INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Rhabdoviruses contain a single-stranded negative-sense RNA genome (11 to 15 kb) which is tightly encapsidated with the viral nucleoprotein (N) to form an RNP (nucleocapsid) template for transcription and replication. During transcription, a 47-nucleotide-long leader RNA and five capped and polyadenylated mRNAs are synthesized (32). The replication process yields nucleocapsids containing full-length antigenome-sense RNA which in turn serve as templates for the synthesis of genome-sense RNA. The active virus-encoded RNA polymerase complex is composed of the large protein (L) and its cofactor, the phosphoprotein (P) (14). The L protein is a multifunctional enzyme and is the RNA-dependent RNA polymerase. This protein may carry out all enzymatic steps of transcription, including initiation and elongation of transcripts as well as cotranscriptional modifications of RNAs such as capping, methylation, and polyadenylation (1). The sequences of rhabdovirus L-protein amino acids have been compared with those of other negative-strand RNA viruses (9, 27, 33). Four motifs (A to D) constitute the so-called polymerase module and are conserved in all viral RNA-dependent DNA and RNA polymerases (24). Part of the highly conserved motif C which is located within the amino-terminal half of the rabies virus L protein has recently been shown to be involved in the formation of the catalytic center of the protein (26). Functions of the P protein are not well defined. Studies with vesicular stomatitis virus (VSV), the best-characterized rhabdovirus, have shown that the P protein is a noncatalytic cofactor and a regulatory protein: it associates with the L protein in the polymerase complex and interacts with both soluble and genome-associated N protein (13, 21, 30). The P protein has different phosphorylation states and is believed to bind with different affinities to the RNP template and to have different transcription activities (2, 3, 17). Furthermore, the VSV P protein has been shown to form multimers, and multimerization seems to be necessary for binding both to the L protein and to the template (12, 17). Rabies virus and VSV are structurally similar. Thus, by analogy, their RNA polymerase complexes may have similar properties. In vitro and in vivo studies have shown that rabies virus P protein forms specific complexes with N proteins (8, 15). We have previously demonstrated the existence of two N-protein binding sites on the P protein; one is located between amino acids 69 and 177, and another requires the carboxy-terminal region comprising the amino acids 268 to 297 (8). The rabies virus P protein has at least two differently phosphorylated forms (34). Four additional proteins (P2, P3, P4, and P5) translated from the P mRNA have been found in purified virus, in infected cells, and in cells transfected with a plasmid encoding the complete P protein. Translation of these proteins is initiated from internal in-frame AUG initiation codons by a leaky scanning mechanism (7).

To characterize functional domains of the rabies virus RNA polymerase, we have expressed P and L proteins from plasmids in cultured cells, and we show that they form a complex that can be immunoprecipitated. Analyses of P-protein deletion mutants reveal that the amino-terminal residues of the P protein are involved in the interaction with the L protein and that they are sufficient to mediate binding to L of a green fluorescent protein (GFP) fusion protein. Analysis of the P-L complex formation with truncated L proteins shows that binding to the P protein is mediated by the carboxy-terminal part of the L protein.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cells and virus. BSR cells, cloned from BHK-21 (baby hamster kidney) cells, were grown in Eagle's minimal essential medium supplemented with 10% calf serum. The CVS strain of rabies virus was cultivated and purified as previously described (18).

Antibodies. The anti-L antibody is a mixture of two rabbit polyclonal antisera, S94 and S97. S94 was raised to a synthetic peptide whose sequence corresponds to the 24 amino-terminal residues of the strain SAD B19 L protein (26). S97 was made against a peptide corresponding to the 19 carboxy-terminal residues (not containing the very last Leu residue) of the SAD B19 L protein. Two anti-P monoclonal antibodies (MAbs), A17 and 25E6, were used as mouse ascites preparations. A17 recognizes an epitope located on the P protein between amino acids 83 and 138, and 25E6 is directed against the first 19 residues of the P protein (25). The MAb raised to GFP was supplied by Clontech.

Plasmid constructions. Plasmids encoding the wild-type P protein and the truncated P proteins PN19, PN52, PN68, PN82, PC30, and PC120 of the CVS strain have been described previously (7, 8). The P19-GFP and P52-GFP fusion constructs were generated by insertion of the 5'-terminal 57 nucleotides (encoding the 19 amino-terminal residues of P) or the 5'-terminal 156 nucleotides (encoding the 52 amino-terminal residues of P), respectively, in frame with the 5' end of the GFP gene in the amino-terminal protein fusion vector (pEGFP-N1; Clontech). The P-GFP fusion genes were then transferred to the HindIII-XbaI sites of plasmid pCDNA1 (Invitrogen) downstream of the T7 promoter sequence. These constructs were called pT7-P19-GFP and pT7-P52-GFP.

DNA transfection. Proteins were transiently expressed by using a T7 vaccinia virus expression system as described by Fuerst et al. (16). BSR cells were grown in 3.5-cm-diameter dishes to 80% confluency and infected with vTF7-3 at a multiplicity of infection 5 PFU/cell. After 1 h of adsorption, the cells were transfected with 5 µg of plasmid DNA by the calcium phosphate coprecipitation procedure (22).

Radiolabeling and immunoprecipitation of viral proteins. Twenty-four hours after infection, proteins were labeled with 1 ml of methionine-free medium containing 50 µCi of [³⁵S]methionine and [³⁵S]cysteine (PRO-mix; specific activity, >1,000 Ci/mmol; Amersham) for 3 h. Cells were harvested by scraping into cold TD buffer (137 mM NaCl, 5 mM KCl, 0.7 mM Na₂HPO₄, 25 mM Tris-HCl [pH 7.5]) and lysed on ice in 1 ml of buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.5% Nonidet P-40, and an antiprotease cocktail (2 µg of leupeptin per ml, 2 µg of antipain per ml, 2 µg of pepstatin per ml, 2 µg of chymostatin per ml, 16 µg of aprotinin per ml). Nuclei were eliminated from the lysate by centrifugation at 12,000 × g for 10 min at 4°C. The cytoplasmic fractions were incubated overnight at 4°C with antibodies (anti-P, anti-L, or anti-GFP). Protein A-Sepharose was then added for 1 h at 4°C. The immune complexes were centrifuged and washed three times in lysis buffer and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by autoradiography.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Interaction between L and P proteins. To study the interaction between the L and the P proteins, we infected BSR cells with vTF7-3 (16) and transfected these cells with a plasmid encoding either L of the SAD B19 strain (pT7T-L) (9) or P of the CVS strain (pMcP) (8) downstream of the T7 RNA polymerase promoter. In parallel experiments, cells were cotransfected with both plasmids. Radiolabeled proteins were extracted under nondenaturing conditions and then immunoprecipitated with monoclonal anti-P (A17) or rabbit anti-L antibodies. A17 recognizes an epitope located on the P protein between amino acids 83 and 138 (8). The immunoprecipitates were analyzed by SDS-PAGE (Fig. 1). The expressed L protein (Fig. 1, lane 7) comigrated with L protein detected in extracts of infected cells (lane 5). Two additional bands were also present in the extracts of transfected cells (lanes 6 and 7). They could correspond to smaller L proteins initiated at internal in-frame AUG codons or terminated early, to L degradation products, or to cellular proteins which coimmunoprecipitated with the L protein. The expressed P protein (lane 2) comigrated with P protein detected in extracts of infected cells (8). Additional smaller proteins (P2 and P3) immunoprecipitated with anti-P antibody were detected similarly in infected and transfected cells (lane 2); we have previously shown that they correspond to proteins initiated by a leaky scanning mechanism from secondary downstream AUG initiation codons (7).

View larger version (46K): [in this window] [in a new window]   FIG. 1.   Detection of L-P complex in infected and transfected cells by immunoprecipitation. BSR cells were infected with rabies virus (i; lane 5) or with vTF7-3 (lanes 1 to 4 and 6 to 9). vTF7-3-infected cells were then either transfected with 5 µg of a plasmid encoding the P (lanes 2 and 8) or L (lanes 3 and 7) protein or cotransfected with both plasmids (lanes 4 and 6). At 24 h after infection or transfection, proteins were labeled with [35S]methionine for 3 h. Cell extracts were prepared and immunoprecipitated with A17 anti-P (lanes 1 to 5) and anti-L (lanes 6 to 9) antibodies. Immune complexes were collected with protein A-Sepharose and analyzed by SDS-PAGE (12% gel).

View larger version (49K): [in this window] [in a new window]   FIG. 2.   P-L complex formation in the absence of coexpression. A mixture of cytoplasmic extracts from cells separately transfected with P and L plasmids [(L) + (P)] or cell extracts from cotransfected cells (L+P) were immunoprecipitated with anti-P (lanes 1 and 3) and anti-L (lanes 2 and 4) antibodies. The immunoprecipitates were analyzed by SDS-PAGE (14% gel).

Mapping the L binding site on the P protein. To identify domains on the P protein required for binding to the L protein, we used a set of six plasmids encoding amino- or carboxy-terminally truncated P proteins (CVS strain) (Fig. 3A) (7, 8). These P proteins were coexpressed with the L protein and then tested for the ability to bind to L protein in coimmunoprecipitation assays with the anti-L or anti-P antibody (Fig. 3B and C). When cell extracts were immunoprecipitated with the A17 anti-P antibody, the L protein was efficiently brought down with the full-length P protein and with the carboxy-terminally truncated proteins PC30 and PC120 (Fig. 3B, lanes 2, 7, and 8). Similarly, the anti-L antibody coimmunoprecipitated efficiently the complexes L-PC30 and L-PC120 (Fig. 3C, lanes 7 and 8). These results are summarized in Fig. 3A and suggest that PC30 and PC120 interacted with the wild-type L protein whereas all amino-terminally truncated proteins did not (Fig. 3B and C, lanes 3 to 5). As even binding of PN19 was not detected, this finding suggested that the very amino terminus of P is critical for efficient binding to the L protein. This possibility is supported by the observation that the L protein interacted only with the complete P protein and not with P2 and P3 proteins initiated from the second and third AUG codons, respectively (Fig. 3C, lane 2), and which correspond to proteins PN19 and PN52, respectively (7). Consequently it is unexpected that the P2 and P3 of the carboxy-terminally truncated proteins (P2C30 and P3C30) are present in the immunoprecipitates obtained with the anti-L antibody (Fig. 3C, lane 7). However, a similar situation is observed after immunoprecipitation with MAb 25E6 directed against the first 19 amino acids of the P protein (25). This MAb, which recognized only the complete P protein and not P2 and P3 (see Fig. 6, lane 2), is able to coimmunoprecipitate not only PC30 but also P2C30 and P3C30 (data not shown). These results suggest that the truncation of the last 30 amino acids of the P protein may induce a change in the behavior of the protein in addition the formation of multimers, but this requires further investigations.

View larger version (31K): [in this window] [in a new window]   FIG. 3.   Mapping the L binding site on the rabies virus P protein. (A) Schematic representation of the truncated P proteins. Dark bars represent the protein product of each deleted P gene, with amino acid positions indicated. The thin angled lines indicate deleted regions. The plasmids encoding truncated P proteins have been described elsewhere (7, 8). P-L binding data are summarized at the right. (B and C) Analysis of interaction between the L protein and the truncated P proteins by immunoprecipitation. Cells were not transfected (NT; lane 2) or infected with vTF7-3 (lanes 1 to 8) and then cotransfected with 5 µg of plasmids encoding the complete L and P proteins (lane 2) or truncated P proteins PN19 (lane 3), PN52 (lane 4), PN68 (lane 5), PN82 (lane 6), PC30 (lane 7), and P120 (lane 8). Cell extracts (35S labeled) were immunoprecipitated with A17 anti-P antibody (B) or anti-L antibody (C). The immunoprecipitates were analyzed by SDS-PAGE (14% gel).

View larger version (18K): [in this window] [in a new window]   FIG. 4.   The first 19 amino acids of P are sufficient for L-P interaction. (A) Schematic representation of the P-GFP fusion proteins. The 19 or 52 amino-terminal residues of P (white bars) are fused to GFP (dark bars); amino acids are indicated. (B) Cells were infected with vTF7-3 (lanes 1 to 6). vTF7-3-infected cells were then transfected with 5 µg of a plasmid encoding the complete L protein (lanes 1 and 6), P19-GFP (lanes 2), or P52-GFP (lanes 3). Cells were also cotransfected with plasmids encoding L and P19-GFP (lanes 4) or P52-GFP (lanes 5). Cell extracts (35S labeled) were immunoprecipitated with the indicated antibodies (anti-GFP, 25E6 anti-P, and anti-L). Cell extracts from transfected cells with a plasmid encoding L protein were also immunoprecipitated with the anti-L antibody to indicate the position of the L protein [(a)]. The precipitated proteins were analyzed by SDS-PAGE (14% gel). Longer exposures of the controls are included (lanes 1).

Mapping the P binding site on the L protein. To characterize the domain of the L protein required for the interaction with the P protein, we constructed six plasmids encoding truncated L proteins as described in Materials and Methods (Fig. 5). Two carboxy-terminal (L1 and L2) and four amino-terminal (L3, L4, L5, and L6) truncated L proteins were expressed in transfected cells and migrated with the predicted relative mobilities (Fig. 6). The additional band immunoprecipitated with the anti-L antibody and detected above the P protein in cells expressing the L, L1, and L2 proteins (lanes 3 to 5) disappears with the removal of the amino-terminal region of the L protein (lanes 6 to 9). Nonspecific precipitation of truncated L proteins was found to be more important than precipitation of wild-type L protein with the anti-P antibody (not shown). This may be due to some misfolding and aggregation of truncated L proteins. We thus used only the anti-L antibody to immunoprecipitate the complex P-L. The results presented in Fig. 4 show that the carboxy-terminally truncated proteins L1 and L2, lacking 789 and 1,262 residues, respectively, did not bind to P since no P protein was found in the immune complexes obtained with the anti-L antibody (Fig. 6, lanes 4 and 5). On the other hand, proteins L3, L4, L5, and L6, truncated to amino acids 471, 1171, 1438, and 1562, respectively, coimmunoprecipitated with P, forming a complex (Fig. 6, lanes 6 to 9). Thus, the L6 mutant lacking 1,561 N-terminal residues from the 2,127-residue full-length protein is still capable of forming a stable L-P complex in this coimmunoprecipitation assay. These data strongly suggest that the P-protein binding site is located in the last 566 residues of the L protein.

View larger version (8K): [in this window] [in a new window]   FIG. 5.   Schematic representation of the truncated L proteins. Dark bars represent the protein product of each deleted L gene, with amino acid positions indicated. The thin lines indicate deleted regions.

View larger version (70K): [in this window] [in a new window]   FIG. 6.   Mapping the L binding site on the rabies virus P protein. Cells were not transfected (NT; lane 1) or infected with vTF7-3 (lanes 1 to 10) and then cotransfected with 5 µg of plasmids expressing the complete P and L proteins (lane 3) or the complete P protein and truncated L proteins L1 (lane 4), L2 (lane 5), L3 (lane 6), L4 (lane 7), L5 (lane 8), and L6 (lane 9). Cell extracts (35S labeled) were immunoprecipitated with anti-L antibody (lanes 1 and 3 to 9). Extracts of cells transfected with plasmid encoding the complete P protein (lanes 2 and 10) were also immunoprecipitated with anti-P MAb 25E6 as described above (lane 2) and MAb A17 (lane 10). The immunoprecipitates were analyzed by SDS-PAGE (12% gel).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The interaction between the rabies virus P and L proteins was studied in transfected cells by coimmunoprecipitation of both proteins with antibodies specific for one component. We showed that the P and L proteins from two strains of rabies virus (CVS and SAD B19) could interact with each other in this system in the absence of other viral proteins. L-P complexes have also been described for other negative-strand RNA viruses, including VSV, Sendai virus, and measles virus. However, recent findings suggest an important difference in the properties of the P-L polymerase complex of rhabdoviruses and paramyxoviruses. Coexpression of P and L proteins in the same cell was necessary for Sendai virus polymerase activity (20) but not for that of VSV (5) on, as shown here, rabies virus. An explanation for the P and L coexpression requirement was suggested by results which indicated that the Sendai virus and measles virus L proteins are unstable unless they are coexpressed with P protein (20, 28). The finding reported here suggests that the rabies virus L protein can be stably expressed in the absence of the P protein. So far, we cannot exclude a protective effect of the P protein on L-protein stability, as reported recently for the polymerase of VSV (5). The results of coimmunoprecipitation experiments indicate that the L-P association is not as strong as the N-P interaction (8). Sedimentation analysis also suggests a weak P-L complex since the interaction was completely disrupted during cell extract preparation and/or during centrifugation (not shown). We cannot exclude that this weak association is due to difference of strain of both partners, although the amino acid homology of the SAD B19 and CVS L proteins is high (9, 33). However, studies using the cell two-hybrid system also show that the VSV P-L interaction is not as strong as the P-N association (30, 31).

Using a deletion mutant analysis, we identified domains of the P and L proteins involved in the formation of the L-P complex. The interaction of P with L protein was not disrupted by large deletions within the carboxy-terminal part of the P molecule. In contrast, all of the deletion mutants that were altered in the amino-terminal half of P were defective in the formation of complex with L. We proposed that the first 19 residues of P may be important for L-protein binding, and we tested this hypothesis by using fusion protein constructs. P19-GFP still interacted with L, demonstrating that the L binding site lies in the 19 first residues of the P protein. These results are in agreement with the previous finding reported for VSV that the negatively charged amino-terminal region of P binds to the L protein in vitro (13) and in vivo (30). However, another region of the VSV P protein located near the carboxy terminus of P has also been shown to contribute to L-protein binding (30). From our data, we cannot exclude the possibility that another binding site exists on the P protein. The fact that PC30 interacts more efficiently with L than the truncated PC120 or the complete P protein (Fig. 3B and C) suggests that a region upstream of the last 30 amino acids may be involved in strengthening the association with L. We can speculate that this site is masked in the complete P protein and becomes accessible after truncation of the 30 carboxy-terminal residues. Data for the RNA polymerase complexes of other negative-strand viruses such as Sendai virus have mapped a single binding site in the carboxy-terminal half of the P protein (28). These observations reveal differences between rhabdoviruses and paramyxoviruses in protein interaction involved in RNA synthesis, although the processes are very similar.

Several reports have implicated an essential role for P-protein phosphorylation in VSV transcription and in formation of the L-P complex (2, 3). Recent data have shown that without phosphorylation, the bacterially expressed P protein is unable to multimerize in vitro (12, 17) and cannot bind to the L protein (17). It has been proposed that P-protein phosphorylation induces a conformational change associated with oligomerization of P and then facilitates P interaction with the L protein (29). The Sendai virus P protein has also been shown to be trimeric (11). We can speculate that the rabies virus P protein is oligomeric too. We have shown that the first 19 amino acids of P were sufficient for the binding to L, but oligomerized N termini might bind more easily or efficiently. From our experiments, we cannot conclude that oligomerization of rabies virus P protein is required for L-P complex formation.

The rabies virus P protein forms a complex with the N protein also (8). Two N binding sites exist on P; one is located between the amino acids 69 and 138, and the other requires the 30 last residues of the protein. The N- and L-protein binding sites on the P protein, therefore, do not overlap. This correlates well with the dual functionality of the P protein: P can interact simultaneously with both L and N proteins to act as a transcription factor when complexed with the L protein and as a replication factor when complexed with the N protein.

We also mapped the region of L involved in the interaction with P; the results show that the carboxy-terminal 566 amino acids of the L protein are sufficient to permit P-L polymerase complex formation. This finding is reinforced by the demonstration that deletions of 789 residues within this region eliminate binding to P protein. This conclusion is consistent with the recent data reported for VSV that a small 36-amino-acid-long deletion in the carboxy-terminal region of L (amino acids 1638 to 1673) abolished transcription activity and abrogated binding to the P protein (4, 5). The authors mentioned that they could not exclude the possibility that the failure to bind P might also be due to misfolding or aggregation of the mutant L protein rather than deletion of the putative binding site. However, our results, which are based on positive binding, demonstrated that the carboxy terminus of the L protein contains the P-protein binding site. This again is in contrast to the data obtained with paramyxoviruses that showed that the P binding site is located in the amino-terminal portion of L for simian virus 5 (23), Sendai virus (6), and measles virus (20). This discrepancy reveals again that binding sites or conformations of the polymerases of related viruses differ in this regard.