The Rabies Virus L Protein Catalyzes mRNA Capping with GDP Polyribonucleotidyltransferase Activity

The large (L) protein of rabies virus (RABV) plays multiple enzymatic roles in viral RNA synthesis and processing. However, none of its putative enzymatic activities have been directly demonstrated in vitro. In this study, we expressed and purified a recombinant form of the RABV L protein and verified its guanosine 5′-triphosphatase and GDP polyribonucleotidyltransferase (PRNTase) activities, which are essential for viral mRNA cap formation by the unconventional mechanism. The RABV L protein capped 5′-triphosphorylated but not 5′-diphosphorylated RABV mRNA-start sequences, 5′-AACA(C/U), with GDP to generate the 5′-terminal cap structure G(5′)ppp(5′)A. The 5′-AAC sequence in the substrate RNAs was found to be strictly essential for RNA capping with the RABV L protein. Furthermore, site-directed mutagenesis showed that some conserved amino acid residues (G1112, T1170, W1201, H1241, R1242, F1285, and Q1286) in the PRNTase motifs A to E of the RABV L protein are required for cap formation. These findings suggest that the putative PRNTase domain in the RABV L protein catalyzes the rhabdovirus-specific capping reaction involving covalent catalysis of the pRNA transfer to GDP, thus offering this domain as a target for developing anti-viral agents.


Introduction
Rabies virus (RABV), a member of the Lyssavirus genus belonging to the Rhabdoviridae family in the order Mononegavirales, is a causative agent of classic rabies, a fatal neurological disease, in humans and animals [1,2]. Rabies kills more than 50,000 people each year worldwide, especially in developing countries [3]. Furthermore, RABV-related lyssaviruses (e.g., Mokola virus, Duvenhage virus, European bat lyssaviruses 1 and 2, Australian bat lyssavirus) are known to cause rabies-like diseases in humans [4]. Therefore, it is important to develop therapeutic targets against these lyssaviruses.
RABV possesses an 11.9 kb negative-strand RNA genome, which contains five genes encoding nucleocapsid (N), phospho-(P), matrix (M), glyco-(G), and large (L) proteins [1,2]. RABV particles deliver ribonucleoprotein complexes into the cytoplasm of host cells, where RNA-dependent RNA polymerase (RdRp) complexes composed of the L and P proteins synthesize viral mRNAs from the genomic RNA template encapsidated with the N proteins. However, the inability to establish efficient in vitro transcription systems [5,6] has hampered progress in understanding the molecular mechanisms of RABV mRNA biosynthesis.
Structural homology of the L protein (2127 amino acids) of RABV to that of vesicular stomatitis virus (VSV), a well-characterized rhabdovirus belonging to the Vesiculovirus genus, suggests that it proteins were obtained. To generate mutant L genes, site-directed mutagenesis was performed using the Quick Change Lightning mutagenesis kit according to the manufacturer's instruction (Agilent Technologies, Santa Clara, CA, USA).

Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE) and Western Blotting
Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting were performed as described previously [16]. Mouse anti-His-tag monoclonal antibody (GenScript, Piscataway, NJ, USA), horseradish peroxidase-conjugated goat anti-mouse IgG polyclonal antibody (Santa Cruz Biotechnology, Dallas, TX, USA), and ECL detection system (GE Healthcare Life Sciences, Pittsburgh, PA, USA) were used to detect His-tagged proteins.

The RABV L Protein Catalyzes Unconventional RNA Capping
To investigate the putative cap-forming activities of the RABV L protein, we expressed and purified its C-terminal His-tagged form. The purified RABV L protein was analyzed by 7.5% SDS-PAGE followed by staining with Coomassie Brilliant Blue (Figure 1a). The full-length RABV L protein was co-purified with small amounts of proteins with molecular masses of 110 and 120 kDa, and, therefore, its purity was estimated to be~70%. Western blotting with anti-His antibody suggested that the 110 kDa protein is a C-terminal fragment of the C-terminal His-tagged RABV L protein (Figure 1b, marked by a closed arrowhead).
The purified recombinant RABV L protein was subjected to the VSV in vitro capping assay with a 5 1 -triphosphorylated oligo-RNA with a RABV mRNA-start sequence (5 1 -AACAC), instead of the VSV mRNA-start sequence (5 1 -AACAG) (Figure 2a). The RABV L protein capped 5 1 -triphosphorylated (lane 2) but not diphosphorylated (lane 3) AACAC with [α-32 P]GDP to generate the G(5 1 )ppp(5 1 )A cap structure. The RABV L protein (0.1 µg) produced 0.91˘0.15 fmol (mean˘standard deviation of three independent determinations) of GpppA with GDP for 2 h. This result indicates that the specific activity (9 fmol GpppA/µg protein/2 h) of the RABV L protein is significantly lower than the reported specific activity of the VSV L protein (5 pmol GpppA/µg protein/2 h [27]), although these activities cannot be directly compared because of the results from independent experiments with the different RNA substrates for the respective L proteins. When using [α-32 P]GTP as a substrate (lane 4), G(5 1 )ppp(5 1 )A (0.40˘0.04 fmol) and the tetraphosphate-containing cap structure G(5 1 )pppp(5 1 )A (0.19˘0.00 fmol) were produced with the RABV L protein as reported for the VSV L protein [17]. G(5 1 )pppp(5 1 )A appears to be formed as a by-product by the pRNA transfer to GTP rather than to GDP [17]. As expected, the RABV L protein hydrolyzed the γ-phosphate of GTP into P i (specific activity, 0.6 pmol P i /µg protein/2 h) in a dose-dependent manner (Figure 2b, lanes 2 and 3). These results suggest that the RABV L protein catalyzes the unconventional capping reaction with the GTPase and PRNTase activities. purified its C-terminal His-tagged form. The purified RABV L protein was analyzed by 7.5% and, therefore, its purity was estimated to be ~70%. Western blotting with anti-His antibody 123 suggested that the 110-kDa protein is a C-terminal fragment of the C-terminal His-tagged RABV L 124 protein (Figure 1b, marked by a closed arrowhead).

145
RABV L protein as reported for the VSV L protein [17]. G(5')pppp(5')A appears to be formed as a 146 by-product by the pRNA transfer to GTP rather than to GDP [17]. As expected, the RABV L protein

169
As expected, the RABV leader RNA-start sequence, ACGCU, did not serve as a substrate for the 3.2. The RABV L Protein Specifically Caps the RABV mRNA-Start Sequence mRNAs of RABV and RABV-related lyssaviruses are known to start with the 5 1 -AACAY (Y, C or U) sequence which is similar to the mRNA-start sequence AACAG of vesiculoviruses [28]. To identify key nucleotide residues in the RABV mRNA-start sequence required for capping with [α-32 P]GDP by the RABV L protein, we measured substrate activities of oligo-RNAs with various sequences (Figure 3a). Replacement of the first three nucleotides (A 1 , A 2 , and C 3 ) in AACAC (lane 2) with different nucleotides abolished (lanes [3][4][5] or significantly diminished (lane 6) their substrate activities, while replacement of A 4 and C 5 showed no or modest effects on their activities (lanes 7-9). As expected, the RABV leader RNA-start sequence, ACGCU, did not serve as a substrate for the RABV capping enzyme (lane 10). In contrast, the conventional mRNA capping enzyme of vaccinia virus efficiently capped all these RNA substrate with [α-32 P]GTP (Figure 3b, lanes 2-10). These results indicate that the RABV L protein strictly recognizes the first three nucleotides of the RABV mRNAs to form the cap structure.

194
in motif D of the VSV L protein has been identified as the covalent pRNA attachment site [19]. Using the recombinant VSV L protein, we have previously identified key amino acid residues in PRNTase motifs A-E that are required for RNA capping in the step of the L-pRNA intermediate formation [19,22]. All the motifs are strictly conserved in the L proteins of RABV ( Figure 4) and other lyssaviruses [22]. 194 in motif D of the VSV L protein has been identified as the covalent pRNA attachment site [19].  [19]. Consensus amino acid sequences of the PRNTase motifs A-E [22] are shown on the bottom (x, Φ, and ζ indicate any, hydrophobic, and hydrophilic amino acids, respectively).
In order to investigate whether these motifs participate in RNA capping with the RABV L protein, we generated L proteins with alanine mutations in selected residues. Similar to the WT L protein (Figure 1a; Figure 5a, lane 1), all these mutant L proteins were co-purified with small amounts of the 110 kDa and 120 kDa proteins (Figure 5a, lanes 2-12). As shown in Figure 5b, while the D729A mutation in the RdRp active site [29] 12) in motif E, abolished the capping activity. Similar to the S1155A mutation in motif B of the VSV L protein [22], S1168A (lane 5) significantly reduced the capping activity of the RABV L protein to 9% of the WT activity.

198
In order to investigate whether these motifs participate in RNA capping with the RABV L 199 protein, we generated L proteins with alanine mutations in selected residues. Similar to the WT L 200 protein (Figure 1a; Figure 5a, lane 1), all these mutant L proteins were co-purified with small 201 amounts of the 110-kDa and 120-kDa proteins (Figure 5a, lanes 2-12). As shown in Figure 5b

233
The recombinant RABV L protein specifically capped pppRNA, but not ppRNA, with GDP, 234 generated from GTP with GTPase (Figure 2), as reported for the VSV L protein [16]. In contrast, Using the VSV and CHPV L proteins, we have previously identified a rhabdovirus-specific arginine residue (R1221 for VSV; R1211 for CHPV) in the vicinity of motif D as essential for the PRNTase activity [18,19]. Consistent with these findings, the RABV L protein with an alanine substitution (R1235A) for the counterpart of R1221 of the VSV L protein was found to be inert in RNA capping (lane 8). Interestingly, the alanine mutation in the non-conserved P1287 residue in motif E increased the capping activity 3.6-fold (lane 13). It should be noted that the VSV L protein has an alanine residue at this position in motif E (see Figure 4). Taken together, these results demonstrated that the PRNTase motifs in the RABV L protein are essential for the capping activity.

Discussion
Earlier studies reported that RNA synthesis activities of RABV ribonucleoproteins and detergent-disrupted virions are significantly lower than those of VSV [5,6]. Therefore, it has remained challenging to further characterize RABV enzymes involved in mRNA biogenesis using in vitro transcription systems. Here, we demonstrated that the purified recombinant RABV protein catalyzes the unconventional RNA capping reaction with the PRNTase activity, although to a much lesser degree than the VSV L protein. This is the first example to directly show the enzymatic activities of the RABV L protein in vitro, thus opening up the opportunity to investigate the functions of this poorly characterized protein.
Thus, as in the case of the VSV L protein [16,19], the RABV L protein was strongly suggested to transfer pRNA from pppRNA to GDP via a covalent L-pRNA intermediate with the PRNTase activity ( Figure 6). However, so far, our attempts to demonstrate the formation of the putative intermediate with the RABV L protein have failed because of the low specific activity of the PRNTase domain.

248
Similar to the recombinant VSV L protein as well as the native VSV L-P RdRp complex [16,17], 249 the GTPase activity was co-purified with the recombinant RABV L protein (Figure 2b). This activity 250 is required to hydrolytically convert GTP into GDP, which acts as the pRNA acceptor for the 251 following PRNTase reaction (Figure 6). Although some mutations (e.g., Y1152A in motif B, W1188F 252 in motif C) in the PRNTase domain of the VSV L protein were found to abolish and diminish the 253 PRNTase and GTPase activities, respectively [22], it still remains unclear whether the L protein itself 254 catalyzes GTP hydrolysis. Further studies are needed to identify the putative GTPase domain/active 255 site in the rhabdoviral L proteins or possibly cellular GTPase associated with these L proteins.

256
As reported for the recombinant VSV L protein [17], the recombinant RABV L protein produced 257 the G(5')pppp(5')A cap structure on RNA probably as a by-product when using GTP as a substrate 258 (Figure 2a). G(5')pppp(5')A seems to be generated by the transfer of pRNA from pppRNA to GTP 259 prior to hydrolysis of GTP into GDP under the optimal conditions for the PRNTase activity, but not 260 for the GTPase activity. Although G(5')pppp(5')A is known to be formed on a small part of mRNAs 261 synthesized by native VSV ribonucleoproteins in vitro [17], its biological significance, if any, is not 262 known.

263
We have previously reported that the VSV L protein efficiently caps pppRNAs with the VSV 264 mRNA-start sequence (5'-AACAG). In this sequence, the A residue at position 1 (A1) and the 265 pyrimidine residue at position 3 (Y3, C > U) are critical for the substrate activity with VSV PRNTase, 266 but the A residue at position 2 (A2) can be functionally replaced with G or partially with C [16,17].
Since the mRNA-start sequences of RABV and RABV-related lyssaviruses (5'-AACAY) are very 268 similar to that of VSV [28], lyssaviral L proteins were suggested to recognize their mRNA-start 269 sequences with VSV-like RNA-binding sites in their PRNTase domains [16]. In fact, similar to VSV, 270 the A1 and C2 residues in pppAACAC were found to be essential for RNA capping with the RABV L 271 protein (Figure 3). To our surprise, however, we found that the A2 residue in pppAACAC cannot be  Similar to the recombinant VSV L protein as well as the native VSV L-P RdRp complex [16,17], the GTPase activity was co-purified with the recombinant RABV L protein (Figure 2b). This activity is required to hydrolytically convert GTP into GDP, which acts as the pRNA acceptor for the following PRNTase reaction (Figure 6). Although some mutations (e.g., Y1152A in motif B, W1188F in motif C) in the PRNTase domain of the VSV L protein were found to abolish and diminish the PRNTase and GTPase activities, respectively [22], it still remains unclear whether the L protein itself catalyzes GTP hydrolysis. Further studies are needed to identify the putative GTPase domain/active site in the rhabdoviral L proteins or possibly cellular GTPase associated with these L proteins.
As reported for the recombinant VSV L protein [17], the recombinant RABV L protein produced the G(5 1 )pppp(5 1 )A cap structure on RNA, probably as a by-product when using GTP as a substrate (Figure 2a). G(5 1 )pppp(5 1 )A seems to be generated by the transfer of pRNA from pppRNA to GTP prior to hydrolysis of GTP into GDP under the optimal conditions for the PRNTase activity, but not for the GTPase activity. Although G(5 1 )pppp(5 1 )A is known to be formed on a small part of mRNAs synthesized by native VSV ribonucleoproteins in vitro [17], its biological significance, if any, is not known.
We have previously reported that the VSV L protein efficiently caps pppRNAs with the VSV mRNA-start sequence (5 1 -AACAG). In this sequence, the A residue at position 1 (A 1 ) and the pyrimidine residue at position 3 (Y 3 , C > U) are critical for the substrate activity with VSV PRNTase, but the A residue at position 2 (A 2 ) can be functionally replaced with G or partially with C [16,17]. Since the mRNA-start sequences of RABV and RABV-related lyssaviruses (5 1 -AACAY) are very similar to that of VSV [28], lyssaviral L proteins were suggested to recognize their mRNA-start sequences with VSV-like RNA-binding sites in their PRNTase domains [16]. In fact, similar to VSV, the A 1 and C 2 residues in pppAACAC were found to be essential for RNA capping with the RABV L protein (Figure 3). To our surprise, however, we found that the A 2 residue in pppAACAC cannot be substituted with G or C for RNA capping with the RABV L protein. These results suggest that the RABV PRNTase domain recognizes the 5 1 -AAC sequence more strictly than VSV.
We verified that conserved amino acid residues (G1112, T1170, W1201, H1241, R1242, F1285, and Q1286) in the PRNTase motifs A to E of the RABV L protein are critical for RNA capping ( Figure 5). Significant homology of the putative PRNTase domain in the RABV L protein to that in the VSV L protein (Figure 4) suggests that these residues constitute the active site of the RABV PRNTase domain and are required for the L-pRNA intermediate formation in the pRNA transfer reaction. H1241 in motif D (HR motif), the counterpart of H1227 of the VSV L protein, most probably serves as a nucleophile to form the putative L-pRNA intermediate. For the L-pRNA intermediate formation, the basic amino acid residue R1242 in motif D may interact with the 5 1 -triphosphate group of pppRNA and facilitate the release of PP i , possibly by acting as a proton donor. Other residues in motifs A, B, C, and E appear to be involved in pppRNA binding and/or structural maintenance of the PRNTase domain. Although Li et al. [23] reported that G1100 in motif A of the VSV L protein is not required for mRNA capping, our recent [22] and current ( Figure 5) studies showed that G1100 of the VSV L protein and its RABV counterpart, G1112, play an important role in mRNA capping.
In this study, using our in vitro RNA capping assay with the recombinant RABV L protein, we demonstrated that the RABV L protein catalyzes the unique RNA capping reaction with the putative PRNTase domain in an mRNA-start sequence-dependent manner. Studies are under way to further characterize the mRNA capping activity and other enzymatic activities, such as RNA synthesis and cap methylation activities, of the RABV L protein. In vitro RNA synthesis and processing systems with the recombinant RABV L protein will be useful for elucidating the molecular mechanisms of lyssaviral mRNA biogenesis and developing antiviral agents against lyssaviral L proteins.