RNA Packaging Device of Double-stranded RNA Bacteriophages, Possibly as Simple as Hexamer of P4 Protein*

Genomes of complex viruses have been demonstrated, in many cases, to be packaged into preformed empty capsids (procapsids). This reaction is performed by molecular motors translocating nucleic acid against the concentration gradient at the expense of NTP hydrolysis. At present, the molecular mechanisms of packaging remain elusive due to the complex nature of packaging motors. In the case of the double-stranded RNA bacteriophage φ6 from the Cystoviridae family, packaging of single-stranded genomic precursors requires a hexameric NTPase, P4. In the present study, the purified P4 proteins from two other cystoviruses, φ8 and φ13, were characterized and compared with φ6 P4. All three proteins are hexameric, single-stranded RNA-stimulated NTPases with α/β folds. Using a direct motor assay, we found that φ8 and φ13 P4 hexamers translocate 5′ to 3′ along ssRNA, whereas the analogous activity of φ6 P4 requires association with the procapsid. This difference is explained by the intrinsically high affinity of φ8 and φ13 P4s for nucleic acids. The unidirectional translocation results in RNA helicase activity. Thus, P4 proteins of Cystoviridae exhibit extensive similarity to hexameric helicases and are simple models for studying viral packaging motor mechanisms.

Genome packaging is best understood for the dsDNA bacteriophage 29, whose portal complex serves as a packaging engine (8). The complex contains three components: 1) the head-tail connector, a dodecamer of protein p10, attached to a unique 5-fold prohead vertex; 2) a multimeric ring of packaging RNAs; and 3) a homomultimeric ATPase composed of protein p16 (6,8). The portal of 29 can translocate DNA against loads up to 57 piconewtons and therefore is one of the most powerful molecular motors in nature (9).
Portal complexes of many other DNA bacteriophages are organized similarly to 29, although the packaging RNA requirement is limited to 29-related phages (10). In addition, DNA translocation in these systems is often accompanied by the processing of genomic DNA concatemers (terminase activity) (3,11). Because the 29 p16 and terminases of others bacteriophages have been shown to catalyze NTP hydrolysis in vitro, these proteins are thought to generate the energy for the packaging reaction (12). It has been suggested that the energy is somehow transmitted from the transiently associated terminase to the packaging RNA ring and connector, which in turn effects unidirectional DNA translocation (6,13). However, the complexity of the portal structures in DNA viruses has made it difficult to elucidate the packaging mechanism experimentally.
In vitro packaging of dsRNA viruses has been accomplished only for bacteriophages 6 and 8 of the Cystoviridae family, which infect the plant pathogenic bacterium Pseudomonas syringae (14,15). All cystoviruses contain a lipid envelope and a tripartite dsRNA genome composed of S, M, and L segments.
Similarly to 29, 6 packaging can be assayed in vitro using recombinant empty particles, called procapsids (PCs) and (ϩ) sense ssRNA genome precursors s ϩ , m ϩ , and l ϩ , respectively (14). Segment s ϩ is packaged first, followed by m ϩ and l ϩ (16). The specificity of packaging is due to the pac sites located within the 5Ј-terminal ϳ300-nt regions of all three RNA precursors (14,17).
The 6 P4 forms doughnut-shaped hexamers in the presence of divalent cations and ATP or ADP, and the NTPase activity is associated only with the multimeric form. The NTPase activity is weakly stimulated by ssRNA (27). Cryoelectron microscopy reconstruction of 6 PC has localized P4 hexamers to the particle 5-fold vertices, thus constituting a symmetry mismatch (28). Our recent micrographs of 8 subviral particles show that the localization of P4 is likely to be similar for other cystoviruses (29). The symmetry mismatch is also typical for packaging motors of tailed bacteriophages, and it has been proposed to facilitate rotation of the portal during nucleic acid translocation (11).
Although P4 NTPase activity is essential for RNA packaging (24,27), it is still unknown whether P4 alone constitutes the packaging motor or if other PC proteins are also involved. In addition, the details of the mechanochemical coupling during RNA packaging in dsRNA viruses are poorly understood, a situation similar to the genome packaging in DNA viruses.
The existing RNA packaging assays are based on precursor RNase protection, cosedimentation with the PC, and replication inside the PC (14,22,24). Obviously, these assays do not permit direct demonstration of ssRNA translocation, and new approaches are needed to monitor the motor action. To address these issues, we developed a novel packaging assay based on complementary oligonucleotide displacement (COD), which is conceptually similar to helicase assays (30). Using this approach, we showed that P4 proteins, isolated from bacteriophages 8 and 13 (31,32), specifically translocate ssRNA in the 5Ј to 3Ј direction at the expense of NTP hydrolysis. Other types of nucleic acids are bound but not translocated. We also demonstrated that the packaging motors of the Cystoviridae share structural and biochemical characteristics with hexameric helicases. Because of their simplicity, these P4 proteins constitute an attractive model system for delineating the translocation mechanism and mechano-chemical coupling.
NTPase Assays-The NTPase activity of the purified P4 proteins was assayed in 20 mM Tris (pH 8.0), 75 mM NaCl, 7.5 mM MgCl 2 , 4 mM unlabeled NTP, and 0.5 Ci of the relevant [␣-32 P]NTP (3000 Ci/mmol; Amersham Biosciences). The final P4 concentration was 0.1 mg/ml. Reactions were incubated for 1 h at 37°C, and 1 l of the reactions were analyzed on polyethyleneimine-cellulose F TLC plates (Merck). Chromatograms were developed as described (26). The amount of radioactivity in the nucleotide spots was measured using a phosphor imager (Fuji BAS 1500).
The steady-state rate of NTP hydrolysis was studied as the rate of phosphate release using the EnzChek phosphate assay kit (Molecular Probes, Inc., Eugene, OR). The concentration of P i was proportional to the absorbance at 360 nm. The calibration was done using standard KH 2 PO 4 solutions. Steady-state kinetics of phosphate release was measured using a Victor 2 plate reader (Wallac-PerkinElmer Life Sciences). ATP was purchased from Amersham Biosciences, ribohomopolymers (poly(G), poly(A), poly(U), and poly(G)) were from Sigma (concentrations reported in moles of nucleotide), and E. coli rRNA (16 and 23 S) was from Roche Applied Science.
To produce the RNA1 probe, plasmid pEM21 was constructed by inserting the duplex of two complementary oligonucleotides, 5Ј-CTA-GATCGGTAACCTCGGTCAGGTAC-3Ј and 5Ј-CTGACCGAGGTTAC-CGAT-3Ј, at the XbaI-KpnI sites of the pGEM3Zf(ϩ) vector (Promega). The resultant plasmid was linearized with HindIII and transcribed with T7 RNA polymerase in the presence of [␣-32 P]UTP. All transcripts were precipitated with 3 M LiCl, reprecipitated with ethanol, and dissolved in sterile water.
COD Assays-The COD activity of PCs was assayed under the conditions optimized for RNA packaging (37). Complete and protein-deficient PCs were produced as described (24). To prepare RNA duplex substrates, unlabeled RNAs were mixed with 32 P-labeled RNA probes in hybridization buffer containing 15 mM HEPES-KOH, pH 7.5, 150 mM NaCl, and 0.75 mM EDTA. The solution was boiled for 2 min and slowly cooled to room temperature. If necessary, residual nonannealed probe was removed by passing the mixture through a Sephacryl S-300 spin column (Amersham Biosciences) equilibrated with 10 mM Tris-HCl (pH 7.5), 2 mM EDTA. COD assay mixtures (25 l) contained 50 mM Tris-HCl (pH 8.9), 2 mM dithiothreitol, 0.1 mM EDTA, 5 mM MgCl 2 , 80 mM NH 4 Ac, 6% polyethyleneglycol 4000, 1 mM ATP or its nonhydrolyzable analog AMP-PCP, 1 unit/l rRNasin (Promega), ϳ1.5 g of PCs, and 0.2 g of an RNA duplex. The reactions were usually incubated for 1-2 h at 30°C and transferred to ice. The volume was then brought to 100 l with 0.13 mg/ml yeast tRNA (Sigma) in water, followed by phenolchloroform extraction and ethanol precipitation. The samples were dissolved in Tris-EDTA and analyzed by native 7% PAGE (in standard Tris borate-EDTA buffer) at room temperature. The gels were dried and analyzed by autoradiography and/or phosphorimaging (Fuji BAS 1500).
FIG. 1. Cystoviral P4 proteins contain two highly conserved regions. Amino acid sequences of P4 NTPases from 6, 8, and 13 (NCBI accession numbers P11125, AAF63301, and AAG00445, respectively) were aligned using the ClustalW algorithm (50). The similarity plot, built up on the alignment data using a 20-amino acid sliding window, reveals two peaks (black) corresponding to the Walker A and B motifs (51). Interestingly, the two spans show detectable similarity to the corresponding regions of hexameric DNA helicase gp4 of bacteriophage T7 (NCBI accession number CAA24407). In the alignment, residues identical in all four proteins are highlighted black; similar residues are shown in gray.
For P4 proteins, the COD assay was performed in 10-l reaction mixtures containing 20 mM Tris-HCl (pH 7.5), 7.5 mM MgCl 2 , 75 mM NaCl, 5 mM nucleotide (NTP, NDP, or AMP-PNP), 0.1 g of RNA duplex, and 0.05 g of P4. The reactions were incubated at 23°C for 1 h; stopped by adding 2 l of 15% glycerol, 20 mM EDTA, 3% SDS; and analyzed by native PAGE as above.
Dynamic Light Scattering-The hydrodynamic properties of individual proteins were studied using a batch dynamic light scattering instrument (Precision Detectors) equipped with deconvolution software for correlation function analysis (33).
Raman Spectroscopy-Protein solutions were concentrated (to ϳ20 mg/ml each), sealed in sterile capillaries (Kimax 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: YVO4 laser (model Verdi em ϭ 532 nm; Coherent, Santa Clara, CA) was used to obtain 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 (38). 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.
Small Angle Neutron Scattering-8 P4 was concentrated to 4 mg/ml and dialyzed against D 2 O buffer solution (20 mM Tris-DCl, 50 mM NaCl, 7.5 mM MgCl 2 ). The sample was filtered and centrifuged (Beckman Airfuge; rotor A-95, 90,000 rpm, 10 min), and supernatant was transferred into a thermostated quartz sample cell (2-mm path, 20°C). The SANS data were collected on beam line D22 (Institute Laue Langevin; Grenoble, France) using a collimation distance of 4.4 m, neutron wavelength of 0.6 nm, and a 4-m sample-detector distance. The data were processed as described (39) to obtain an averaged and corrected scattering curve. The curve was modeled by scattering from a set of dummy atoms using the DAMMIN program (40) and imposing 6-fold symmetry.
The micrographs were scanned on a Z/I imaging Photoscan TD flat bed scanner in 12-bit gray scale mode at a raster step of 7 m, corresponding to 1.4 Å/pixel (8 P4) and 1.13 Å/pixel (13 P4) in the image. Individual particles were picked manually from the scans using EMAN (41) and analyzed using the Spider program package (42). Referencefree alignment was used for the averaging (43). No symmetry constraints were applied.

RESULTS
P4 Proteins Are Ring-shaped Hexamers with an ␣/␤ Fold-6 P4 has been previously shown to assemble into do-nut-like hexamers in the presence of nucleotides and divalent metal ions (27,28). The quaternary structures of the purified P4s were characterized by several methods (Table I, Fig. 2). The apparent molecular masses of 6, 8, and 13 P4s deduced by gel filtration and light scattering were ϳ212, ϳ202, and ϳ220 kDa, respectively (Fig. 2C), as expected for the hexameric forms (predicted masses of 210, 204.6, and 225.6 kDa, respectively).
The oligomeric status and shape of each P4 was confirmed by electron microscopy. Examination of several thousand vitrified P4 molecules revealed primarily ring-shaped molecules, indicating a preferential orientation for the particles in a vitrified water layer. Reference-free alignment and averaging images revealed hexameric rings for both 8 and 13 P4 (Fig. 2D). The inner diameter of the 8 P4 ring was ϳ3.0 nm, and the outer diameter was ϳ12.1 nm. Similarly, the 13 P4 ring exhibits an inner diameter of ϳ2.8 nm and outer diameter of ϳ12.1 nm. These data agree well with those published earlier for the 6 P4 (27, 28). The average image of 8 P4 is in good agreement with a small angle solution scattering-based model, which revealed a funnel shaped hexamer with a two-dimensional projection similar to the electron microscopy average (Fig. 2E).
We also examined the secondary structure of the three P4 proteins using Raman spectroscopy. Raman spectra have been widely employed to compare the secondary structures of viral proteins (44,45). In particular, Raman spectral intensities of the backbone amide I vibrations (range 1630 -1700 cm Ϫ1 ) corresponding to pure classes of secondary structures (i.e. ␣-helices, ␤-sheet, and undefined structures) are well documented both experimentally and theoretically (46). The amide I signatures of P4 proteins from the three cystoviruses are compared in Fig. 2B. Each exhibits a broad amide I band with an apparent maximum between 1660 and 1670 cm Ϫ1 (␤-strand) and shoulders at both higher (turns and extended structures) and lower (␣-helix) wave number values. This spectral signature is typical of ␣/␤ proteins that bind and hydrolyze nucleotides (27).
Despite the overall similarity of the subunit folds, significant differences are apparent. The P4 amide I band of 8 is much sharper and displays a distinct peak position (1663 cm Ϫ1 ) reflecting lesser ␣-helical structure and greater ␤-strand structure than the P4s of 6 and 13. On the other hand, the broader amide I band of the 13 P4 spectrum indicates a greater ␣-helical content.
P4 Proteins Are RNA-stimulated NTPases-6 P4 is known to be an unspecific NTPase (26). We assayed NTPase activities of P4s from the other two phages. The proteins were incubated for 1 h at 37°C in the presence of either purine (ATP) or pyrimidine (UTP) NTPs, and the reaction products were analyzed by TLC (Fig. 3A). It is obvious that isolated 13 P4 also possesses a readily detectable NTPase activity, hydrolyzing both types of NTPs into NDP and inorganic phosphate (Fig. 3A,  lanes 2 and 5). In comparison, hexameric 8 P4 appears to be a very poor NTPase (Fig. 3A). In order to establish nucleotide substrate specificity, we have determined the k cat and K m for the P4 proteins and nucleotides ATP and UTP ( Fig. 3B and Table II). Notably, no hydrolysis was detected for 8 P4; therefore, K m constants have not been determined. The K m value for ATP and UTP hydrolysis is similar for 6 P4, whereas 13 P4 exhibits higher affinity for UTP (Table II and Fig. 3B). However, judged by k cat /K m , both enzymes prefer the UTP substrate (Table II).
It has been reported that the NTPase activity of 6 P4 was enhanced 1.6-fold by the addition of ssRNA (27). We therefore examined the effect of ssRNA on the NTPase activities (k cat ) of the P4 proteins (Fig. 3C). A weak stimulation of the 13 P4 NTPase activity, comparable with that of the 6 P4, was observed in the presence of ribosomal RNA. Interestingly, an NTPase rate comparable with that of the 6 and 13 proteins was detected for 8 P4 in the presence of rRNA. This suggests that the 8 NTPase is strongly coupled to the presence of ssRNA. NTPase activity also became cooperative in the presence of RNA (data not shown).
To find the best ssRNA cofactor for each P4 NTPase, we analyzed the effect of synthetic polyribonucleotides on k cat (Fig.  3C). Poly(C) and poly(A) acted as the best cofactors for the 8 P4. 6 P4 was also stimulated by poly(C) and poly(A) with a preference for poly(A). On the other hand, all polynucleotides except poly(G) stimulated 13 P4. Poly(G) had an inhibitory effect.
Nucleic Acid Binding Properties of P4 Proteins-NA binding to P4 proteins was assessed by an electrophoretic mobility shift assay (gel shift), which detects stable nucleoprotein complexes (nanomolar affinity). P4 proteins from 6, 8, and 13 were incubated with long ssRNA, dsRNA, ssDNA, or dsDNA and analyzed by nondenaturing agarose gel electrophoresis (Fig.  4A). No stable complexes were detected for 6 P4. In the case of 8 and 13 proteins, slowly migrating complexes were formed with all four types of NA. The extent of the retardation increased with the P4 amount added. Notably, the ssDNA probe used in this assay was the circular genome of bacteriophage M13, which indicated that free NA ends were dispensable for P4 binding.
Since 8 and 13 P4s bind all types of NA, we studied the possible effect of dsRNA, ssDNA, and dsDNA on the NTPase activity of these proteins. Of these, only dsRNA slightly stimulated the 8 P4 NTPase, whereas no activation was detected in other combinations (Fig. 4B).
COD Assay Strategy-The current 6 packaging assays require the presence of intact PC particles. We sought to develop a more direct method for detecting RNA translocation activity of isolated as well as procapsid-associated P4 proteins. Electron microscopy reconstructions indicated that the genomic RNA might be packaged via a narrow central opening in the P4 protein (28). Given the extensive secondary structure of ssRNA precursors (17), it is likely that packaging would lead to separation of short complementary regions.
Accordingly, an RNA substrate was prepared by annealing an unlabeled 700-nt-long 5Ј-terminal fragment of the s ϩ segment (sR5 RNA) with the 32 P-labeled 66-nt-long RNA oligonucleotide (RNA1). RNA1 was designed to target the 3Ј-proximal region of sR5, ϳ300 nt downstream of the pac site. Only the middle 18 nt of RNA1 can form a duplex with the corresponding sequence (590 -607 nt) of sR5, the 5Ј and 3Ј termini forming single-stranded overhangs (Fig. 5A). If the packaging apparatus recognizes sR5/RNA1, RNA1 may be displaced in the course of sR5 translocation. The liberated RNA1 probe can be separated from the input duplex in PAGE analysis under native conditions (30).
P4 Proteins of 8 and 13 but Not 6 Can Translocate RNA Substrates-We examined the COD activity of isolated P4 proteins. For this purpose, P4 enzymes were incubated with the R5/RNA1 duplex under various conditions. In contrast to 6 P4 (Fig. 5B, lane 1), both 8 and 13 proteins displaced the RNA1 probe from the duplex in the presence of ATP (Fig. 5B, lanes 4  and 7), 8 P4 showing noticeably higher activity. No COD activity was detected when ATP was omitted or substituted with either ADP or AMP-PNP (Fig. 5B, lanes 5, 6, 8, and 9, and data not shown). Thus, isolated 8 and 13 P4 can translocate RNA. 8 P4 Translocates in the 5Ј to 3Ј Direction-To determine the direction of RNA translocation, two additional RNA duplex substrates were designed containing unlabeled RNAs (positions 72-136) (a 65-nt fragment of 6 s ϩ ) and either of the two 32 P-labeled RNA oligonucleotides, RNA2 and RNA3, complementary to the 5Ј and 3Ј regions of s (positions 72-136), respectively (Fig. 5C). Both duplexes were incubated with 8 P4, and the reaction products were separated by nondenaturing PAGE. Only the RNA3 and not the RNA2 probe was displaced from the duplex, thus indicating that P4 translocates s (positions 72-136) RNA in the 5Ј to 3Ј direction. Notably, genome precursors have been shown to enter 6 PC in the 5Ј to 3Ј direction (14).
RNA Packaging into 6 PC-No displaced radioactive probe was detected for isolated 6 P4 even after prolonged incubations (not shown), thus suggesting that 6 P4 either requires additional PC component(s) or needs to be physically associated with the particle to support RNA translocation.
When the COD activity of the WT 6 PC was assayed in the presence of ATP, ϳ20% of the RNA1 was indeed released from the sR5/RNA1 duplex (Fig. 5D, lane 3). No RNA1 bands appeared when PC particles were omitted from the reaction or when ATP was replaced by a nonhydrolyzable analog AMP-PCP (Fig. 5D, lanes 1 and 2). Prolonged incubations in the presence of ATP but not AMP-PCP led to the displacement of Ͼ40% of RNA1 (Fig. 5E).
RNA Translocation by PC Requires the Presence of P4 -Previous experiments with mutated 6 PC have shown that P4 is indispensable for RNA packaging (24). We therefore tested the COD activities of several recombinant PC variants. PC containing ϳ10% of the normal P4 amount due to the point mutation S250Q in P4 (P1P2P4*P7) displaced the RNA1 probe with an efficiency close to that of the WT PC (Fig. 6A). Furthermore, PC containing the normal amount of P4 but missing protein P2 (P1P4P7) also supported oligonucleotide displacement, albeit with 10-fold reduced efficiency compared with the WT particles. As expected, PCs missing P4 (P1P2P7) did not displace RNA1.
We also assayed the COD activity of PCs assembled in vitro (22). For this purpose, different combinations of individual PC proteins were incubated with sR5/RNA1 under conditions promoting PC assembly (22). The RNA1 probe was efficiently displaced in the mixture containing all four PC proteins, whereas other mixtures showed very little COD activity (Fig. 6D).

DISCUSSION
Many viruses encapsidate their genomes through NTP-dependent packaging into preformed capsids. This process requires a portal complex that operates as a molecular motor converting chemical energy into mechanical work. Although DNA packaging in tailed bacteriophages has been studied to a great extent, the molecular basis of mechanochemical coupling is far from being understood. This may reflect the complexity of the packaging motor in which the terminase only transiently associates with the portal protein and the consequent need to study translocation in the context of assembled capsids.
RNA translocation of dsRNA bacteriophages (Cystoviridae) seems to be a much simpler system. As demonstrated here, it can be assayed without the need to monitor RNA inclusion to the PC particles using the COD technique (Figs. 5 and 6), which can also be applied to purified packaging proteins.
Only two short amino acid spans, apparently important for NTP binding and catalysis, are noticeably conserved between 6, 8, and 13 P4 sequences (Fig. 1). Nevertheless, all three P4 proteins share a common hexameric ringlike architecture and ␣/␤ fold (Fig. 2). They differ only in the detailed shape and the amount of helical secondary structure ( Fig. 2 and Table I), indicating the presence of accessory ␣-helical domains for 13 and 6. Although the NTPase activity of all three proteins is stimulated by RNA, only the 8 and 13  proteins have nanomolar affinity for nucleic acids (Figs. 3  and 4). The high affinity for nucleic acids correlates with the ability of these proteins to translocate along RNA (Fig. 5 and Table I).
The results of this study indicate that RNA packaging in Cystoviridae is catalyzed by a single protein species, P4, which therefore represents one of the simplest nucleic acid pumps described so far. The isolated 8 P4 exhibits the properties of a tightly coupled motor hydrolyzing NTP only in the presence of RNA, whereas 6 P4 may need to be attached to the PC particle to become an RNA-dependent translocase. In line with this idea, previous Raman spectroscopy studies have shown that 6 P4 undergoes extensive conformational changes upon its binding to the PC particle (38). FIG. 5. RNA translocase activity of P4 proteins. A, schematics of the COD assay. Portal corresponds to either procapsid-associated P4 or isolated P4 hexamer. B, P4 COD activity was assayed with sR5/RNA1 in the presence of ATP or ADP or without nucleotides, as indicated. Lanes 1-3, 6 P4; lanes 4 -6, 13 P4; lanes 7-9, 8 P4. Migration of the sR5/RNA1 duplex (D) and liberated RNA1 probe (P) is shown on the right. C, directionality of translocation was assayed by COD activity of 8 P4 using specially designed substrates s (positions 72-136)/RNA2 and s (positions 72-136)/RNA3 depicted at the top of the panel. Reactions contained ATP or ADP, as indicated below the panel. D, COD activity of wild-type PC particles was assayed with RNA substrates where RNA1 probe (P) was duplexed (D) with either sR5 RNA, containing the intact packaging region of the 6 s ϩ segment (wt) or one of the three sR5 variants containing deletions in the packaging region (⌬11-23, ⌬11-32, and ⌬11-43). The reactions were carried out at 30°C for 1 h in the presence of either ATP or its nonhydrolyzable analog AMP-PCP, as indicated. Lane P, RNA1 probe only. E, time course of the COD reaction catalyzed by WT PC particles in the presence of ATP or AMP-PCP.
Although all or most of the PC 5-fold vertices seem to be occupied by P4 hexamers (18,28), a model has been suggested, where only one of the 12 vertices operates as a packaging portal (24). Consistently, particles containing only ϳ10% of P4 displace complementary oligonucleotides almost as efficiently as WT PCs (Fig. 2D). It is also notable that no specific sequence requirements were observed in the COD assays for purified 8 and 13 P4s. Because packaging in Cystoviridae requires the presence of pac sites, located at the 5Ј termini of genome precursors (14), other PC proteins (e.g. P1) are likely to be responsible for the RNA selection.
Our data suggest that cystoviral P4s are structurally and functionally related to other hexameric enzymes that can couple NTP hydrolysis with unidirectional translocation along nucleic acids and other types of mechanical work. These include replicative helicases (e.g. bacterial DnaB and phage T4 gp41), transcription termination factor Rho, cellencoded DNA pumps, and the AAA family ATPases (47). For some of these proteins, such as the DNA helicase gp4 of bacteriophage T7, a high resolution structure is available, offering a model for translocation (48). Interestingly, gp4 translocates DNA in the 5Ј to 3Ј direction, similarly to the polarity of RNA translocation by 8 P4 (Fig. 5C). Furthermore, cystoviral P4 proteins show substantial sequence homology with T7 gp4 in both of the two conserved regions (Fig.  1). This encourages us to propose that the molecular basis of P4 catalyzed RNA translocation could be similar to that suggested for T7 gp4 (48).
In conclusion, isolated P4 proteins of cystoviruses 8 and 13 function as NTP dependent RNA translocases in vitro, which argues that the packaging device of at least some viruses can be exceptionally simple and perhaps mechanistically similar to the hexameric helicases. Notably, a helicase-like protein has been implicated in genome encapsidation of the adenoassociated virus 2 (49), indicating a wider spread of this paradigm. P4 hexamers of 8 and 13 can bind all types of nucleic acids, but only ssRNA stimulates the NTPase activity and supports translocation. Thus, the nucleic acid is likely to play an active role in the coupling between NTPase activity and translocation.