Role of RNA Structure and RNA Binding Activity of Foot-and-Mouth Disease Virus 3C Protein in VPg Uridylylation and Virus Replication

Production and modification of plasmids.

The plasmids containing FMDV O1Kaufbeuren (O1K) cDNA sequences, namely, pUb-3D^pol, pQE30/3CD^pro(C163G), pGC-cre, and pT7S3, have been described previously (12, 29). Similarly, the plasmid pETm11-3C, which encodes the 3C protease from FMDV A10₆₁, has also been described previously (6, 7). Mutations producing single-amino-acid substitutions in this 3C protease were introduced by QuikChange mutagenesis (Strategene) using the primers listed in Table 1. A region (nt 403 to 1120) of the FMDV 5′ UTR (O1Kaufbeuren) was amplified using primers poly(C)-down-For and IRES/REV and cloned into pT7blue (Novagen) to create pT7(PK+cre+IRES). The manipulation, analysis, and growth of plasmids were achieved using standard techniques (37).

Expression and purification of FMDV 3D^pol.

Expression and purification of the untagged FMDV 3D^pol was achieved by production and cleavage of ubiquitin-tagged FMDV 3D^pol by coexpression with the Ubp2 protease as described previously (16, 29).

Production and purification of FMDV 3C^pro(C163G).

The coding sequence of O1K 3C^pro was amplified by PCR using oligonucleotides O1K/3CFOR and O1K/3CREV, with BamHI and HindIII restriction sites, respectively (Table 1), using pQE30/3CD^pro(C163G) as a template (29). The fragment was inserted into the pT7-blue vector (Novagen), and the resulting plasmid was digested with BamHI and HindIII; the released insert was ligated between the same sites of the pQE30 vector. The expression of this protein and the protease-inactive forms of His-tagged 3CD(C163G) and the 3C(A10₆₁) proteins plus their derivatives were achieved and purified as described previously (6, 7, 29). The 3C from the A10₆₁ strain of FMDV has amino acid substitutions C95K, C142S, and C163A to inactivate the proteolytic activity and enhance its solubility (6, 7).

Production of FMDV 3B₃3C and 3B₁₂₃3C.

For expression of the FMDV nonstructural precursor proteins 3B₃3C and 3B₁₂₃3C in the ubiquitin system (16), the cDNA fragment was amplified by PCR using the appropriate primers (Table 1) and cloned into the pT7-blue vector. Modification within the 3C protease sequence to change the codon for residue C163 to Gly in the pT7-3B₃3C and pT7-3B₁₂₃3C constructs was carried out using the QuikChange site-directed mutagenesis protocol as described previously (29), using appropriate primers (listed in Table 1). The modified fragments were excised from the pT7-blue clones using SacII and BamHI and cloned into a similarly digested pET26b-Ub vector. The presence of the desired mutation was confirmed by sequencing. Coexpression with the ubiquitin protease and purification of the native 3B₃3C and 3B₁₂₃3C proteins was performed essentially as described for the native 3D^pol (16, 29). The phosphocellulose column-purified fractions were directly loaded onto a 1-ml Hi-trap SP Sepharose column (Amersham Bioscience), eluted by gradient elution, and dialyzed using buffer A (50 mM Tris, pH 8.0, 20% glycerol, 10 mM β-mercaptoethanol, and 0.1% NP-40) with 750 or 500 mM NaCl, respectively.

Construction of infectious cDNAs for FMDV encoding single-amino-acid substitutions within the 3C protease.

A PCR fragment (1,750 bp) encompassing the FMDV 3C coding sequence (O1K), along with the preexisting BclI and BlpI restriction sites, was amplified using primers O1KBclI and O1KBlpI (Table 1) and ligated into the pT7-blue vector. The resultant clone (pT7-BB) was modified by site-directed mutagenesis (QuikChange mutagenesis kit; Stratagene) to encode serines at the codons for R92, R95, and R97 using appropriate primers (Table 1). Positive clones were grown using a dam and dcm mutant Escherichia coli strain (ER2925 from New England Biotechnology). The purified DNA was digested with BclI and BlpI to release an insert of 1,729 bp and introduced into the similarly digested FMDV infectious cDNA clone pT7S3 (12). The presence of each mutation in the full-length cDNA was confirmed by sequencing.

Electroporation of FMDV RNA into BHK cells.

Wild-type (WT) or mutant full-length FMDV RNA transcripts (ca. 4 μg) were prepared using a Megascript kit (Ambion) containing T7 RNA polymerase and were added to BHK 21 cells (2 × 10⁶) suspended in siPORT siRNA Electroporation Buffer (Ambion) (0.75 ml), which were shocked (1 kV; 0.6 ms; two pulses) using an Electro Square Porator T820 (BTX Electronics). Following incubation at room temperature for 10 min, the cells were transferred to a 25-cm² flask containing 5 ml of virus growth medium with 1% fetal calf serum and incubated for 8 h at 37°C. The recovered virus was released by freeze-thawing, and the yield was determined by plaque assay on BHK cells. The plaques were visualized after 40 to 48 h, using methylene blue with 4% formaldehyde in phosphate-buffered saline.

Uridylylation assays.

The synthesis of VPgpU(pU) was measured essentially as described previously (18, 29, 33). Unless otherwise indicated, the reaction mixtures contained 50 mM HEPES (pH 7.5), glycerol (8%), MgCl₂ (2 mM), 3B3/VPg3 (12.5 μM), cre/bus RNA transcripts (1 μM), [α-³²P]UTP (0.75 μCi; Amersham Bioscience), UTP (10 μM), His-3CD(C163G) (1 μM), and 3D^pol (1 μM). The final NaCl concentration in the reaction mixture (from the protein solutions) was kept at 6.25 mM. Reaction mixtures were incubated for 30 min at 30°C and analyzed by Tris-tricine sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) with 14% polyacrylamide. The gels were dried and autoradiographed. The products were quantitated with a PhosphorImager (Molecular Imager FX; Bio-Rad) to determine the level of [³²P]UMP incorporation. Other RNA transcripts were produced when required using T7 RNA polymerase on appropriately linearized plasmids.

Preparation of BHK S10 extracts.

Monolayers of BHK cells (five 175-cm² flasks) were harvested using a cell scraper, washed in phosphate-buffered saline, resuspended in cold hypotonic buffer (20 mM HEPES, pH 7.5, 10 mM KCl, 1.5 mM magnesium acetate, 1 mM dithiothreitol [DTT]) (5 ml), and allowed to swell on ice for 15 min. The cells were lysed with a Dounce homogenizer (40 strokes) and centrifuged (5,000 × g for 5 min) to remove the cellular debris and nuclei. Cytoplasmic extract (S10) was prepared by centrifuging the postnuclear fraction at 10,000 × g for 15 min at 4°C. The supernatant was dialyzed for 3 h against 40 mM HEPES, pH 8.0, 120 mM potassium acetate, 5.5 mM magnesium acetate, 10 mM KCl, and 6 mM DTT; aliquoted; and stored at −70°C.

Cell-free translation of FMDV RNA in BHK S10 extracts.

In vitro translation reactions were performed using mixtures (25 μl) containing 70% BHK S10, 1 μg of FMDV RNA transcripts (wild type or mutant), 50 mM HEPES (pH 8.0), 50 mM potassium acetate, 1.0 mM ATP, 0.25 mM GTP, 0.25 mM CTP, 0.25 mM UTP, 5 mM creatine phosphate, 50 μg/ml creatine phosphokinase (Sigma), 800 U of RNasin/ml in the presence of [³⁵S]methionine. The reaction mixtures were incubated for 90 min at 30°C and analyzed by SDS-PAGE and autoradiography.

RNA isolation, reverse transcription-PCR amplification, and sequencing.

Total RNA from virus-infected BHK cells was extracted with Trizol (Invitrogen, Life Technologies). Viral cDNA was made using SuperScript III Reverse Transcriptase (Invitrogen) with selective viral primers and amplified using PCR. The products were purified from agarose gels and sequenced using the CEQ 8000 Genetic Analysis System (Beckman Coulter).

RNA-protein interaction assays.

The binding of WT and mutant 3C proteins to ³²P-labeled FMDV cre/bus RNA transcripts was analyzed using a filter binding assay (10, 49). ³²P-labeled RNA transcripts corresponding to the FMDV cre/bus were produced using a Maxiscript kit (Ambion) with pGC-cre (29) as a template with [α-³²P]UTP. The reaction was performed using buffer conditions similar to those of the in vitro uridylylation assay (50 mM HEPES, pH 7.5, 8% glycerol, 2 mM magnesium chloride); total volume, 20 μl). The RNA (2.5 × 10⁴ cpm) was incubated with different concentrations of 3C protein (0.5 to 10 μM). Assays were performed for 20 min at 30°C and spotted onto Protran BA-85 nitrocellulose filters (Schleicher & Schuell). The filters were extensively washed with 50 mM HEPES, pH 7.5, 8% glycerol, 2 mM magnesium chloride and dried, and the bound RNA was measured using liquid scintillation. The percentage of RNA bound was calculated. UV cross-linking assays were performed using 0.4 μg of purified recombinant proteins [3C(wt), 3C(R92S), 3C(R95S), and 3C(R97S)] or bovine serum albumin with uniformly radiolabeled FMDV S-fragment RNA transcripts (2 × 10⁵ cpm) in 20 μl of 10 mM HEPES-KOH (pH 7.9), 25 mM KCl, 2 mM MgCl₂, 10% glycerol, 0.05% NP-40, 0.5 mM DTT, 10 μg tRNA. After a 30-min incubation on ice, the reaction mixtures were exposed to UV light for 30 min on ice. The samples were treated with 2 μl RNase cocktail (Ambion) for 30 min at 37°C, and the proteins were analyzed using 10% SDS-PAGE and autoradiography.

Identification of the optimal RNA sequences required as the template for FMDV VPg uridylylation.

In our previous studies (29), it was shown that the FMDV cre/bus transcript (ca. 50 nt), at relatively high concentration (1 μM), acts as an efficient template for VPg uridylylation; however, larger RNA transcripts (e.g., the FL RNA) were significantly more efficient. To define the optimal RNA sequences required for this reaction, a series of RNA transcripts were prepared, including different regions of the FMDV 5′ UTR; they are indicated in Fig. 1A. Each transcript was used at 35 nM within a standard uridylylation assay using VPg3 as a substrate (Fig. 1B). As shown previously (29), the FL-RNA and the Wt-XbaI transcript were significantly more active than the short cre/bus transcript itself. Furthermore, the entire 5′ UTR and the PK+cre+IRES transcript were even more efficient. However, the PK+cre transcript was not much better than the cre/bus transcript alone. These results suggested that the presence of the IRES sequences within the transcript significantly enhanced the template activity in this reaction. This probably reflects some modification of the structure or availability of the cre/bus in the presence of these additional sequences.

FMDV 3C can substitute for FMDV 3CD for the uridylylation of FMDV VPg.

In our previous studies, it was shown that the uridylylation of FMDV VPgs required UTP plus Mg²⁺ (or Mn²⁺) and was also dependent on the presence of each of the following FMDV components: 3D^pol with 3CD plus the RNA template incorporating the cre/bus (29). This closely mirrors the requirements for uridylylation of the PV or HRV-2 VPg (15, 33, 34). However, it should be noted that, in contrast to the PV system, the FMDV components do not significantly utilize poly(A) as a template even in the presence of Mn²⁺ (29). Since Pathak et al. (31) have demonstrated that the role of PV 3CD in this reaction can be fulfilled by 3C alone, we investigated whether the FMDV 3C protein could replace 3CD in the FMDV VPg uridylylation assay. The results obtained using the cre/bus transcript (ca. 50 nt) as a template showed that the 3C protein from either the type O or type A serotype of FMDV can substitute for 3CD (from serotype O) in this assay (Fig. 2A). To determine the relative efficiency of the reaction in the presence of 3C or 3CD, a series of reactions was performed using a range of 3C concentrations in the presence of VPg3 with the cre/bus transcript as a template (Fig. 2B). Under these conditions, it was found that even at 10 μM 3C, the efficiency of VPg uridylylation was only about 25% of that achieved with 1 μM 3CD (Fig. 2B). Furthermore, using the complete FMDV 5′ UTR as the RNA template, which is much more efficient than using the cre/bus transcript (Fig. 1B), 3C was again less efficient than 3CD in supporting the reaction, but under these conditions, higher concentrations of the 3C protein inhibited the reaction (Fig. 2C).

Precursors of FMDV VPg, termed 3B₃3C and 3B₁₂₃3C, can be uridylylated in vitro.

The VPg peptide is derived from the P3 precursor and can potentially be liberated from various different precursors, e.g., 3AB or 3BC. Since 3C can support the uridylylation of VPg, it seemed possible that 3BC would be a good substrate for this reaction even in the absence of added 3C or 3CD. To test this idea, two different precursors of the FMDV VPg were expressed in E. coli and purified to near homogeneity (Fig. 3A). The recombinant 3B₃3C and 3B₁₂₃3C precursor proteins (which lacked protease activity due to modification of a catalytic residue, C163G) were then assayed within in vitro uridylylation assays using the cre/bus transcript with 3D^pol alone or in the presence of added His-3CD. Both precursors were efficiently uridylylated, both in the presence and absence of 3CD (Fig. 3B and C). However, the addition of 3CD still stimulated the reactions (ca. three- to fivefold), consistent with the enhanced uridylylation of VPg in the presence of 3CD rather than just 3C (Fig. 2). The precursor 3B₃3C was a more efficient substrate than 3B₁₂₃3C, although neither was as good as the 3B₃ (VPg) peptide alone on an equimolar basis (Fig. 3D).

Identification of a putative RNA binding site on the surface of the FMDV 3C protein.

Individual residues within a conserved motif (KFR⁸⁴DI⁸⁶) of PV 3C that affect its binding to the cloverleaf at the 5′ end of the PV genome have been identified (1). Residue R84 within this motif was found to be critical for RNA binding, and modification of R84 and I86 blocked the ability of PV 3CD to support VPg uridylylation (33). A comparison of the PV, HAV, HRV, and FMDV 3C sequences in this region is shown in Fig. 4. The KFRDI sequence is partially conserved as (R/K)VRDI⁹⁹ in FMDV; the PV residue R84 corresponds to R97 within the FMDV 3C sequence. Since the structure of the FMDV 3C protein has been determined recently (7), it is possible to locate this motif, and residue R97 in particular, on the surface of 3C. Consistent with observations of the PV and HAV 3C proteins (5, 27), this motif on the FMDV protein is located on the face of the protein opposite from the protease active site (Fig. 5). It is apparent that other positively charged residues are close to R97 on the surface of the FMDV 3C protein, including residues R92 and K/R95. Residues corresponding to R97 of FMDV are absolutely conserved in the 3C proteins of other FMDVs, PV, HRV, and HAV, while at the position of residue 95 in the FMDV 3C, there is either K or R (both positively charged) in each of these proteins. It is noteworthy that residue R92 in the FMDV 3C protein is also conserved in PV and HRV, but not in HAV. Note that in the original FMDV A10₆₁ incomplete cDNA, the 3C residue 95 was encoded as a Cys (C) but was mutated to Lys (K) (as found in other FMDV sequences) to increase the solubility of the protein (6, 7); it is not known whether the presence of a C residue at this position is compatible with virus viability.

Analysis of putative RNA binding residues in FMDV 3C.

In order to test the roles of these basic residues for the uridylylation function, we generated three mutant 3C proteins in which residues R92, K95, and R97 were each replaced by an uncharged serine (S) residue. The WT and mutant His-tagged 3C proteins were each expressed in E. coli and purified to near homogeneity. Equal amounts of these proteins, as validated by Western blot analysis (Fig. 6A), were then tested in uridylylation assays in conjunction with 3D^pol and either the cre/bus transcript (Fig. 6B) or the entire FMDV 5′ UTR (Fig. 6C) as the RNA template. In each case, the mutation had a very deleterious effect on uridylylation activity. The R97S mutant was almost completely inactive (<5% of the WT) in these assays, and the R92S mutant was also severely debilitated (ca. 6% of the WT). The K95S mutant was the most active mutant but still displayed only about 25% of WT activity.

To confirm that these mutations in FMDV 3C resulted in a defect in RNA binding, the abilities of these proteins to bind ³²P-labeled cre/bus transcripts were determined using a filter binding assay (Fig. 7A). The R92S and R97S proteins were both significantly reduced in their abilities to bind this RNA transcript compared to the WT 3C protein. The K95S mutant was also deficient in cre/bus binding but was slightly more active than the other mutants. These results closely mirrored the abilities of these proteins to support the uridylylation of VPg (Fig. 6). In addition, the ability of the WT and mutant 3C proteins to bind a ³²P-labeled RNA transcript corresponding to the FMDV S fragment (Fig. 1) was analyzed using a UV cross-linking assay (Fig. 7B). The WT 3C protein was efficiently cross-linked to the RNA, but the mutant proteins interacted with the S fragment significantly less well; the results showed a pattern similar to that observed in the filter binding assay using the cre/bus transcript. The defect in uridylylation activity appears greater than the defect in RNA binding; this may reflect the nature of the assays or may suggest that these residues also have a specific role in VPg uridylyation.

Roles of RNA binding residues in FMDV 3C in virus replication.

The codons for residues R92, R95, and R97 of the FMDV 3C were modified individually to encode serine residues within a full-length infectious cDNA of FMDV (O1K strain) termed pT7S3 (12). RNA transcripts were prepared and translated in an in vitro translation system using a BHK cell S10 lysate. The WT and mutant transcripts produced very similar patterns of polypeptides (Fig. 8A), indicating that normal polyprotein processing occurred; this showed that the proteolytic activity of the 3C protein was not greatly affected by any of these mutations. The full-length RNA transcripts were also introduced into BHK cells using electroporation. It was found that the R95S and R97S mutant transcripts were noninfectious; no plaques were observed from cell harvests prepared at 8 h postelectroporation. The R92S mutant transcripts did produce virus, but the yield of virus produced at 8 h postelectroporation was reduced about 10-fold compared to the WT transcript (Fig. 8B). Sequencing of reverse transcription-PCR products obtained from the rescued viruses indicated that they retained the R92S mutation (data not shown), but the presence of second-site mutations has not been examined. These results indicated the importance of each of these residues for efficient virus replication.