure of the s segment (33) slowed down the translocation.
Our simple model (Equation 9) matched the experimental data well (Fig. 4) and gave the value of a processive rate k tr of 8.1 Ϯ 1.7 s Ϫ1 (k end ϭ 0.5 Ϯ 0.2 s Ϫ1 , C ϭ 1990 Ϯ 950). This value is close to the rate of ATP hydrolysis in the presence of poly(C) (k ϭ 6.2 Ϯ 0.6 s Ϫ1 ). Thus, poly(C) is a good model for an infinitely long RNA strand without any secondary structure.
ADP and Inorganic Phosphate Release Are Fast-The fluorescence intensity change upon binding of 30 M TNP-ADP to 3 M P4 was faster than the detection limit of 6 ms (Fig. 1A) irrespective of the presence of poly(C). The value of the TNP-  ADP dissociation constant 28 Ϯ 4 s Ϫ1 (see above) and Equation 2 yield a limit for k off of Ͼ170 s Ϫ1 . The same result was obtained also for 30 M TNP-ADP in 50 mM phosphate buffer. Thus, the release of products (ADP and phosphate) appears to be faster than all other steps of the ATP hydrolysis cycle. Another control experiment was performed to demonstrate that the phosphate release step was not coupled to RNA binding. P4 protein was incubated in an excess of ATP and the unbound nucleotides were removed by a spin column. Subsequently, poly(C) was added and phosphate release was measured. If the phosphate release step was RNA-dependent a burst of inorganic phosphate should be concomitant with poly(C) addition. Instead, a slow linear increase in phosphate concentration (rate 0.03 s Ϫ1 , data not shown) was observed. This slow rate is consistent with the hydrolysis of ATP that was bound to P4. Thus, the ATP hydrolysis is directly coupled to RNA binding.

RNA Binding
To study the ssRNA binding properties of P4, we employed an electrophoretic mobility shift assay (gel-shift) that detects stable nanomolar affinity complexes. P4 protein was incubated with a 32 P-labeled RNA probe with or without nucleotides and the RNA-protein complexes were analyzed by non-denaturing PAGE (Fig. 5A). The strongest complex was formed in the absence of nucleotides, followed by the AMP-PNP complex. No stable complexes were detected in the presence of ADP or ATP.
Application of the gel shift method in the presence of ATP is complicated by active translocation along the RNA and dissociation from the short substrate. Therefore we have estimated the apparent affinity for RNA from the stimulation of ATPase activity by increasing amounts of poly(C) using the phosphate release assay (Fig. 5B). The magnitude of the stimulation was taken as a measure of RNA binding and the apparent dissociation constant estimated for poly(C) binding (using Equation 6) was 2.2 Ϯ 0.2 M (measured in concentration of bases). This value was independent of protein concentration in the range 0.05-0.2 M, showing that the measured curve represent a saturation binding titration and not merely a titration of the P4 protein with poly(C). These results indicate that P4 has apparent submicromolar affinity for poly(C) in the presence of ATP.

Structural Changes Associated with RNA Binding and ATP Hydrolysis
Secondary Structure of P4 Is Invariant to Nucleotide Binding and Hydrolysis-Raman difference spectroscopy has been applied previously to study conformational changes in 6 P4 upon nucleotide binding and triphosphate hydrolysis (12). We have applied a similar analysis to 8 P4 (Fig. 6A) to provide further insight into the nature of the conformational change detected upon ATP binding. The absence of any significant difference in the amide I region of Fig. 6A demonstrates that ATP binding (trace III), hydrolysis (trace V), and release (trace IV) do not lead to significant changes in protein secondary structure. The  level of noise in the amide I difference spectrum sets the limit to the number of residues that may undergo structural changes without being detected. In the case of amide I differences in Fig. 6A the noise level is at 1% of the parent peak intensity, which in turn limits the number of residues with changed but undetected peptide backbone conformation to four.
RNA Backbone Conformation Is Affected by ATP Binding-There was no detectable change in fluorescence upon mixing the P4-TNP-ATP complex with poly(C) (data not shown). This suggests that poly(C) binding does not cause any significant conformational change in the nucleotide binding site. The effect of ssRNA binding was also studied using Raman difference spectroscopy, which permits resolving changes in protein and nucleic acid conformations and thus constitutes an excellent tool to study nucleo-protein complexes in solution (34). From the absence of significant differences in the amide I region it is clear that the protein secondary structure is not appreciably perturbed upon poly(C) binding (Fig. 6B). The noise level in the difference spectra indicates that less than five residues may partake in protein backbone rearrangement.
On the other hand, the Raman bands of wavenumber region 780 -830 cm Ϫ1 (Fig. 6C), which reflects the secondary structure of RNA (35,36), exhibits significant changes assignable to conformational perturbations of the ribosylphosphate backbone (37). AMP-PNP binding induces a decrease in the intensity of the 807 cm Ϫ1 band and an increase in the intensity of the 790 cm Ϫ1 band (Fig. 6C, trace III), signaling a conversion of ordered (A-form) phosphodiester moieties to an alternative state (37). The data are consistent with an order 3 disorder transition localized in the RNA backbone. The magnitude of the spectral changes suggests that ϳ20% of the RNA residues are affected. Qualitatively similar but quantitatively lesser perturbations (Ͻ5%) are also observed in the NTP-free complex (trace  6. Analysis of P4 complexes with nucleotides and RNA by Raman spectroscopy. A, comparison of the Raman amide I region corresponding to different nucleotide-bound states of P4. Protein concentration was 23 mg/ml, nucleotide concentrations were 5 mM. Trace I corresponds to the P4⅐AMP-PNP complex. II, AMP-PNP spectrum that has a negligible contribution within the amide I region (the same holds also for ADP, which is not shown). Traces III-V are normalized differences between various nucleotide states of P4: III, P4⅐AMP-PNP⅐P4; IV, P4⅐ADP⅐P4; V, P4⅐AMP-PNP-P4⅐ADP. B, binding of poly(C) to P4, amide I region. Protein concentration was 11 mg/ml; RNA concentration, 2.2 mM (corresponds to 40 bases per P4 hexamer). Trace I, P4⅐poly(C) complex, which is stable in the absence of nucleotides. II-III, difference spectra between RNA-bound and RNA-free P4 in the nucleotide-free (II) and AMP-PNP (III) states. The difference spectra were corrected for the poly(C) and AMP-PNP contributions. C, binding of poly(C) to P4, RNA structure fingerprint region. Other conditions are as in B except that difference traces II and III were multiplied by a factor of 2 with respect to the parent spectrum (trace I).
II). In addition, RNA binding to NTP-free P4 also produces a Raman difference band at 1099 cm Ϫ1 , which is assigned to the phosphodioxy moiety (38,39). The 1099 cm Ϫ1 feature indicates a strong electrostatic interaction between protein basic groups and the RNA phosphate backbone. Thus, the ssRNA conformation is altered upon NTP binding, which may be important for triggering NTP hydrolysis and RNA translocation.
However, P4-RNA complex formation does not greatly affect base stacking, as evidenced by the largely unperturbed Raman intensities of poly(C) bands in the 1200 -1600 cm Ϫ1 interval (40). The difference at 1301 cm Ϫ1 does not coincide with any of the parent poly(C) bands and thus it likely originates from protein backbone or side chain rearrangements (34). In the absence of amide I difference (see above) the assignment to a side chain represents the likely scenario.

Structural Changes during Translocation
Subunit Movement-Two fundamentally different models for the coupling between nucleotide hydrolysis and conformational changes have been proposed. In one model, which applies to the G-protein-like family (elongation factors, Ras) of small oligomeric nucleotide triphosphatase, at least 8 residues change secondary structure from a mostly ␣-helical to an extended backbone upon NTP binding (41). In the second model, which was proposed for the hexameric helicases (22), secondary and tertiary structures of the subunits are unaffected by nucleotide exchange, whereas relative subunit orientation depends on the type of nucleotide bound. The two models can be set apart by the presence or absence of secondary structure changes that are concomitant with ATP binding and hydrolysis.
The secondary structure of 8 P4 is largely invariant (up to the detection limit of 4 peptide bonds, see "Results") to nucleotide exchange and hydrolysis. Identical results were also observed for 6 P4 (12). This indicates that the P4 proteins may share the same mechanism of hydrolysis, wherein subunits or domains move with respect to one another during translocation (rigid body movement), as proposed previously (12). This mechanism is similar to that proposed on the basis of an analysis of the high-resolution structure of the gene 4 product helicase from bacteriophage T7 (22). Given that T7 gene 4 product exhibits biochemical and structural similarities with the P4 proteins (13), it likely shares the same mechanism of translocation. Thus, we adopt and further elaborate on the T7 gene 4 product model.
Structural Role of Nucleic Acid in ATP Hydrolysis-The present results provide three lines of evidence that the nucleic acid substrate (viz. ssRNA) does not constitute a passive track, along which the P4 motor runs, but participates in coordinating ATP hydrolysis. First, ATP binding is non-cooperative in the absence of ssRNA but becomes cooperative in the presence of ssRNA. Second, although P4 binds other types of nucleic acids (dsRNA, ssDNA, and dsDNA) (13), only ssRNA triggers ATP hydrolysis. Third, the RNA backbone conformation becomes more disordered upon binding of ATP (AMP-PNP). All three findings point toward a structural role for RNA during hydrolysis: RNA is bound to several subunits concurrently, thus providing a link between the subunits of the P4 hexamer. ATP binding triggers the rigid body movement of subunits (see also kinetic analysis below) deforming the bound RNA, and the RNA-P4 complex presumably attains a strained configuration. Such a strained configuration may facilitate ATP hydrolysis.
RNA Binding Is Modulated by ATP Binding and Hydrolysis-The P4 hexamer binds nonspecifically to the sugar-phosphate backbone of RNA. The affinity of P4 for nucleic acids is highest in the nucleotide-free state and decreases with ATP binding and hydrolysis, resulting in a low affinity in the pres-ence of ADP alone. As proposed in the previous section, RNA is likely bound to several subunits at a time. Thus, ATP binding and hydrolysis may modulate the number of RNA-bound subunits per hexamer and their affinity for RNA.

Mechanistic Model
ATP Binding Triggers Subunit Movement-Because TNP-ATP was used throughout this study and is efficiently hydrolyzed by P4, we limit our discussion to the kinetic mechanism of TNP-ATP hydrolysis. A simplified kinetic mechanism of TNP-ATP hydrolysis contains at least three steps: (i) secondorder, TNP-ATP association; (ii) TNP-ATP hydrolysis; and (iii) product (TNP-ADP and P i ) dissociation. In the case of 8 P4 we have evidence for an additional step: a slower, first-order conformational change that follows the fast, second-order TNP-ATP association.
Model of coupling between ATP hydrolysis and RNA translocation. Although not known for 8 P4, in analogy to 6 P4 (12) and T7 gene 4 product (22) we show ATP bound between neighboring subunits. At the high ATP concentrations inside the cell (1 mM) all binding sites apart from one are occupied by ATP. There is a dislocation in P4 hexamer structure located at the subunits interface that is void of ATP (22). The similar "lock washer" architecture was also found for the Rho transcription terminator in its RNA-bound state (44). The rapid, second-order ATP binding is followed by a slower subunit movement within the hexameric ring. The movement leads to a partial disorder in the bound RNA and formation of the strained complex. The strained complex facilitates ATP hydrolysis and RNA release. After hydrolysis, the ADP bound subunit releases the products via a fast, diffusionlimited, process. The cycle is completed by subunit relaxation and rebinding of RNA.
The apparent dissociation constants and the first-order rates of the forward reactions are shown above the arrows, the rates of reverse reactions below the arrows, respectively (all rates and binding constants are reported in the absence of RNA, E represents P4 hexamer, T is TNP-ATP, and D is TNP-ADP). The dissociation constant estimated from the kinetics of TNP-ATP binding in the absence of RNA (Table I) is in good agreement with the equilibrium experiment (Table II).
RNA binding induces cooperativity, presumably because of sequential binding of several ATP molecules with similar equilibrium affinity (Scheme 2, E represents P4 hexamer complexed with RNA). Such binding cannot be fully resolved kinetically by the present data and we consider the cooperative binding only in qualitative terms, whereas a more quantitative description awaits further analyses. The present kinetic analysis provides evidence for sequential binding of at least two ATP molecules, both of which are followed by conformational changes (estimates for the upper limits of the forward rate constants are given in Scheme 2). Hydrolysis follows binding of the second nucleotide (the rate was estimated in the presence of poly(C), Table III). Release of ADP and phosphate remains fast.
P4 Is a Directly Coupled Motor-Using the crystal structure of the related T7 gene 4 product helicase (22) we assign the kinetic steps observed for P4 to the proposed structural changes within the P4-RNA complex (Fig. 7). It is assumed that at the high concentrations of ATP inside a cell (approximately 1 mM) all binding sites apart from one are occupied by ATP (note that the exact stoichiometry of ATP binding is not known) and the first conformational change in Scheme 2 is already over. The first step then constitutes rapid ATP binding (secondorder) followed by subunit movement (first-order) within the hexameric ring. The subunit movement has a high activation energy (63 kJ/mol) that is comparable with the activation energy for TNP-ATP hydrolysis (58 kJ/mol) and thus may involve breaking numerous non-covalent interactions within the P4-RNA complex. The movement also leads to partial disorder in the bound RNA and formation of the strained complex with lower RNA affinity. Only the strained complex proceeds with ATP hydrolysis (e.g. there is no hydrolysis in the absence of RNA) and RNA release from one subunit. After hydrolysis, products are released from the ADP-bound subunit by a fast, diffusion-limited, process. Subunit relaxation to the initial state and rebinding of RNA most likely follows ADP release.
Although the proposed model is far from quantitative and further experiments are needed to delineate a detailed kinetics scheme, it captures all essential features for nucleic acid translocation. The present model, by virtue of considering both cooperativity and the role of the nucleic acid, is a substantial elaboration on the structure-based T7 helicase mechanism (22). To the best of our knowledge, no detailed investigation of nucleotide hydrolysis by the T7 gene 4 product in the presence of DNA has been reported. A different nucleotide hydrolysis cycle involving three catalytic and three non-catalytic nucleotide binding sites was proposed for the T7 gene 4 product based on nucleotide binding and hydrolysis measurements in the absence of DNA (42,25). In contrast, P4 possesses only one type of nucleotide binding site. Similar results were also reported for the large T antigen of SV40 (43).
In conclusion, we have provided evidence that RNA binding induces cooperativity into a hexameric packaging enzyme P4 from bacteriophage 8. Subunit movements within the hexamer are associated with ATP binding and modulate affinity for RNA. We also provide the first evidence for conformational changes of the RNA backbone during translocation.