Binding Site for S-Adenosyl-L-methionine in a Central Region of Mammalian Reovirus l2 Protein EVIDENCE FOR ACTIVITIES IN mRNA CAP METHYLATION*

One or more proteins in mammalian reovirus core particles mediate two RNA methylation activities, (guanosine-7-N)-methyltransferase and (guanosine-2*O)-methyltransferase, that contribute to forming the 5* cap 1 structure on viral mRNA. We used UV irradiation to identify core proteins that bind S-adenosyl-L-methionine (SAM), the methyl-group donor for both methyltransferases. A [methyl-H]SAM-binding site was observed among the reovirus l proteins; was shown to be specific by competition with low levels of S-adenosyl-Lhomocysteine, the product of methyl-group transfer from SAM; and was subsequently localized to protein l2. l2 mediates the guanylyltransferase reaction in cap formation and was previously proposed to mediate one or both methylation reactions as well. SAM binding was demonstrated for both l2 in cores and l2 expressed in insect cells from a recombinant baculovirus. Using three different methods to cleave l2, a binding site for SAM was tentatively localized to a central region of l2, between residues 792 and 1100, which includes a smaller region with sequence similarity to the SAM-binding pocket of other methyltransferases. Alanine substitutions at positions 827 and 829 within this predicted binding region greatly reduced the capacity of baculovirusexpressed l2 protein to undergo UV cross-linking to SAM but had no effects on either the guanylyltransferase activity of this protein or its conformation as judged by partial proteolysis, suggesting that one or both of these residues is essential for SAM binding. Based on these findings, we propose that the two methyltransferase activities involved in mRNA capping by reovirus cores utilize a single SAM-binding pocket within a central region of l2.

One or more proteins in mammalian reovirus core particles mediate two RNA methylation activities, (guanosine-7-N)-methyltransferase and (guanosine-2-O)-methyltransferase, that contribute to forming the 5 cap 1 structure on viral mRNA. We used UV irradiation to identify core proteins that bind S-adenosyl-L-methionine (SAM), the methyl-group donor for both methyltransferases. A [methyl-3 H]SAM-binding site was observed among the reovirus proteins; was shown to be specific by competition with low levels of S-adenosyl-Lhomocysteine, the product of methyl-group transfer from SAM; and was subsequently localized to protein 2. 2 mediates the guanylyltransferase reaction in cap formation and was previously proposed to mediate one or both methylation reactions as well. SAM binding was demonstrated for both 2 in cores and 2 expressed in insect cells from a recombinant baculovirus. Using three different methods to cleave 2, a binding site for SAM was tentatively localized to a central region of 2, between residues 792 and 1100, which includes a smaller region with sequence similarity to the SAM-binding pocket of other methyltransferases. Alanine substitutions at positions 827 and 829 within this predicted binding region greatly reduced the capacity of baculovirusexpressed 2 protein to undergo UV cross-linking to SAM but had no effects on either the guanylyltransferase activity of this protein or its conformation as judged by partial proteolysis, suggesting that one or both of these residues is essential for SAM binding. Based on these findings, we propose that the two methyltransferase activities involved in mRNA capping by reovirus cores utilize a single SAM-binding pocket within a central region of 2.
The first of four enzymes for forming the 5Ј cap 1 structure (m 7-N Gppp m 2Ј-O GpC[pN] n -OH) on each reovirus mRNA is an RNA triphosphate phosphohydrolase (1)(2)(3). This enzyme removes the ␥ phosphate from the 5Ј-terminal nucleotide of the nascent mRNA, a guanosine in all 10 reovirus mRNAs, generating a diphosphorylated terminus that serves as substrate for the RNA guanylyltransferase. The reovirus protein that mediates this activity in cores remains unproven; however, two proteins, 1 and 2, have been shown to influence the ion and temperature dependence of NTP hydrolysis by cores (4,5), and a recombinant 1 protein expressed in yeast has been shown to act in vitro as both NTPase (6) and RNA triphosphatase (7).
The next enzyme to act in cap formation is the RNA guanylyltransferase. The RNA guanylyltransferase activity mediated by core protein 2 (8 -10) was the focus of recent work in our lab. 1 Previous work associated sequences near the N terminus of 2 with this activity, particularly lysine 226, at which covalent linkage to GMP was found to occur as part of the catalytic mechanism exhibited by this and related enzymes (9,11). We have recently shown that a 40-kDa N-terminal proteolytic fragment of 2 can mediate both steps in the guanylyltransferase reaction: covalent linkage to GMP using GTP as substrate and transfer of GMP to an acceptor molecule, GDP or GTP in our experiments. 1 Lastly, two methyltransferase activities act to form the final cap 1 structure on reovirus mRNAs (3,12). Both activities appear to be mediated by reovirus-encoded proteins because no host proteins are known to be present in reovirus cores (13). The first is an RNA (guanosine-7-N)-methyltransferase, which uses S-adenosyl-L-methionine (SAM) 2 as substrate for methylgroup transfer to the N-7 position of the RNA guanylyltransferase-added 5Ј-terminal guanosine (m 7-N GpppGpC[pN] n -OH) (2). The second is an RNA (guanosine-2Ј-O)-methyltransferase, which uses SAM as substrate for methyl-group transfer to the O-2Ј position of the 5Ј-most RNA polymerase-added nucleotide (m 7-N Gppp m 2Ј-O GpC[pN] n -OH). For both reactions, S-adenosyl-L-homocysteine (SAH) is formed as the by-product of methyl-group transfer from SAM. The reovirus protein that mediates these methylations has remained unproven, although the literature contains two suggestive findings. First, an observation reported as unpublished data by Seliger et al. (14) indicated that the 2 protein was the only reovirus core protein that underwent covalent cross-linking to 8-azido-S-adenosyl [ 35 S]methionine. Second, data base sequence comparisons by Koonin (15) revealed that 2 contains sequences between residues 825 and 888 that bear similarities to other SAM-utilizing methyltransferases. Notably, this region of similarity includes sequences found in the SAM-binding pocket of methyltransferases for which such information is known (15).
Identification of the protein(s) that mediates methyltransferase reactions by reovirus cores is important for understanding how the capping enzymes are arranged in these particles. A confounding observation from previous work is that recombinant 2 protein (r2) expressed from a vaccinia virus vector does not exhibit methyltransferase activity, possibly because it remains monomeric in solution (10) and does not adopt the pentameric structure observed for 2 in cores (16). Having confirmed ourselves that r2 expressed from a baculovirus vector remains monomeric in solution and also fails to mediate methyltransferase activity, 1 we turned to an alternative approach to address this potential activity of 2. In this study, we used UV cross-linking in combination with protein fragmentation and site-directed mutagenesis to localize a binding site for [ 3 H]SAM to a central region of 2 sequence, overlapping that predicted by Koonin (15). Our results strongly suggest that 2 mediates at least one of the two methylation reactions in mRNA capping by reovirus cores.

EXPERIMENTAL PROCEDURES
Purification of Virus Particles-Virions of reovirus T1L or T3D were purified as described (17). Virion concentrations were determined from the relationship 1.0 A 260 ϭ 2.1 ϫ 10 12 virions/ml (18). Cores of reovirus T3D were prepared by digesting virions at a concentration of 3 ϫ 10 13 particles/ml with 200 g/ml ␣-chymotrypsin (Sigma) for 1.5 h at 37°C in virion buffer (10 mM Tris, pH 7.5, 10 mM MgCl 2 , 150 mM NaCl). Cores of reovirus T1L were chymotrypsin-digested as described (19). The chymotrypsin was inactivated with 1 mM phenylmethylsulfonyl fluoride (Sigma). To purify cores, digests were loaded onto a preformed CsCl gradient ( ϭ 1.30 -1.55 g/cm 3 ) and subjected to equilibrium centrifugation. The particles were dialyzed into virion buffer. Modified cores were generated as described in Luongo et al. (20). In brief, 1 ϫ 10 13 cores of reovirus T3D in virion buffer were heated to 52°C for 30 min, cooled to 37°C, and treated with 200 g/ml chymotrypsin for 10 min. Following treatment, the chymotrypsin was inactivated with 1 mM phenylmethylsulfonyl fluoride, and the modified cores were isolated following equilibrium centrifugation.
UV Cross-linking-Reactions were UV irradiated using a modification of the procedure of Ahola et al. (21). In brief, 20-l reactions containing the protein sample and 1.7 Ci of S-adenosyl-L-[methyl-3 H]methionine ([ 3 H]SAM) (80 Ci/mmol; DuPont) in 50 mM Tris-HCl, pH 7.5, 2 mM dithiothreitol, and 2 mM EDTA were pipetted into 0.5-ml microfuge tubes. The reactions were incubated on ice while being irradiated for 30 min with 254-nm UV light in a Stratalinker 2400 crosslinking oven (Stratagene, La Jolla, CA) with the samples placed 4 cm from the light source. We tested several variations to the reaction conditions in an effort to optimize the UV cross-linking of [ 3 H]SAM to core proteins. For these studies, we used an assay in which proteinbound [ 3 H]SAM was collected by precipitation with trichloroacetic acid after UV cross-linking and then detected by scintillation counting. The results (data not shown) suggested a pH optimum of pH 7.5, a preference for low concentrations of monovalent salts (Ͻ75 mM NaCl), and an insensitivity to low concentrations (10 mM) of divalent salts (MgCl 2 ). Because these conditions were similar to those we had chosen to begin these studies, we continued to use the original conditions.
Polyacrylamide Gel Electrophoresis-Protein samples were mixed with one-third volume of 3 ϫ Laemmli loading buffer (3% SDS, 9% ␤-mercaptoethanol, 375 mM Tris-HCl, pH 8.0, 30% sucrose, and 0.004% bromphenol blue) and heated at 100°C for 2 min. The two identical samples were resolved on the same gel. The protein sample on one-half of the gel were visualized by staining with Coomassie Brilliant Blue (Sigma). For SDS-PAGE, the proteins in the other half of the gel were electroblotted to nitrocellulose (Bio-Rad) in 25 mM Tris, 192 mM glycine, pH 8.3. The nitrocellulose was coated with En 3 Hance spray surface autoradiography enhancer (DuPont), and the label was detected by fluorography. For the continuous phosphate-urea-SDS gels (7.5% acrylamide, 100 mM Na 2 PO 4 , pH 7.0, 20 mM EDTA, pH 8.0, 6 M urea, 0.1% SDS) 1.5-mm-thick mini-gels (Bio-Rad) were run at 20 V for 18 h. To visualize radiolabel in the unstained half of the gel, it was treated with Intensify universal autoradiography enhancer (DuPont) and dried onto filter paper under vacuum prior to fluorography.
Expression and Proteolysis of r2-The reovirus 2 protein was ex-pressed and purified as described in Luongo et al. 1 In brief, recombinant baculovirus AcMNPV.T3DL2 was generated by transfection of Spodoptera frugiperda insect cells (Sf21, Invitrogen, Carlsbad, CA) with recombinant bacmid generated by cloning the L2 gene (22) into the pFastBacI vector (Life Technologies, Inc.). Trichoplusia ni insect cells (High Five, Invitrogen) were infected with AcMNPV.T3DL2 and harvested 48 h postinfection. R2 protein was purified by anion exchange chromatography. To remove residual protein contaminants and concentrate r2, the column fractions were ammonium sulfate precipitated. For thermolysin-limited proteolysis of r2, the protein was diluted 10-fold with 50 mM Tris-HCl, pH 8.0, and placed on ice for 5 min. The protein was then placed at 4°C and digested with a 5% volume of 2 mg/ml thermolysin (95 g/ml final) (Sigma) for 30 min. The thermolysin was inactivated with 25 mM EDTA, pH 8.0. Immunoblot Analysis-Protein was electroblotted from SDS-polyacrylamide gels to nitrocellulose (Bio-Rad) in 25 mM Tris, 192 mM glycine, pH 8.3. Binding of the monoclonal antibody 7F4 was detected using goat anti-mouse IgG alkaline phosphate conjugate and colorimetric reagents 5-bromo-4-chloro-3 indoyl phosphate p-toluidine salt and p-nitro blue tetrazolium chloride (Bio-Rad). The molecular weight standards (kaleidoscope prestained) were purchased from Bio-Rad.
N-terminal Sequencing-Proteolytic products of 2 were transferred from an SDS-polyacrylamide gel to polyvinylidene fluoride paper (Applied Biosystems, Foster City, CA) (23), and the band corresponding to the 80K fragment was excised from the blot. N-terminal sequencing by Edman degradation was performed at the University of Michigan Protein and Carbohydrate Structure facility (Ann Arbor, MI).
Proteolytic Cleavage by Formate Treatment-T3D cores at high concentration (1.2 ϫ 10 12 ) were incubated with [ 3 H]SAM, subjected to UV cross-linking, and resolved on an 8% SDS-polyacrylamide gel. The protein band was excised and incubated in 75% formate (Sigma) (acid hydrolysis) or distilled water (control) at 37°C for 16 h. The proteins were resolved on a 10% SDS-polyacrylamide gel and visualized as described above for UV cross-linked protein.
Generation of Mutant r2-The L2 gene cloned into the BamHI site of pBluescript (Stratagene) was the template for site-directed mutagenesis utilizing the Quik Change site-directed mutagenesis kit (Stratagene). The complementary mutagenic primers 5Ј-CGTTGTGCTAG-CTCTTGCGACGGGACCAGAGGC and 5Ј-GCCTCTGGTCCCGTCGC-AAGAGCTAGCACAACG were used because the underlined nucleotides cause missense mutations that change both aspartate 827 and glycine 829 to alanine. In addition, these two alterations remove a BglII restriction site at nucleotide 2491. In brief, 50 ng of template was amplified with either 125 or 500 ng of each primer under the following conditions: 1 cycle at 94°C for 30 s followed by 16 cycles at 94°C for 30 s, 55°C for 1 min, and 68°C for 14 min followed by one cycle at 15°C for 5 min. Mutant clones were identified by screening for the loss of the BglII site. The presence of the mutations were confirmed by sequencing both DNA strands from nucleotides 1804 to 3157 using the ABI Prism dye terminator cycle sequencing ready reaction kit (Applied Biosystems). No second site mutations were identified in this region, which encompasses the NsiI to NcoI fragment that was ligated with the 7.4kilobase pair region of pFastBacI-L2 generated by digestion with NsiI and NcoI (New England Biolabs, Beverly, MA). Recombinant bacmid was generated and used to produce recombinant baculovirus by transfection of Sf21 cells. Mutant protein was expressed and purified as described for wild type 2.
Autoguanylylation Assay-The wild type and mutant r2 proteins were assayed for autoguanylylation activity by incubation of each protein with 5 Ci of [␣-32 P]GTP (DuPont) in 50 mM Tris, pH 8.0, 10 mM MgCl 2 , and 2 mM dithiothreitol for 30 min at room temperature. EDTA, pH 8.0, was added to a final concentration of 10 mM to halt the reaction. The proteins were resolved on a Tris-glycine-SDS 8% polyacrylamide gel, and the covalent protein-associated radiolabel was visualized by PhosphorImager analysis (Molecular Dynamics, Sunnyvale, CA).
Expression of Recombinant 3 (r3)-r3 was expressed using a recombinant baculovirus system as follows. The reovirus T3D L1 gene, which encodes the 3 protein, cloned into the PstI site of pBR322 (22) was obtained from Michael Roner (Florida Atlantic University). The L1 gene was PCR amplified using the forward primer 5Ј-CGCGGGTAC-CTCGAGACGACCATGGCATCCATGATAC and the reverse primer 5Ј-CGCGGGTACCCATGGTAGACTCACGCTGAC. These primers add KpnI sites (underlined) 5Ј and 3Ј of the L1 coding sequence and XhoI and NcoI restriction sites (italicized), respectively, 5Ј of the L1 coding sequence. The Extend Long Template PCR system (Boehringer Mannheim) was used to amplify the L1 gene. In brief, buffer 2 supplemented to 5.25 mM MgCl 2 (final) was used in the reaction. The amplification conditions were one cycle at 92°C for 2 min followed by 10 cycles at 92°C for 15 s, 60°C for 35 s, and 68°C for 3.5 min followed by 20 cycles at 92°C for 15 s, 65°C for 35 s, and 68°C for 3 min 50 s with an addition of 20 s/cycle followed by one cycle at 68°C for 7 min. The L1 PCR product was digested with XhoI and KpnI (New England Biolabs) and ligated into the pFastBacI vector digested with the same enzymes. Recombinant bacmid was generated and used to produce recombinant baculovirus by transfection of Sf21 cells. For protein expression, High Five cells were infected with 0.6 plaque forming units of L1 recombinant baculovirus and harvested 62-70 h postinfection. Cells were placed on ice and lysed in a hypotonic buffer (10 mM Tris, pH 7.4, 2.5 mM MgCl 2 , 0.5% Triton X-100, 100 mM NaCl, 2% complete protease inhibitor mixture (Boehringer Mannheim)). Nuclei and large cellular debris were pelleted at 200 ϫ g for 15 min at 4°C. The 3 in the supernatant was partially purified by removal of protein contaminant by pretreatment with 20% (saturation at 0°C) ammonium sulfate and precipitation. The protein precipitated by 35% ammonium sulfate contained 3 and was resuspended in 50 mM Tris, pH 8.0. The resuspended protein was then used for UV cross-linking.

UV Cross-linking of [ 3 H]SAM to 2 and/or 3 Proteins in
Reovirus Cores and Virions-To identify which reovirus core proteins contain binding sites for SAM, core particles of reovirus T3D were incubated with [methyl-3 H]SAM and irradiated with UV light (254 nm) to cross-link this molecule to its binding site(s). Following electrophoresis in a Tris-glycine-SDS gel (8% acrylamide), blotting to nitrocellulose to improve detection of 3 H, and fluorography, radiolabel was strongly detected in the protein band (Fig. 1A). In this type of gel, the band comprises core proteins 1 (120 copies/particle), 2 (60 copies/particle), and 3 (12 copies/particle). When virions of reovirus T3D were treated in the same protocol, radiolabel was again strongly detected in the band and at similar levels as seen with cores (Fig. 1A). The latter result concurs with findings that the capping methyltransferases have nearly full activity in reovi-rus virions (24). Only minor amounts of radiolabel were associated with the major outer capsid proteins in virions, 1C and 3 (600 copies each per virion), providing evidence that the binding of [ 3 H]SAM is specific to one or more of the proteins. The same results were obtained with cores and virions of reovirus T1L (data not shown), indicating that this property is not unique to a single reovirus strain.
To identify which proteins contain binding sites for SAM, core particles of reovirus T3D were incubated with [ 3 H]SAM, irradiated with UV light to achieve cross-linking, and subjected to electrophoresis in a phosphate-urea-SDS gel (7.5% acrylamide). Such gels provide good separation of the 1 protein from the 2 and 3 proteins, which still comigrate (18). Following fluorography, radiolabel was strongly detected in the 2/3 band but not in the 1 band (Fig. 1B). When virions of reovirus T3D were treated in the same protocol, radiolabel was again strongly detected in the 2/3 band and at similar levels as observed with cores (Fig. 1B). The same results for cross-linking to [ 3 H]SAM were obtained with cores and virions of reovirus T1L (data not shown). These results indicate that the primary binding sites for SAM identified by UV cross-linking are found in 2 and/or 3, and not in 1, in both particle types of the two reovirus strains.

Competition by SAH and Optimization of [ 3 H]SAM Binding to Reovirus Protein(s)-
Before proceeding with other studies to localize the binding site(s), we addressed the specificity of binding to SAM by the protein(s) in cores by performing competition experiments with nonradiolabeled SAH, the stable by-product of methyl-group transfer from SAM in the capping methyltransferase reaction (2). The capacity of SAH to inhibit methyltransferase activities in reovirus cores was previously demonstrated (25). In addition, competition by SAH has been used to demonstrate the specificity of binding to SAM by other methyltransferases (26 -28). Using SDS-PAGE, electroblotting, and fluorography, we showed that radiolabeling of the protein(s) in reovirus T3D cores upon UV cross-linking in the presence of 1 M [ 3 H]SAM was inhibited by SAH with an IC 50 of ϳ0.5 M (Fig. 2A). In contrast, S-adenosyl-D-homocysteine (dextro form of SAH) had little or no effect on cross-linking to [ 3 H]SAM at the highest concentration tested (5 M) (data not shown), providing further evidence for the specificity of competition by the physiological, laevo form of SAH. These results provide evidence that cross-linking to [ 3 H]SAM reflects the presence of one or more specific binding sites for SAM in 2 and/or 3.
In an effort either to compete or to enhance SAM binding to the proteins in cores, we tested the effects of GTP and the cap analog GpppG. We hypothesized that GTP binding to the guanylyltransferase region of 2 (9) 1 or GpppG binding to one or both putative methyltransferase regions in 2 (Refs. 14 and 15 and this study) might enhance the observed SAM binding activity. Nevertheless, when cross-linking to [ 3 H]SAM was performed in the presence of up to 20 M GTP (data not shown) or GpppG (Fig. 2B), SAM binding to 2 and/or 3 in T3D cores was unchanged, indicating that these compounds do not modulate this activity.

Localization of [ 3 H]SAM Binding to a 120K N-terminal Portion of 2 Protein in Modified
Cores-In an attempt to localize the binding site(s) for SAM in reovirus cores, we used modified cores containing 2 protein that lacks 25K of C-terminal sequences due to sequential treatments with mild heat (52°C) and chymotrypsin (20). After subjecting these modified cores to UV cross-linking in the presence of [ 3 H]SAM, followed by SDS-PAGE and fluorography, we found the 120K N-terminal fragment of 2 that is present in these particles to be strongly radiolabeled (Fig. 3). Because the complementary, 25K C-terminal region of 2 is degraded as the modified cores are made (20), we could not address whether it, too, might contain a SAM-binding site; nonetheless, the available data demonstrate that at least one SAM-binding site is located within the 120K N-terminal portion of 2. The data also indicate that this large N-terminal portion of 2 is sufficient for at least one SAM-binding activity of 2, consistent with previous findings that removing the C terminus of 2 by treatment with heat and chymotrypsin does not reduce the capacity of modified cores to methylate newly transcribed viral mRNAs (20).
The small amount of radiolabel that was seen to migrate at the position of full-length 2 protein after cross-linking with modified cores (Fig. 3) could represent the small amount of uncleaved 2 protein that remains present in these particles and/or the 3 protein, which fully resists cleavage during the heat and chymotrypsin treatments (20). Nonetheless, because radiolabel was mostly associated with the 120K fragment of 2 after cross-linking with modified cores, the findings strongly suggest that any cross-linking of [ 3 H]SAM to 3 that may also have occurred with modified cores or cores was minor compared with that to 2 (see last section of "Results" for additional analysis of 3).

Localization of [ 3 H]SAM Binding to a 55K C-terminal Portion of 2 in Cores by Limited Hydrolysis with Formic Acid-
The deduced amino acid sequence of T3D 2 protein includes four aspartate-proline pairs that should be sensitive to limited acid hydrolysis using 75% formic acid (29). The locations of these sites and the sizes of fragments that result from partial cleavage are shown in Fig. 4. After incubating with 75% formic acid to obtain cleavage of gel-isolated 2 protein that had been previously cross-linked to [ 3 H]SAM in T3D cores and then subjecting the products to a second round of SDS-PAGE, followed by electroblotting and fluorography, we observed a number of radiolabeled bands consistent with the expected fragments (Fig. 4). In particular, most radiolabel was associated with a band near 55K, the size of predicted fragment V from the C terminus of 2. The next most prominent bands were ones near 60K (predicted fragment IV-V), 90K (predicted fragment III-V), and 120K (predicted fragment II-V). Fragments of similar sizes and relative intensities to these were detected in parallel samples analyzed by immunoblot using monoclonal antibody 7F4, confirming that these fragments all contain the C terminus of 2. Gel-isolated 2 protein, which had been previously cross-linked to [ 3 H]SAM in T3D cores but which was incubated without formic acid prior to the second round of SDS-PAGE, served as a marker for full-length protein and was observed to be both radiolabeled and bound by 7F4 (Fig. 4). These findings strongly suggest that SAM is bound to one or more site within the C-terminal, 55-kDa fragment resulting from limited acid hydrolysis of 2.

UV Cross-linking of [ 3 H]SAM to r2 Protein and Localization to an 80K C-terminal Portion of the Protein by Cleavage with
Thermolysin-To confirm the capacity of 2 to bind SAM and to further localize the site of SAM binding in 2, we performed experiments using a soluble, partially purified form of r2 that had been expressed in insect (Sf21 or High Five) cells from a recombinant baculovirus (AcMNPV) and shown to mediate the capping guanylyltransferase activity assigned to an N-terminal region of 2. 1 The r2 protein obtained in this manner was clearly visualized as a stained band after SDS-PAGE and was also strongly radiolabeled after UV cross-linking in the presence of [ 3 H]SAM (Fig. 5, 0 time points). These findings demonstrated that the monomeric r2 binds SAM.
In an effort to localize the site(s) of SAM binding in r2, we screened a variety of proteinases for limited proteolysis of r2 to yield distinct fragments. Thermolysin initially cleaved r2 to yield fragments with approximate sizes of 40K (data not shown) and 100K (Fig. 5). The 40K fragment was concluded to be derived from N-terminal sequences of 2 based on its spontaneous, covalent labeling with [␣-32 P]GTP (data not shown), characteristic of the guanylyltransferase activity of this region (9). 1 The 100K fragment was concluded to be the C-terminal complement of the 40K fragment based on its binding by monoclonal antibody 7F4 (20) (Fig. 5). As determined from time course analysis (data not shown), while this initial cleavage by thermolysin was continuing to occur in uncleaved copies of r2, both the 40K and 100K fragments became subject to additional cleavages. In the case of the 100K fragment, secondary cleavages occurred such that a stable 80K fragment was generated (Fig. 5). Because the 80K fragment is also recognized by 7F4 (Fig. 5), it must have retained its C-terminal sequences (20) and lost sequences from the middle of 2. When thermolysincleaved r2 protein was subjected to UV cross-linking in the presence of [ 3 H]SAM, radiolabeling of both the 100K and 80K fragments was observed (Fig. 5), localizing one or more SAMbinding site to the 80K C-terminal portion of 2. To confirm which region of 2 the 80K fragment represents, we subjected it to N-terminal sequencing. The sequence obtained by Edman degradation was leucine-alanine-arginine-proline-phenylalanine-proline (data not shown), corresponding to residues 562-567 in the deduced amino acid sequence of the T3D 2 protein (14) and consistent with cleavage by thermolysin (N-terminal to leucine). The calculated mass of a fragment spanning residues 562-1289 (the C terminus of 2) is 80.7 kDa, consistent with the observed M r (80K).
The findings for localizing SAM-binding sites within either 2 from cores or r2 using three different methods of proteinase or chemical cleavage can be summarized as follows. The SAMbinding fragments resulting from each of these treatments overlap sequences between positions 792, in the most C-terminal aspartate-proline pair in 2, and ϳ1100, the site of 2 truncation in cores treated with heat and chymotrypsin. Thus, if 2 were to contain a single site for binding SAM that can be identified by UV cross-linking, this region (amino acids 792 to ϳ1100) would represent the likely location for this binding site. It is notable that this region includes the SAM-binding motif and adjacent sequences with similarity to known SAM-utilizing methyltransferases (residues 825-888) as described by Koonin (15). Koonin (15) regarding binding to SAM by residues near position 830 in 2, as well as our own evidence that SAM is bound in this region, we created mutations in the pBluescript-inserted cDNA copy of T3D L2 (see "Experimental Procedures") such that the encoded 2 protein contained two amino acid substitutions: alanine 827 (for aspartate in wild type 2) and alanine 829 (for glycine in wild type 2) in the B-B-aspartate-B-glycine-X-glycine motif (where B represents a hydrophobic residue and X represent any residue) shared by many methyltransferases (15). The mutant L2 gene was then used to generate a recombinant baculovirus for expressing the mutant 2 protein in insect cells. Upon demonstrating that the mutant 2 was expressed in these cells, we partially purified it using the same protocol as described for wild type 2. The purified mutant and wild type 2 proteins were then analyzed for SAM binding by incubation with [ 3 H]SAM and UV cross-linking. The results indicate that the mutant 2 protein is greatly diminished in its capacity to bind SAM (Fig. 6A). By quantitating the amounts of radiolabel associated with the wild type or mutant 2 proteins with the amounts of each protein loaded in the gel lanes, we estimated that the mutant protein was labeled with [ 3 H]SAM at Յ4% the level of wild type. The greatly reduced binding to SAM by the mutant 2 was also demonstrated in a sample to which the 100K fragment of wild type 2 (generated by prior thermolysin treatment) was added prior to cross-linking with [ 3 H]SAM. In this case, the 100K fragment of wild type protein was strongly radiolabeled, and the full-length mutant protein again showed little detectable labeling above background. In contrast, the mutant 2 protein was efficiently radiolabeled by covalent linkage to GM[ 32 P] (Fig. 6B) as part of the guanylyltransferase reaction mediated by an N-terminal region of 2 (9) 1 ; was recognized by monoclonal antibody 7F4 (data not shown), which binds to a C-terminal epitope in 2 (20,30); and exhibited an identical pattern of cleavage upon treatment with thermolysin as the wild type protein (data not shown), indicating that other regions of 2 were not substantially affected by the alanine substitutions at 827 and 829. Because mutations at these two closely spaced positions in 2 so greatly reduced binding to SAM in the cross-linking assay but did not affect other properties, the results provide evidence that a single SAM-binding site in 2, involving residues near To address more directly whether 3 might also bind SAM, we used a recombinant form of the T3D 3 protein that was expressed in insect cells from a recombinant baculovirus. The r3 protein was partially purified using a combination of ammonium sulfate precipitation and anion exchange chromatography (see "Experimental Procedures") and was shown to have poly(C)-dependent poly(G) polymerase activity (data not shown) (31). When partially purified preparations of r2 (see above) and r3 were subjected to UV cross-linking in the presence of [ 3 H]SAM and then analyzed by SDS-PAGE, electroblotting, and fluorography, SAM bound to r2 and its 100K fragment (generated by a contaminating proteinase in this protein preparation 1 ) but not to r3 (Fig. 7). Thus, the results for baculovirus-expressed r2 and r3 proteins, coupled with the results for proteins in cores, suggest that 3 exhibits little if any binding to [ 3 H]SAM using the UV cross-linking assay.

DISCUSSION
Three different fragmentation methods were used to localize binding sites for SAM in the reovirus 2 protein. In the first, involving C-terminally truncated 2 protein in modified cores, a site was localized between the N terminus and a position near residue 1100. In the second, involving limited acid hydrolysis of 2 in cores, a site was localized between position 792 and the C terminus. In the third, involving thermolysin cleavage of r2, a site was localized between position 562 and the C terminus. Recognizing that these localizations overlap a region of 2 between positions 792 and ϳ1100, we hypothesized that one or more binding site for SAM is located within this specific 35-kDa, central region of the protein. Although it is formally possible that more than one SAM-binding site is contained within this region, no candidate sequences are evident besides those identified by Koonin (15) between positions 825 and 888. Moreover, because the alanine substitutions we introduced at positions 827 and 829 within this region so greatly reduced the capacity of r2 to undergo UV cross-linking to [ 3 H]SAM, the findings suggest that a single SAM-binding site involving amino acids near position 830 in 2 has been identified by UV cross-linking in this study. Although it is tempting to state more definitively that 2 contains only the one SAM-binding site, an important caveat is that a second SAM-binding site in 2 may be less efficiently cross-linked to [ 3 H]SAM in our protocol than the site we have studied and may in fact account for the very low level of activity (Յ4% wild type levels) we found remaining in the alanine substitution mutant. Another possibility is that the alanine substitutions at positions 827 and 829 knocked out two distinct or overlapping SAM-binding sites in 2 at the same time. Thus, whereas we obtained no strong evidence for a second SAM-binding site in 2, we cannot definitively rule out that possibility.
Region of 2 Sufficient for SAM Binding-We conceived of other experiments involving expression of 2 or portions of that protein in Escherichia coli in an effort to identify the minimal region of 2 that is necessary for SAM binding. The full-length 2 protein and six deletion mutants expressed to date, however, have been found to be mostly insoluble, not readily solubilized, and inactive at SAM binding in the cross-linking assay (data not shown). Similar difficulties with 2 expression in E. coli have been described by other investigators (9). Thus, we abandoned that approach for any further experiments in the current study.
On the other hand, the findings from partial proteolysis of intact 2 described in this paper, which suggest that a central region of 2 contains a SAM-binding site, can be considered in light of other findings to suggest that a region from 562 to ϳ1100 is sufficient for SAM binding. Because the ϳ25K Cterminal portion of 2 is degraded in generating the 120K N-terminal fragment in cores treated with heat and chymotrypsin (20) and because the 120K fragment can subsequently bind SAM and undergo cross-linking to it (Fig. 3), the ϳ25K C-terminal portion must be dispensable for at least one SAMbinding activity. In addition, because the 40K N-terminal and 100K C-terminal fragments that can be generated from baculovirus-expressed 2 protein by mild proteolysis with several proteinases 1 including thermolysin (Fig. 5) are physically separated according to both native gel electrophoresis and gel filtration chromatography 1 , because the 20K N-terminal portion of the 100K thermolysin-generated fragment is degraded in generating the 80K fragment at later times of thermolysin treatment (data not shown), and because the 80K fragment can  bind SAM and undergo cross-linking to it after proteolysis (Fig.  5), the N-terminal ϳ60K region of 2 must also be dispensable for for at least one SAM binding activity. Thus, our data suggest that an ϳ60-kDa portion of 2 spanning amino acids 562 to ϳ1100 constitutes the smallest region of the protein defined to date as being sufficient for SAM binding and UV cross-linking.
Implications for 2 Function as a Capping Methyltransferase-Biochemical evidence for SAM binding by 2 supports a general expectation (14,15) that 2 mediates one or both of the RNA methylation activities (RNA (guanosine-7-N)-methyltransferase and RNA (guanosine-2Ј-O)-methyltransferase) in forming a 5Ј cap 1 structure on reovirus mRNA. Nevertheless, the failure of r2 to exhibit methylation activity (Ref. 10 and data not shown) still limits studies to dissect the putative methyltransferase activities of 2 at a molecular level. This failure may reflect the incapacity of r2 expressed in isolation from other reovirus proteins to adopt the pentameric structure observed for 2 in cores (10) (see next section).
Despite the caveat mentioned in the preceding section, an interesting possibility that can be inferred from the evidence for a single SAM-binding site in 2 is that the same region of 2 sequence may provide binding to SAM for the two different cap methylation activities in reovirus cores. An example similar to this possibility is seen in the multi-specific DNA methyltransferases of certain bacteriophages. In these enzymes, one polypeptide chain includes a single SAM-binding pocket and a single set of catalytic residues for methylation but multiple DNA-binding sites that allow specific recognition and subsequent methylation of several different DNA sequence motifs (32). Thus, different DNA-binding regions are arranged in those proteins such that each of the different DNA targets can be effectively bound and presented to the same catalytic machinery for methylation of defined bases within them. A similar hypothesis for 2 would be that there are two different sequence regions in the protein that bind the nascent cap at the 5Ј end of reovirus mRNA, one region that binds GpppG-pC[pN] n -OH and presents the base of the terminal guanosine for methylation at the N-7 position and another region that binds m 7-N GpppGpC[pN] n -OH (cap 0) and presents the sugar of the penultimate guanosine for methylation at the O-2Ј position (Fig. 8). Both substrates are methylated by catalytic machinery using the same SAM-binding pocket. Another possibility is that once the first methylation occurs, that is, once m 7-N GpppG-pC[pN] n -OH is generated, this new RNA acceptor may undergo a change in orientation rather than translocating the RNA to a distinct binding site, such that the sugar of the penultimate guanosine is now properly presented for methylation at the O-2Ј position. This rearrangement might involve conformational changes in 2, m 7-N GpppGpC[pN] n -OH, or both, possibly due to the presence of the (guanosine-7-N)-methyl group. The presence of one or two sites in 2 for binding RNA acceptors is testable, and such experiments are in progress.
The circumstance suggested in Fig. 8 for 2, one SAM binding site for the two methylation reactions, has not been described for a capping methyltransferase to date. For example, the (guanosine-7-N)-methyltransferase and (nucleoside-2Ј-O)methyltransferase activities of vaccinia virus are mediated by distinct proteins (33,34). Similarly, the capping methyltransferase encoded by the ABD1 gene of Saccharomyces cerevisiae has been shown to act as a (guanosine-7-N)-methyltransferase but not as a (nucleoside-2Ј-O)-methyltransferase (27). Thus, a detailed analysis of the SAM binding and methyltransferase activities of 2 may provide unique insights into this class of enzymes.
Relationship of Findings to Pentameric Structure of Core-bound 2-The arrangement of 2 as homopentamers in reovirus cores has been known for some time (35,36) although visualized at higher resolution only recently by cryoelectron microscopy with (16,20) or without (37) three-dimensional image reconstruction. The regions of 2 that are essential for pentamerization remain undefined; however, the aforementioned three-dimensional reconstructions suggest that regions at the intermediate radii spanned by 2 (which extends radially over ϳ100 Å in reovirus cores) may be most important. 1 Although the SAM-binding pocket is functionally constituted by a 2 monomer, as shown in this study, it is conceivable that residues in 2 that contribute to catalysis and/or to binding RNA acceptor molecules in association with the SAM-binding pocket (Fig. 8) may be contributed by an adjacent monomer in the core-bound, pentameric 2. This requirement would provide one possible explanation for why monomeric 2 does not exhibit methyltransferase activity. A related, though distinct, possibility is that the residues necessary to constitute each active methyltransferase unit are contributed by the same 2 monomer but that interactions with adjacent monomers in the 2 pentamer are necessary for permitting these residues to assume their catalytically active conformations. These putative interactions between adjacent 2 monomers are apparently not required for forming a functional SAM-binding site, however, as we show in this study using monomeric r2 obtained from insect cells.