Nucleoside and RNA Triphosphatase Activities of Orthoreovirus Transcriptase Cofactor μ2

The mammalian Orthoreovirus (mORV) core particle is an icosahedral multienzyme complex for viral mRNA synthesis and provides a delimited system for mechanistic studies of that process. Previous genetic results have identified the mORV (cid:1) 2 protein as a determinant of viral strain differences in the transcriptase and nucleoside triphosphatase activities of cores. New results in this report provided biochemical and genetic evidence that purified (cid:1) 2 is itself a divalent cation-dependent nucleoside triphosphatase that can remove the 5 (cid:1) (cid:2) -phosphate from RNA as well. Alanine substitutions in a putative nucleotide binding region of (cid:1) 2 abrogated both functions but did not affect the purification profile of the protein or its known associations with microtubules and mORV (cid:1) NS protein in vivo . In vitro microtubule binding by purified (cid:1) 2 was also demonstrated and not affected by the mutations. Purified (cid:1) 2 was further demonstrated to interact in vitro with the mORV RNA-dependent RNA polymerase, (cid:3) 3, and the presence of (cid:3) 3 mildly stimulated the triphosphatase activities of (cid:1) 2. These findings confirm that (cid:1) 2 is an enzymatic component of the mORV core and may contribute several possible functions to viral mRNA synthesis.

The mammalian Orthoreovirus (mORV) core particle is an icosahedral multienzyme complex for viral mRNA synthesis and provides a delimited system for mechanistic studies of that process. Previous genetic results have identified the mORV 2 protein as a determinant of viral strain differences in the transcriptase and nucleoside triphosphatase activities of cores. New results in this report provided biochemical and genetic evidence that purified 2 is itself a divalent cation-dependent nucleoside triphosphatase that can remove the 5 ␥-phosphate from RNA as well. Alanine substitutions in a putative nucleotide binding region of 2 abrogated both functions but did not affect the purification profile of the protein or its known associations with microtubules and mORV NS protein in vivo. In vitro microtubule binding by purified 2 was also demonstrated and not affected by the mutations. Purified 2 was further demonstrated to interact in vitro with the mORV RNAdependent RNA polymerase, 3, and the presence of 3 mildly stimulated the triphosphatase activities of 2. These findings confirm that 2 is an enzymatic component of the mORV core and may contribute several possible functions to viral mRNA synthesis.
The mammalian Orthoreovirus (mORV) 1 core particle is an icosahedral multienzyme complex for viral mRNA synthesis (for review, see Ref. [1][2][3]. The core is released from the infectious virion during cell entry and is also formed from newly synthesized gene products as an assembly intermediate within infected cells. In addition to the 10 centrally condensed viral double-stranded RNA genome segments cores contain five different viral proteins in defined copy numbers (Table I). Virions contain three additional viral proteins that play key roles in particle stability and cell entry. The viral mRNAs are fulllength plus-strand copies of the genome segments 1200 -3900 nucleotides in length, modified by a dimethylated cap 1 struc-ture at their 5Ј ends and not polyadenylated. The manner in which the double-stranded RNA genome segments and mRNA products interact with the transcription and capping machinery in mORV cores is a topic of interest for understanding structural and mechanistic aspects of these processes in this and other viral and cellular systems.
X-ray crystallography has recently advanced our understanding of the core. A 3.6-Å crystal structure of the 52-MDa core particle has provided an atomic model for most residues of the three major core proteins: 1 in the T ϭ 1 shell, 2 in the pentameric surface turrets, and 2 in the monomeric surface nodules (4). Not visualized in that structure are the two stoichiometrically minor proteins, 3 and 2, which are located beneath the 1 shell near the icosahedral 5-fold axes (5). To obtain an atomic model for 3, which is the viral RNA-dependent RNA polymerase (6 -8) and, thus, a functionally key element, a 2.5-Å crystal structure of recombinant 3 has been determined (9). In addition, the 3 structure has recently been fit into a 7.6-Å electron cryomicroscopy reconstruction of the mORV virion, revealing how 3 associates with the 1 shell (10). These studies leave 2 as the only core protein for which no crystal structure is yet available ( Table I). The roles of 2 in core functions are also poorly defined.
Two previous genetic studies have implicated 2 in the enzymatic activities of cores. Yin et al. (31) show that the M1 genome segment, which encodes 2, determines in vitro differences between cores of mORV strains type 1 Lang (T1L) and type 3 Dearing (T3D) in both the temperature optimum of transcription and the amounts of transcripts produced. Because 3 is the viral polymerase (6 -9), these genetic findings suggest an auxiliary role for 2 in 3 function. Noble and Nibert (24) show that M1/2 determines in vitro differences between T1L and T3D cores in both the response of the ATPase activities to temperature and the rate of GTP hydrolysis at certain conditions. The latter genetic associations are consistent with 2 functioning as an NTPase or playing a regulatory role in the NTPase function of another core protein, most likely 1 (23,24,30). Both 2 and 1 in vitro have also been shown to bind RNA (32)(33)(34). Thus, questions remain about the relative roles of 2 and 1 in the NTPase activities of cores and the specific roles each plays in viral mRNA synthesis.
Recent studies have used immunofluorescence microscopy to identify in vivo associations between 2 and other proteins. Initially, 2 was shown to associate with and stabilize microtubules in both infected and transfected cells (35). Subsequently, 2 was shown to associate also with the major mORV nonstructural protein NS (36). These findings suggest important roles for 2 and NS in building the cytoplasmic "factories" in which viral genome replication and assembly are thought to occur (35)(36)(37)(38).
To learn more about the activities of 2, we devised a purification protocol and undertook functional studies of the purified protein. Results in this report provide biochemical and genetic evidence that 2 is itself a divalent cation-dependent NTPase and RTPase. Alanine substitutions in a putative nucleotide binding region of 2 abrogated both functions but did not substantially affect the purification profile of the protein or its known associations with microtubules and NS protein in vivo. In vitro microtubule binding by purified 2 was also demonstrated and not affected by the mutations. Purified 2 was further demonstrated to interact in vitro with the viral polymerase, 3, and the presence of 3 mildly stimulated the triphosphatase activities of 2. These findings suggest that 2 is an enzymatic component of the mORV core and may contribute several possible functions to viral mRNA synthesis.

EXPERIMENTAL PROCEDURES
Generation of a Recombinant Baculovirus for Expression of Wild-type (wt) 2-Generation of plasmid pBS-M1(T1L), encoding wt T1L 2 protein, was previously described (35). The complete 2-encoding region was excised from this plasmid by digestion with BamHI and HindIII and ligated into the pFastBac vector (Invitrogen) under transcriptional control of the baculovirus (Autographa californica nuclear polyhedrosis virus) polyhedrin promoter. The resulting plasmid pFB-M1(T1L) was used to generate a recombinant baculovirus with the Bac-to-Bac system (Invitrogen). This baculovirus was amplified into high titer stocks by serial passage in Spodoptera frugiperda 21 cells (Sf21 cells).
Generation of a Recombinant Baculovirus for Expression of Mutant K415A/K419A 2-To obtain an M1 gene encoding alanine substitutions for lysines 415 and 419 in 2, we first removed the BamHI site from the multiple-cloning region of pBS-M1(T1L) by cutting with BamHI, filling the overhangs with Klenow polymerase, and religating.
The resulting plasmid pBSЈ-M1(T1L) was used as template in an inverse polymerase chain reaction with forward primer 5Ј-GGGGATC-CTTTGCATCAACAATTATGAGAGTTC-3Ј and reverse primer 5Ј-GGGGATCCCGCAGGCAGCACAGCGCC-3Ј (bold italics indicate nucleotide changes for Ala-419 and Ala-415, respectively). In each primer one silent nucleotide change was also introduced to generate a BamHI site (underlined) without affecting the 2 amino acid sequence. The amplification conditions were the same as those described in Kim et al. (39). The resulting plasmid was subjected to nucleotide sequencing to confirm that between the BsmI and SphI sites in M1, and the BsmI-SphI fragment was then swapped for the same region of pBS-M1(T1L). The XbaI-HindIII fragment from the resulting plasmid pBS-M1(T1L)K415A/K419A was swapped for the same region of pFB-M1(T1L). The resulting plasmid pFB-M1(T1L)K415A/K419A was used to generate a recombinant baculovirus as indicated in the preceding section.
Cloning M1 Genes into the pCI-neo Vector-Generation of plasmid pCI-M1(T1L), encoding wt T1L 2 protein, was previously described (35). The plasmid pCI-M1(T1L)K415A/K419A was generated in the same manner in that the M1 gene from pBS-M1(T1L)K415A/K419A was excised by digestion with SpeI and XhoI and ligated into the pCI-neo vector (Promega) that had been digested with NheI and SalI.
Purification of wt and K415A/K419A 2-Sf21 cells were prepared in a 1-liter spinner at a concentration of 1 ϫ 10 6 cells/ml, into which the recombinant baculovirus to express either wt or K415A/K419A 2 was inoculated at 5-10 plaque-forming units/cell. At 60 h post-infection, the cells were harvested and washed 3 times with cold phosphate-buffered saline (137 mM NaCl, 3 mM KCl, 8 mM Na 2 HPO 4 , 1 mM KH 2 PO 4 , pH 7.5). Nuclear extracts were then prepared from these cells as follows. The cell pellet was washed twice with 50 ml of ice-cold Nonidet P-40 buffer and then twice with 50 ml of ice-cold CaCl 2 buffer. To make the Nonidet P-40 and CaCl 2 buffers 100 ml of Chelsky buffer (10 mM Tris, pH 7.0, 10 mM NaCl, 3 mM MgCl 2 , 30 mM sucrose) was supplemented with Nonidet P-40 to 0.5% or CaCl 2 to 10 mM. The nuclear pellet was resuspended in 50 ml of lysis buffer (20 mM Tris, pH 7.9, 100 mM KCl, 0.2 mM EDTA, 20% glycerol, 1ϫ protease inhibitor mixture (Roche Diagnostics)). Nuclei were lysed on ice with a syringe, and the nucleoplasm was separated from the postnuclear pellet by centrifugation at 15,000 rpm for 30 min in an RC-5B centrifuge with an SS34 rotor (DuPont Sorvall). The nucleoplasm was dialyzed overnight at 4°C against A 0 buffer (20 mM Tris acetate, pH 8.5, 50 mM KCl, 5 mM MgCl 2 , 10% glycerol, 2 mM ␤-mercaptoethanol).
All subsequent column work was performed at room temperature with an Akta-FPLC system (Amersham Biosciences). At each step the column fractions containing 2 were identified by SDS-polyacrylamide gel electrophoresis and immunoblotting with 2-specific polyclonal antiserum (36) and an alkaline phosphatase-based color detection method (39). The dialyzed nucleoplasm was applied to a 5-ml HiTrap DEAE-Sepharose column (Amersham Biosciences) preequilibrated with A 0 buffer. 2 was eluted from this column at ϳ100 mM NaCl in a 0 -1 M NaCl gradient in A 0 buffer as identified. The fractions containing 2 were pooled, mixed 1:1 with A 0 buffer, and applied to a 1-ml HiTrap Blue (Cibacron Blue)-agarose column (Amersham Biosciences) preequilibrated with A 50 buffer (A 0 buffer supplemented with 50 mM NaCl). 2 eluted from this column at 0.8 -1.2 M NaCl in a 0 -2 M NaCl gradient in A 0 buffer. The fractions containing 2 were pooled, dialyzed over- night at 4°C against A 0 buffer, and applied to a 1-ml HiTrap heparin-Sepharose column (Amersham Biosciences) preequilibrated with A 50 buffer. 2 eluted from this column at ϳ200 mM NaCl in a 0 -1 M NaCl gradient in A 0 buffer. The fractions containing 2 were pooled, dialyzed against A 100 buffer (A 0 buffer supplemented with 100 mM NaCl), and used for all subsequent work as purified 2. Protein concentration was estimated from absorbance at 280 nm. Yields from this procedure were routinely in the range of 300 -400 g of 2 per 4 ϫ 10 8 cells.
Purification of wt 3 Protein-The T3D 3 protein was expressed and purified as previously described (9). Immediately before use in each experiment, 3 (or bovine serum albumin used in parallel samples) was dialyzed against A 100 buffer for 2 h at 4°C using a Slide-A-Lyzer 10K cassette (Pierce).
NTPase Assays-Reactions for analysis in the colorimetric NTPase assay were performed in A 100 buffer lacking ␤-mercaptoethanol. The reactions also contained 2 mM ATP and the indicated amount(s) of protein(s) in a total reaction volume of 60 l. For some experiments ingredients were altered as specifically described in the figure legends. After incubation at room temperature for an appropriate time (standard, 45 min), the reaction was stopped with 60 l of 10% trichloroacetic acid. 100 l of the stopped reaction was then mixed with 100 l of colorimetric reagent prepared by mixing 6 N sulfuric acid, 0.8% ammonium molybdate, and 10% ascorbic acid in a 1:3:1 ratio. This mixture was incubated at 37°C for 30 min, and absorbance at 655 nm (A 655 ) was measured with a microplate colorimeter (Molecular Devices). Standard curves generated with a dilution series of KH 2 PO 4 were used to convert the A 655 values to amounts of inorganic phosphate (P i ) released (23).
For the radiographic NTPase assay reaction conditions were the same as for the colorimetric assay except that the reaction volume was only 20 l, and 5 Ci of [␣-32 P]ATP (3000 Ci/mmol) (PerkinElmer Life Sciences) was used instead of 2 mM ATP. After incubating at room temperature for 30 min, the reaction was stopped with 10 mM EDTA. 1 l of the stopped reaction was spotted onto a polyethyleneimine-cellulose thin-layer chromatography (TLC) plate (EM Science). After drying the TLC plate was developed in 1.2 M LiCl solvent, and the separated reaction products were visualized by either phosphorimaging (Molecular Dynamics) or exposure to x-ray film (Fuji or Kodak) in the presence of an intensifying screen.
RTPase Assay-MAXIscript in vitro transcription kit (Ambion) was used to generate ␥-labeled 45-nucleotide RNA substrates following the manufacturer's directions. Each 20-l reaction mixture contained 30 units of T7 RNA polymerase (from the kit), 1ϫ transcription buffer (from the kit), 1 mM ATP, 1 mM CTP, 1 mM UTP, 0.03 mM GTP, 50 Ci of either [␥-32 P]GTP (6000 Ci/mmol) or [␣-32 P]GTP (3000 Ci/mmol) (PerkinElmer), 1 g of pGEM-4Z (Promega) (linearized with SmaI and purified from an agarose gel), and 1 unit/l RNasin (Promega). After a 2-h incubation at 37°C the reaction was adjusted to 50 l with H 2 O. Template DNA was degraded by treatment for 30 min at 37°C with 2 units of DNase I (from the kit). Free nucleotides were removed with a nucleotide removal kit (Qiagen) and then a Sephadex G-25 column (Amersham Biosciences). The quality and purity of the purified 45-mer [␥-32 P]-or [␣-32 P]RNA (i.e. appropriate RNA size and essential absence of [␥-32 P]-or [␣-32 P]RNA and small abortive transcripts) were verified on 10% polyacrylamide gels containing 7 M urea (data not shown). Efficient removal of GTP and small abortive transcripts from the [␥-32 P]RNA was also verified by TLC, in which [␥-32 P]GTP was found to migrate slightly faster than the upper spot of the radiolabeled RNA (data not shown). RNA concentration was estimated by absorbance at 260 nm.
For the RTPase assay reaction conditions were the same as for the radiographic NTPase assay except that specified amounts of 32 P-labeled RNA were added instead of ATP. After incubating for the appropriate time at room temperature (standard, 1 h), the reaction was stopped with 10 mM EDTA. For Fig. 6, RNAs were resolved on a 10% polyacrylamide gel containing 7 M urea and then visualized by exposure to x-ray film in the presence of an intensifying screen. Otherwise, reaction products were separated by TLC as described for the radiographic NTPase assay and visualized by phosphorimaging. For quantitation from TLC, the intensity of the 32 P i spot obtained in each 2and/or 3-containing sample was expressed as a fraction of that obtained by treatment of the RNA with calf intestinal phosphatase (New England Biolabs). The upper spot attributable to substrate RNA in the TLC plates (see Fig. 7, A and C, for results with [␥-32 P]RNA substrate) was unexpected, suggesting that despite the purification steps and tests of purity described in the preceding section, the samples might be contaminated with GTP. However, the [␣-32 P]RNA substrate was identically resolved into two spots by TLC, the upper of which was not abolished by prior treatment of the sample with calf intestinal phos-phatase (data not shown), demonstrating that the upper spot is indeed attributable to RNA, not GTP.
Microtubule Spin-down Assay-A spin-down kit (Cytoskeleton, Inc.) was used according to the manufacturer's suggestions. Briefly, the indicated amounts of purified wt or K415A/K419A 2 were mixed with microtubules stabilized with 20 M taxol after polymerizing purified ␣and ␤-tubulin by incubation at 37°C for 20 min. After 40 min at room temperature, the mixtures were overlaid on a 100-l sucrose cushion and spun at 100,000 ϫ g for 40 min at room temperature in a TL-100 tabletop ultracentrifuge with a TLA-100 rotor (Beckman). The supernatant was removed, and the pellet was resuspended in 50 l of phosphate-buffered saline. 10 l of the resuspended pellet was then separated by SDS-polyacrylamide gel electrophoresis and analyzed by immunoblotting with 2-specific polyclonal antiserum (36) and a horseradish peroxidase-based enhanced chemiluminescence detection method (Pierce).
Immunofluorescence Microscopy-Transfection was performed using 2 g of plasmid DNA and 6 l of LipofectAMINE reagent (Invitrogen) as previously described (35). At 18 h post-transfection cells were fixed with 100% methanol and stained with either Oregon Green-conjugated 2specific polyclonal antibodies, Texas Red-conjugated NS-specific polyclonal antibodies, or both, as previously described (36). Fluorescence was visualized with a TE-300 inverted light microscope (Nikon), and the collected images were processed and optimized with Photoshop software (Adobe Systems).
Immunoprecipitation-Each of the indicated amounts of purified wt or K415A/K419A 2 was mixed with 300 ng of purified 3 in binding buffer (20 mM Tris, pH 8.5, 50 mM NaCl, 50 mM KCl, 5 mM MgCl 2 , 10% glycerol, 0.01% Triton X-100, 2 mM ␤-mercaptoethanol, 1ϫ protease inhibitor mixture). The mixtures were incubated overnight with rocking at 4°C. 2-specific polyclonal antibodies (purified from antiserum using a protein A column (36)) were added to a final concentration of 33 g/ml, and the mixtures were incubated for an additional 2 h at 4°C. Each mixture was supplemented with 10 l of protein A Dynabeads (Dynal Biotech) preequilibrated with binding buffer and then incubated for 30 min at room temperature. Bead-bound antigen-antibody complexes were isolated with the use of a magnetic stand (Dynal Biotech) and washed twice with binding buffer and twice with washing buffer (binding buffer supplemented with NaCl to 200 mM and Triton X-100 to 0.05%). The washed beads were resuspended in 50 l of 1ϫ gel sample buffer (125 mM Tris, pH 8.0, 1% SDS, 2% ␤-mercaptoethanol, 10% sucrose, 0.01% bromphenol blue), boiled, separated by SDS-polyacrylamide gel electrophoresis, and analyzed by immunoblotting with either 2or 3-specific polyclonal antiserum (36,40) and an alkaline phosphatase-based color detection method (39).

RESULTS
Purification of mORV 2 Protein-A protocol for large scale purification of 2 protein has not been previously reported, and we, therefore, set out to devise one. Recombinant 2 was expressed in Sf21 cells using a baculovirus vector in which the M1 gene was under control of the polyhedrin promoter. About half of 2 expressed in this way appeared in the nuclear fraction of lysed cells (data not shown). This was not wholly unexpected since some nuclear localization of recombinant 2 has been previously noted in mammalian cells by immunofluorescence microscopy (35,36,38), although the basis of this localization has remained unclear. The nucleoplasm (postnuclear supernatant) of 2-expressing Sf21 cells was obtained and found to contain larger amounts of 2 relative to other proteins than did the cytoplasm. The nucleoplasm was, therefore, used for further purification of 2 on a series of ion-exchange and affinity columns: DEAE, Cibacron Blue, and heparin. SDSpolyacrylamide gel electrophoresis and immunoblotting with 2-specific polyclonal antiserum were used to identify the column fractions containing 2. Fig. 1A summarizes representative gel results from the series and shows that one major protein band (M r ϳ 80,000) was strongly enriched. This band comigrated with the 2 protein from mORV cores during electrophoresis (Fig. 1B, left) and was also recognized by the 2 antiserum in immunoblots (Fig. 1B, right). Another band sometimes seen migrating above 2 (Fig. 1, B, left, and C) was a contaminant found in different amounts in different prepara-tions; it was not recognized by the 2 antiserum. A minor band migrating below 2 (M r 50,000 -60,000) but recognized by the 2 antiserum was seen in samples of both cores and purified protein (Fig. 1B), suggesting it is a common degradation product of 2. Preliminary evidence from dynamic light scattering and gel filtration chromatography suggested the purified form of 2 is most likely a dimer. 2 We also found that 2 expressed from this same baculovirus vector was competent to be packaged into core-like particles when coexpressed with other core proteins. 3 NTP Hydrolysis by Purified 2 Protein-To determine whether purified 2 has ATPase activity we examined aliquots of the heparin-column fractions with a colorimetric assay that detects the P i released by ATP hydrolysis (23,24). Fig. 1, C and  D, show results of a representative experiment. The assay revealed that fraction numbers 11-13 in this experiment had a level of ATPase activity substantially above the background level of other fractions (Fig. 1D). These same fractions were shown to contain the highest amounts of 2 protein by gel analysis (Fig. 1C). Moreover, the level of ATPase activity of each fraction was proportional to the amount of 2 the fraction contained (fraction 12 Ͼ 13 Ͼ 11), strongly suggesting that 2 is required for the activity. Other colorimetric assays with the peak fractions provided evidence that 2 is able to hydrolyze GTP, CTP, and UTP as less preferred substrates (data not shown), suggesting that 2 is a nonspecific NTPase.
Mutant 2 Protein (K415A/K419A) with Alanine Substitutions in a Putative Nucleotide Binding Motif-After the preceding results we remained concerned that the NTPase activity might be attributable to a contaminant and not 2 in the purified preparation. To prove that NTP hydrolysis was mediated by 2 we sought to create a 2 mutant that lacked this activity. We have previously noted a conserved region of the 2 primary sequence with some similarities to nucleotide binding A and B motifs (24) (Fig. 2A). Comparison with recently reported sequences for the homologous VP5 proteins of Aquareovirus (41) showed these motifs are conserved despite an overall 2-VP5 homology of only 24% 4 ( Fig. 2A). Moreover, the putative A motif of Ortho-and Aquareovirus is very similar to that of Alphavirus Nsp2, a known NTPase and RTPase (28,42) (Fig.  2A). We, therefore, introduced two alanine-for-lysine substitu-FIG. 1. Purification of mORV 2 protein and analysis of its ATPase activity. Denatured proteins were resolved by electrophoresis on SDS-polyacrylamide mini-gels (10% acrylamide), which were then either stained with Coomassie Brilliant Blue R-250 or electroblotted as indicated. A, mORV 2 protein was obtained from the nucleoplasm of Sf21 cells and progressively purified over DEAE, Cibacron Blue, and heparin columns. Aliquots from the postnuclear pellet, the nucleoplasm, and the peak fractions from each column were evaluated in stained gels. Positions of size markers (kDa) (Invitrogen) are indicated at the left. B, samples of purified mORV cores and purified 2 protein were evaluated in adjacent lanes of a stained gel (left). An identical gel was prepared for immunoblot analysis with 2-specific polyclonal antiserum (right). C, aliquots of numbered fractions from the heparin column were evaluated in adjacent lanes of a stained gel. Positions of size markers (kDa) (Bio-Rad) are indicated at left. D, all fractions shown in C were analyzed for their capacity to release P i from ATP as measured by A 655 in a colorimetric assay. tions at positions 415 and 419 in the putative A motif ( Fig. 2A) in an effort to eliminate the apparent NTPase activity of 2.
The 2 mutant K415A/K419A was expressed using a baculovirus vector and purified through the same procedures as wt 2. Gels showed no mobility difference between the purified K415A/K419A and wt proteins ( Fig. 2B; data not shown). Over the series of columns in the purification, the mutant behaved almost identically to wt protein, except that K415A/K419A 2 was eluted at slightly lower salt concentration from the Cibacron Blue column (data not shown). The similar purification patterns suggest the mutant is folded and structurally similar to wt protein (see Fig. 3 and "Microtubule Binding and NS Association by wt and K415A/K419A 2" below for more evidence). The colorimetric assay revealed that the ATPase activity of purified wt 2 increased in concert with protein concentration, whereas the activity of K415A/K419A 2 remained at background level across the range of concentrations (Fig. 2C). These results indicate that 2 is itself an ATPase and suggest that its putative nucleotide binding A motif is involved in this activity. Based on the comparison with Alphavirus Nsp2 (Fig.  2A), we consider it likely that the conserved Lys-419 residue in mORV 2 is especially important. It is also interesting that the proline residue conserved in Ortho/Aquareovirus 2/VP5 and Alphavirus Nsp2 is one of two residues mutated in the temperature-sensitive 2 protein of mORV mutant tsH11.2 (43).
Microtubule Binding and NS Association by wt and K415A/ K419A 2-The preceding results with the K415A/K419A mutant would be most convincing if the effect were limited to NTP hydrolysis, whereas other activities of 2 are spared. Such findings would indicate that the K415A/K419A mutations do not globally disrupt the folding and structure of 2. The fact that K415A/K419A 2 could be purified in a similar manner to wt 2 provided some evidence in this regard, but we sought further evidence. Recent work from our laboratory has shown the capacity of 2 to associate with both microtubules and mORV NS protein in infected and transfected cells (35,36). We, therefore, tested the mutant for these activities. In transfected CV-1 cells, K415A/K419A 2 associated with microtubules similarly to wt 2 (Fig. 3A). A substantial portion of both 2 proteins localized to the nucleus (Fig. 3A), as previously noted for wt 2 (35,36). The exact role if any of this nuclear localization remains unknown. To complement the in vivo results demonstrating association between 2 and microtubules, we performed an in vitro spin-down assay with a mixture containing taxol-stabilized microtubules (44,45) and increasing concentrations of either wt or K415A/K419A 2. After proteins in the pellets were separated on SDS-polyacrylamide gels, immunoblots revealed that the K415A/K419A mutant was spun down with microtubules as well as wt 2, whereas neither was spun down in the absence of microtubules (Fig. 3B). These results provide strong evidence that 2 binds directly to microtubules and also show that the K415A/K419A mutant has a microtubule binding capacity similar to that of wt protein.
When transiently expressed in CV-1 cells, the NS protein, a major component of viral factories in mORV-infected cells (36), formed globular inclusions within the cytoplasm (Fig. 3C, bottom), as previously reported (36). When wt 2 was coexpressed with NS, on the other hand, NS localized to microtubules along with 2 (Fig. 3C, upper left), as also previously reported (36). When K415A/K419A 2 was coexpressed with NS, NS again localized to microtubules along with 2 (Fig. 3C, upper  right). In sum, these results demonstrate that K415A/K419A 2 behaves like wt 2 with regard to microtubule binding both in vivo and in vitro, nuclear localization, and interaction with NS in vivo, indicating that the two forms of 2 are functionally similar in many regards. We conclude that the K415A/ K419A mutations have little or no effect on the overall conformation of 2 but specifically abrogate its ATPase activity.
Specificity of wt 2 for Release of ␥-Phosphate from NTPs-The 2and 1-influenced NTPase activities in mORV cores are specific for triphosphorylated nucleotides and release only the ␥-phosphate (21)(22)(23)(24)30). We, therefore, tested for such specificity with the purified 2 proteins. [␣-32 P]ATP was used as substrate in a radiographic assay in which the reaction products were analyzed by TLC and fluorography. In this assay were also analyzed. Proteins in the sedimented pellets were resolved in adjacent lanes of an SDS-polyacrylamide mini-gel (10% acrylamide), which was then subjected to immunoblot analysis with 2-specific polyclonal antiserum. C, the wt 2, K415A/K419A 2, and/or wt NS proteins were expressed in cells transfected with pCIneo plasmids containing the respective M1 or M3 genes. NS was separately coexpressed with each 2 protein (top) and was also expressed in the absence of 2 (bottom). At 18 h post-transfection, cells were fixed, and 2 and NS were respectively immunostained with Oregon Green-conjugated 2-pecific polyclonal antibodies (green) and Texas Red-conjugated NS-specific polyclonal antibodies (red). wt 2 generated ADP as the only radiolabeled product (Fig. 4A,  lane 2), indicating that 2 specifically attacks the ␤-␥ phosphodiester linkage of ATP, generating ADP (labeled) and P i (not labeled) as products. As expected from previous results, K415A/K419A 2 showed no activity in this assay (Fig. 4A,  lane 3). The addition of EDTA to the reaction containing wt 2 completely abolished the ATPase activity (Fig. 4A, lane 4), indicating that the activity is dependent on divalent cations, as are many other NTPases (25)(26)(27)(28)(29) including those in mORV cores (23) (see below for additional evidence). Specific cleavage of the ␤-␥ phosphodiester linkage was further indicated by experiments in which the colorimetric assay was used to examine the capacity of wt 2 to hydrolyze GTP, GDP, or GMP and which showed that only GTP could serve for P i release (Fig.  4B).
Other Characteristics of the NTPase Activities of wt 2-NTPase activities are commonly sensitive to the type and concentration of divalent cations in the reaction mixture (25)(26)(27)(28)(29). We, therefore, tested the ATPase activity of 2 in the presence of increasing concentrations of different cations. The results demonstrated that mM concentrations of a divalent cation are necessary for the activity and that Ca 2ϩ , Mg 2ϩ , or Mn 2ϩ can each serve similarly well in this capacity (Fig. 4C). Synergistic effects of Mg 2ϩ and Mn 2ϩ (26) were not observed (data not shown). In other experiments we found that the pH optimum for ATP hydrolysis by 2 was 6.5-7.0 (Fig. 4D). However, the activity remained high, above 50% of maximum, over a broad range of pH values between ϳ5.7 and Ͼ9.0 (Fig. 4D). At higher temperatures (45 or 55°C) little activity was seen, and the activity at 25°C was higher than that at 35°C (Fig. 4D). Experiments to estimate the V max , apparent K m , and K cat values of wt 2 in hydrolysis of each NTP were performed with increasing concentrations of ATP, GTP, CTP, or UTP (Fig. 4E). The results confirmed ATP as the preferred substrate for hydrolysis and suggested this preference is based in modest increases in both substrate binding affinity and catalytic efficiency relative to GTP or CTP (Table II). UTP was a relatively poorer substrate for hydrolysis (Table II). These estimated values for the kinetic properties of 2 in NTP hydrolysis are similar to published findings for several other viral NTPases (see "Discussion").
3 Interaction with wt and K415A/K419A 2-The two stoichiometrically minor core proteins, 2 and 3, have been hypothesized to interact within cores (46 -48). Direct evidence has remained lacking, but evidence from electron cryomicroscopy has suggested that the two proteins are at least juxtaposed in the core interior (5). In addition, the genetic evidence that 2 can influence core transcriptase behaviors has sug-  4) were also analyzed. ori, sample origins. B, guanosine nucleotides (2 mM GTP, GDP, or GMP) were tested for hydrolysis by wt 2. Test samples contained 6.0 pmol of wt 2. C, indicated concentrations of different cations (chloride salts) were tested for the effects on ATP hydrolysis by wt 2. D, buffers at the indicated pH values were tested at different temperatures for effects on ATP hydrolysis by wt 2. Each buffer was made as a 5ϫ stock by mixing 100 mM Tris and 100 mM MES until the desired pH was reached. Other contents of the reaction mixtures were the same as in other experiments. E, the indicated concentrations of the different NTPs were tested for the effects on NTP hydrolysis by wt 2. Initial velocities (V 0 ) were determined from time points within the linear range of each reaction time course. In B-E, P i released from the nucleotide substrate was measured by A 655 in a colorimetric assay.  gested a 2-3 interaction (31). A protocol is available for large scale purification of 3 (8,9) and was used in this study to obtain purified 3 for investigating its possible interaction with purified 2 in vitro. In the absence of 2, 3 was not immunoprecipitated by 2-specific polyclonal antibodies (Fig. 5A, lanes  1 and 5). In the presence of increasing concentrations of either wt or K415A/K419A 2, however, the amount of 3 precipitated by the 2 antibodies was progressively increased (Fig. 5A,  lanes 2-4 and 6 -8). These results indicate that 2 can indeed interact with 3 in vitro. They also indicate that the K415A/ K419A mutations do not affect this interaction, providing further evidence for the largely intact conformation of this mutant 2.
3 Stimulation of ATPase Activity of wt 2-Having demonstrated 2-3 interaction, we hypothesized that 3 binding to 2 may stimulate the enzymatic activities of the latter. In the colorimetric assay for ATPase activity, neither 3 alone nor mixtures of K415A/K419A 2 with increasing concentrations of 3 showed activity (Fig. 5B). These findings suggested that 3 does not have intrinsic ATPase activity (that is, in the presence of ATP but no other nucleoside triphosphates or template RNA) and also that K415A/K419A 2 could not be stimulated to demonstrate this activity by its interaction with 3. On the other hand, mixtures of wt 2 with increasing concentrations of 3 exhibited increasing levels of ATPase activity (Fig. 5, B and C), suggesting that 2-3 interaction mildly stimulated this activity of wt 2. Increasing concentrations of bovine serum albumin had no effect on the ATPase activity of wt 2 (Fig. 5C), providing evidence that the stimulatory effect was specific to 3. An experiment to estimate the effects of 3 on the kinetic properties of 2 in ATP hydrolysis (data not shown) suggested that increasing amounts of 3 progressively increased the K cat of the reaction but had little effect on its apparent K m (Table  II). This last finding suggests that 3 acted to increase the catalytic efficiency of 2 and not its substrate binding affinity.
RTPase Activity of wt 2-The capacity of wt 2 to release the ␥-phosphate from an NTP led us to hypothesize that 2 may also release the 5Ј ␥-phosphate from an RNA molecule. mORV cores mediate such an RTPase activity, which yields a diphosphorylated RNA 5Ј end (11,49) for linkage to GMP by the 2-associated guanylyltransferase (4,11,14,15,17,18), as part of the mRNA capping process. The RTPase activity has been previously attributed to the 1 shell protein (12). We nevertheless considered it possible that this activity may instead be represented by the NTPase activity of 2, and we, therefore, tested purified 2 for its capacity to release the ␥-phosphate from RNA. An in vitro transcription reaction driven by T7 RNA polymerase was performed in the presence of [␥-32 P]GTP to generate RNA transcripts 45 nucleotides in length in which only the 5Ј-terminal ␥-phosphate was radiolabeled. Alternatively, the reaction was performed in the presence of [␣-32 P]GTP to generate 45-mer transcripts with the radiolabel in internal positions. The quality of these substrate RNAs and their essential lack of contamination with [␥-32 P]-or [␣-32 P]RNA were confirmed by electrophoresis in denaturing polyacrylamide gels (data not shown). After incubation in the presence of purified wt 2 followed by denaturing gel analysis, loss of radiolabel from the 45-mer RNA was observed with the [␥-32 P]GTP-labeled substrate but not the [␣-32 P]GTP-labeled substrate (Fig. 6). These results indicate that 2 indeed displays RTPase activity and not a nonspecific nuclease activity that would have degraded both substrates.
The RTPase activity of 2 was further analyzed by TLC. This assay revealed that wt 2 in the presence of increasing concentrations of [␥-32 P]GTP-labeled substrate RNA released increasing amounts of 32 P-labeled P i (Fig. 7A, lanes 3-7). The addition of EDTA abolished the ␥-phosphate release from RNA by wt 2 (Fig. 7A, lane 8), indicating that the RTPase activity is dependent on divalent cations, as is the NTPase activity. K415A/ K419A 2 showed no activity in the RTPase assay (Fig. 7A, lane 9). This last result suggests that the RTPase activity involves the same putative nucleotide binding region of 2 as the NTPase, since the K415A/K419A mutations abolished both activities. Consistent with the release of ␥-phosphate by wt 2, the amount of uncleaved RNA substrate was decreased relative to that in the RNA-alone, EDTA, or K415A/K419A 2 lane (Fig.  7A, compare lane 7 with lanes 1, 8, or 9). The decreased amount of uncleaved [␥-32 P]RNA substrate, after incubation with wt 2, was confirmed by electrophoresis in a denaturing 10% polyacrylamide gel, as was the failure of wt 2 to degrade [␣-32 P]RNA (data not shown). An experiment to estimate the kinetic properties of 2 in the RTPase reaction (Fig. 7B) suggested that its V max , apparent K m , and K cat values were all substantially lower than those for NTP hydrolysis (Table II). This is consistent with published findings for other viral proteins having both NTPase and RTPase activities (see "Discussion").
3 Stimulation of RTPase Activity of wt 2-Preceding evidence for the capacity of 3 to stimulate the ATPase activity of wt 2 (Fig. 5, B and C) led us to expect a similar effect on RTPase activity. Indeed, when [␥-32 P]RNA was used as substrate neither 3 alone (Fig. 7C, lane 5) nor mixtures of K415A/ K419A 2 with 3 (data not shown) exhibited RTPase activity, but mixtures of wt 2 with increasing concentrations of 3 exhibited mildly increasing levels of RTPase activity (Fig. 7C,  lanes 3 and 4; Fig. 7D). These findings suggest that 3 does not have intrinsic RTPase activity, that K415A/K419A 2 cannot be stimulated to demonstrate RTPase activity by its interaction with 3, and that the presence of 3 mildly stimulates the RTPase activity of wt 2.

DISCUSSION
Results in this report demonstrate that mORV 2 protein in the presence of divalent cations can catalyze removal of the ␥-phosphate from either an NTP or the 5Ј end of an RNA. When a purified protein is shown to exhibit a new enzymatic activity, it is always possible that the activity is attributable to a contaminant in the preparation and not the protein of interest. To rule out that possibility we introduced alanine substitutions in 2 that we expected to reduce its NTPase and/or RTPase activities because the substitutions were located in a putative nucleotide binding region of 2. These substitutions did not substantially affect the purification profile of the protein but did abolish its capacity for ␥-phosphate removal from either type of substrate. Thus, both activities are attributable to the wt 2 protein. Moreover, both activities involve the same putative nucleotide binding region of 2 in which the alanine substitutions were located. An asparagine substitution for Lys-192 in Semliki Forest virus (Alphavirus) Nsp2 (see Fig. 2A) has been shown to abrogate both its NTPase and its RTPase activities (28). Similarly, alanine substitutions that inactivate the baculovirus and vaccinia virus RTPases also abrogate their NTPase activities (25,26,29).
Further evidence that the NTPase and RTPase activities are attributable to 2 is that 3 mildly stimulated both activities of wt but not mutant, 2. A viral component such as 3 would be unlikely to have a stimulatory effect on the activities of a cellular contaminant, but since 3 interacts with 2 (Fig. 5), its stimulatory effect on the NTPase and RTPase activities of 2 is not surprising. Because (i) 3 alone did not show these activities, (ii) no increase in activities was seen when 3 was incubated with mutant 2, and (iii) 3 interacts with both wt and mutant 2, the increase in activities when 3 was incubated with wt 2 cannot be attributed to an effect of 2 binding on latent activities of 3. The new evidence for 2-3 interaction in vitro also supports the hypothesis that 3 and 2 interact within mORV cores, as previously suggested by electron cryomicroscopy and genetic results (5,31). Such an interaction would further manifest the close juxtaposition of different transcription and capping enzymes through protein-protein contacts within viral particles (4, 5) (see Table I), as has been recently shown for analogous cellular enzymes (for review, see Ref. 50 and 51).
Our estimated values for the kinetic properties of 2 in NTP hydrolysis (Table II) are similar to published ones for several other divalent cation-dependent viral NTPases. For example, the apparent K m for ATP hydrolysis by NTPase/RNA helicase NPH-II from vaccinia virus has been reported as 1.2 mM (52), and the K cat for ATP hydrolysis by NTPase/RTPase D1 from vaccinia virus has been reported as 606 min Ϫ1 (27). In addition, we found apparent K m and K cat values for the triphosphatase activities of 2 to be much lower for RNA than for NTP substrates (Table II), and this trend has also been seen with other viral proteins. For example, the apparent K m for the RTPase activity of vaccinia virus D1 has been reported as 1.0 M compared with 0.8 mM for ATP hydrolysis (27), and the K cat for the RTPase activity of Semliki Forest virus Nsp2 has been reported as 5.5 min Ϫ1 , compared with 230 min Ϫ1 for GTP hydrolysis (28).
The literature indicates that it is common for RTPases to exhibit NTPase activities (25)(26)(27)(28)(29). The reciprocal statement is harder to support, however, because NTPases have been less routinely tested for RTPase activity. We nonetheless expect that many NTPases could act as RTPases in vitro without the latter being a normal aspect of their functions in cells. Thus, our demonstration that 2 exhibits RTPase activity in vitro does not necessarily mean that it acts as an RTPase in cores. The same logic applies to the in vitro RTPase activity reported for the mORV 1 protein (12).
The findings in this report corroborate previous genetic evidence for a role of 2 in mORV core NTPase activities (24). Future investigations can now be focused on the specific role(s) of the triphosphatase activities of 2 in core functions. For example, it will be important to dissect whether 2 functions as an NTPase, an RTPase, or both within cores. The 1 shell protein has been proposed to be the capping RTPase in cores based on its reported in vitro RTPase activity (12). However, given our new evidence for in vitro RTPase activity of 2, there appears to be little reason from enzymatic or genetic data to conclude that 1 is more likely than 2 to represent the capping RTPase. The apparent K m values for RNA substrate reported for 1 and 2 are similarly low (0.26 and 0.32 m, respectively), although the reported K cat of the 1-RNA reaction (3.1 min Ϫ1 ) is 10-fold higher than that of the 2-RNA reaction (0.3 min Ϫ1 ) (Ref. 12; this study, Table II). An in vitro system for coupled transcription and capping reconstituted from wt or mutant mORV proteins would be useful for dissecting the relative roles of 1 and 2 but has not been reported to date. We are currently testing core-like particles assembled in insect cells from baculovirus-expressed mORV proteins (39) 3 for this purpose. New information about the precise structural positions and orientations of proteins within the core may also be informative since whether the catalytic regions of 1 or 2 have access to the triphosphorylated 5Ј end of nascent mRNA within the crowded core interior may be a key determinant of whether either protein truly acts as an RTPase during viral mRNA synthesis (4,9,10). 4 What other types of functions might 2 or 1 mediate that could be associated with NTP hydrolysis? Many enzymes are known to couple NTP hydrolysis to protein conformational changes; RNA and DNA helicases, myosins, kinesins, and G proteins are well known examples. In many cases the conformational changes allow movement of the NTPase along a nucleic acid or protein polymer track (53,54). The 3 crystal structure suggests some such possible roles for 2 or 1 in melting, translocating, and/or reannealing the genomic plusstrand RNA as it is passed around the outside of the 3 polymerase, whereas the genomic minus-strand RNA is passed through the central cavity (and catalytic site) of 3 during transcript elongation (9). Alternatively, an NTP-dependent activity by either 2 or 1 could be required during a limited part of each transcription cycle, such as for template melting or translocation before the transcript has reached a certain length or for reinitiation after termination. To understand the functions of the mORV core as a molecular machine for mRNA synthesis, the roles of 2 and 1 must be characterized in more detail. The capacity of 2 to bind to microtubules (this study, Fig. 3B; also see Refs. 35 and 36) dictates that we must also be alert to possible NTP-dependent functions of 2 during mORV genome replication and assembly in infected cells.