A Role for the Sendai Virus P Protein Trimer in RNA Synthesis

Abstract

The SeV P protein is found as a homotrimer (P₃) when it is expressed in mammalian cells, and trimerization is mediated by a predicted coiled-coil motif which maps within amino acids (aa) 344 to 411 (the BoxA region). The bacterially expressed protein also appears to be trimeric, apparently precluding a role for phosphorylation in the association of the P monomers. I have examined the role of P trimerization both in the protein’s interaction with the nucleocapsid (N:RNA) template and in the protein’s function on the template during RNA synthesis. As with the results of earlier experiments (32), I found that both the BoxA and BoxC (aa 479 to 568) regions were required for stable binding of P to the N:RNA. Binding was also observed with P proteins containing less than three BoxC regions, suggesting that trimerization may be required to permit contacts between multiple BoxC regions and the N:RNA. However, these heterologous trimers failed to function in viral RNA synthesis, indicating that the third C-terminal leg of the trimer plays an essential role in P function on the template. We speculate that this function may involve the movement of P (and possibly the polymerase complex) on the template and the maintenance of processivity.

The paramyxoviruses are enveloped animal viruses containing a nonsegmented negative-stranded RNA genome. Together with the rhabdoviruses and filoviruses, they constitute the superfamily Mononegaloviridae. Sendai virus (SeV) is a prototype member of the paramyxoviruses and has served as a model system for examining the mechanisms that regulate viral genome expression. The ca. 15-kb genome is never found as free RNA inside an infected cell but rather is associated with the nucleocapsid protein (N protein) in the form of a helical ribonucleoprotein complex, the nucleocapsid (N:RNA). N:RNAs are 96% protein by mass (10) and serve as the templates for mRNA synthesis and genome replication. The SeV polymerase is a complex of two virally encoded proteins, the phosphoprotein (P) and the large protein (L) (17, 31). Upon entry into the cell, holonucleocapsids containing ca. 50 L and 200 P proteins (25) can initiate a virus infection in the cytoplasm by the sequential transcription of the six viral mRNAs (so-called primary transcription). As viral protein products accumulate, these same N:RNAs are used as templates for genome replication, via the synthesis of an antigenome N:RNA intermediate which is also assembled with N protein. Hence, the availability of unassembled N (N°) protein is considered to be a crucial element in regulating the switch between replication and transcription (reviewed in reference 24).

Paramyxovirus P genes are polycistronic, expressing proteins from at least two, and frequently all three, open reading frames (ORFs). The SeV P protein (568 amino acids [aa]) is expressed from the largest ORF on the P mRNA. In addition, a nested set of four C proteins (C′, C, Y1, and Y2) are translated from an ORF which overlaps the N terminus of the P ORF (in the +1 frame) by a mechanism of ribosomal choice during initiation (26). An internal V ORF (in the −1 phase) is accessed via the programmed insertion of a G residue at position 1053 during mRNA synthesis (cotranscriptional editing), which results in a ribosomal frameshift and the expression of V as a fusion protein with the N terminus of P (23). This same region of P is also found in another viral protein, called W, which is generated by the addition of +2Gs at the editing site, a reading frame switch that fuses only 2 aa to the N-terminal half of P (i.e., W is effectively the N-terminal half of P). Of all these P gene products, only the P protein is essential for genome amplification (see below). The C proteins are promoter-specific inhibitors of RNA synthesis and exert this activity indirectly, possibly through transient interactions with P or L protein (2, 6, 20, 35). Likewise, the V and W proteins are negative regulators of viral amplification, inhibiting replication but not transcription in a dose-dependent manner (4).

The SeV P protein is the central component of the viral replication machinery, forming (i) a complex with the L protein and (ii) a complex with N°. The L protein is the catalytic component of the viral polymerase (P-L). This complex serves to both stabilize the L protein (8, 33) and place the polymerase complex (P-L) on the N:RNA template via interactions with the exposed C-terminal tail of the assembled N protein (L alone is unable to interact with the N:RNA) (21, 27). The complex with N protein (P-N°) is the active form of soluble N protein used during assembly. Formation of this complex prevents nonspecific aggregation of the N protein, and consequently, P can be viewed as a chaperone for N° (7). Independently of stable complex formation with the L protein, P also has an additional role to play in viral RNA synthesis. This function involves, at least in part, the binding of additional copies of P to the N:RNA template (3). These supplemental P proteins may play a role during elongation (e.g., by enhancing the processivity of the polymerase or by modulating the conformation of the coiled N:RNA template and thereby facilitating reading of the bases).

Replication, which can be viewed as RNA synthesis plus concurrent assembly, can be reconstituted in vitro by combining extracts in which P-N (the assembly complex) and P-L (the polymerase complex) have been separately coexpressed (19). This two-part system has served to map domains on P involved in RNA synthesis and assembly (Fig. 1), and the results highlight the multifunctional nature of the P protein. Domains specific for RNA synthesis (interaction with the L protein and the N:RNA template) map to the C-terminal half of the polypeptide (aa 344 to 568) (8, 9), whereas a domain involved in formation of a stable complex with N°, and hence in nascent chain assembly, maps to aa 33 to 41 in the N terminus (7). Indeed, the entire N-terminal half of the P protein (aa 1 to 320) is dispensable for RNA synthesis (3). Likewise, the region required for stable complex formation with L (aa 412 to 445) is required for the RNA synthesis step of replication (with L) but not for the assembly process (with N) (8).

FIG. 1.

Schematic representation of the SeV P protein. The P protein (568 aa) is shown as a rectangle, with the various functional domains indicated by shaded boxes. Residues 33 to 41 are required for chaperoning N° during the nascent chain assembly step of genome replication. Two blocks within the C-terminal 40% of the protein (aa 344 to 411 and 479 to 568, previously designated segments A and C) are involved in binding to N:RNA. Segment A is required for trimerization (indicated as 3′mer), and only trimers bind to the template. Residues 412 to 445 represent the stable binding site for the L protein. The bent arrow above the diagram indicates the site at which the alternate C-terminal ORFs of the V and W proteins are fused to the N-terminal half of P by mRNA editing.

The SeV P protein is found as a homotrimer (P₃) when it is expressed in mammalian cells. Computer analysis of 13 paramyxovirus P proteins revealed a predicted coiled-coil region which may be aligned throughout the entire virus subfamily, and it appears that this region is sufficient for oligomerization (9). One exception is the Newcastle disease virus P protein, but recent experiments in my lab indicate that it too forms a stable homotrimer (32a), indicating that trimerization is a general property of all the paramyxovirus P proteins. In this paper, I examine both the role of trimerization in the interaction between P protein and the N:RNA template and its role in P function on the template during RNA synthesis.

MATERIALS AND METHODS

Construction of subclones.

The wild-type P (Pwt) and P deletion mutants used throughout this study were N-terminally tagged with the influenza virus HA1 epitope (12), and such tagging is indicated with a superscript “HA” before the protein designation. A detailed description of the construction of the tagged Pwt clone (pGEM-^HAP) and the deletion mutants is to be found in Curran et al. (7, 8). Throughout this paper, C- and N-terminal P deletion mutants are indicated with the amino acids that remain (e.g., P^78-568) whereas internal deletion mutants are indicated with the amino acids deleted (e.g., P^Δ344-411).

Purification of bacterially expressed P protein.

An N-terminally six-histidine (His₆)-tagged P (^HisP) gene construct (the His₆ plus an ATG start codon was positioned just upstream of the authentic ATG start codon of P) was inserted into the bacterial expression vector pT7-7 (pT7-7 ^HISP). Escherichia coli BL21 transformed with this construct was grown in L broth supplemented with 0.3% glucose at 37°C to an optical density at 600 nm of 0.7. Gene expression was then induced by the addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside), followed 90 min later by the addition of rifampin (200 μg/ml). Incubation was then continued for a further 4 h. Bacteria were pelleted and resuspended in lysis buffer (20 mM Tris-HCl [pH 8.0], 100 mM NaCl, 8 M urea). ^HisP was purified on a Talon metal affinity column (Clontech) according to the manufacturer’s instructions. The bound protein was eluted in lysis buffer containing 150 mM imidazole. The protein was renatured by a gradual removal of the urea (in 0.5 M steps) by dialysis against 100 mM NaCl, 20 mM Tris-HCl (pH 8.45), 1 mM EDTA, and 1% Nonidet P-40 (NP-40). The protein was concentrated by binding to a Hi-Trap Q column (Pharmacia) and elution in 300 mM NaCl plus 20 mM Tris-HCl (pH 8.45).

Purification of His-tagged proteins from transfected mammalian cells.

A549 cells infected with a vaccinia virus recombinant expressing T7 polymerase (vTF7-3 [13]) were transfected with plasmid pT7-7 ^HISP. Cytoplasmic extracts were prepared in 20 mM Tris-HCl (pH 8.0)–100 mM NaCl–0.6% NP-40. The His-tagged proteins were purified on a Talon metal affinity column (Clontech) according to the manufacturer’s instructions. Bound proteins were eluted in 20 mM Tris-HCl (pH 8.0), 100 mM NaCl, and 0.6% NP-40 buffer containing 150 mM imidazole. They were then dialyzed against RM salts (100 mM HEPES [pH 7.4], 50 mM NH₄Cl, 7 mM KCl, 4.5 mM magnesium acetate, 1 mM dithiothreitol [DTT]).

N:RNA binding assay.

Binding of P and P deletion mutants to the N:RNA was monitored essentially as outlined in reference 30). Briefly, cytoplasmic extracts prepared from A549 cells transfected with the plasmids indicated in the figure legends were mixed with 1 μg of SeV core N:RNA (isolated by purification on linear CsCl gradients). After incubation on ice for 60 min, N:RNAs were recovered by pelleting through 50% glycerol–TNE (10 mM Tris-HCl [pH 7.4], 30 mM NaCl, 1 mM EDTA) at 16,000 × g for 1 h at 4°C. The presence of N:RNA and bound P was confirmed by immunoblotting with anti-N monoclonal antibody and a monoclonal antibody to an epitope of the influenza virus HA1 protein, designated 12CA5 (12), herein referred to as anti-HA monoclonal antibody. As a negative control, a duplicate assay was performed in the absence of N:RNA.

In vitro RNA synthesis.

RNA synthesis in vitro was performed essentially as described in reference 8 with the modifications outlined in reference 3. N:RNA nondefective templates were isolated from infected egg allantoic fluid (strain Z) by banding twice on 20 to 40% CsCl gradients. Templates were resuspended at a concentration of ca. 250 ng/μl in TE (10 mM Tris [pH 7.4], 1 mM EDTA) containing 1 mM DTT–10% glycerol and stored at −70°C.

To isolate the P-L complex, A549 cells (5-cm-diameter petri dish) were transfected with 2.5 μg of pGEM-^HAP and 1.0 μg of pGEM-L. Cytoplasmic extracts were prepared by solubilizing the cells in 250 μl of lysis buffer (150 mM NaCl, 50 mM Tris-HCl [pH 7.4], 10 mM EDTA, 0.6% NP-40), and P-L complexes were immunoprecipitated with 1 μl of anti-L monoclonal antibody. Antibody complexes were recovered by the addition of 100 μl of a 50% suspension of protein A-Sepharose beads (Pharmacia) equilibrated in RM salts. This mixture was then incubated for a further hour at 4°C, after which the Sepharose beads were recovered by pelleting and washed three times with RM salts. The beads were then resuspended in 100 μl of transcription buffer (100 mM HEPES [pH 7.4], 150 mM NH₄Cl, 4.5 mM magnesium acetate, 1 mM DTT, 0.5 mM ATP-CTP-UTP, 40 U of creatine phosphokinase per ml, 1 mM creatine phosphate). In vitro RNA synthesis was generally carried out in 250-μl reaction mixtures containing 25 μl of the bead suspension, 5 μl of the template, 100 μl of vaccinia virus-T7 (vac-T7)-infected A549 cell extract, 30 μCi of [α-³²P]GTP, 20 μg of actinomycin D per ml, and purified P protein (see legend to Fig. 5) at 30°C for 3 h. After the reaction, 500 μl of lysis buffer (150 mM NaCl, 50 mM Tris [pH 7.4], 10 mM EDTA, 0.6% NP-40) was added and the RNA was recovered by pelleting through 5.7 M CsCl. Products were analyzed directly on 1.5% agarose–HCHO gels.

FIG. 5.

Mixed P trimers do not function in RNA transcription. A549 cells were transfected with ^HisP or doubly transfected with Pwt plus ^HisP^1-445. Tagged proteins were isolated on a Talon metal affinity column and dialyzed against RM salts containing 1 mM DTT. The yields of homotrimer (^HisP) and mixed trimer (^HisP^1-445-Pwt) were estimated by immunoblotting with a polyclonal anti-P antibody, and the final volumes were adjusted such that the concentration of protein in each lane was the same. (A) Cytoplasmic extracts from A549 cells (5-cm-diameter petri dish) transfected with pGEM-^HAP (2.5 μg) and pGEM-L (1 μg) were immunoprecipitated with an anti-L monoclonal antibody, and immune complexes were recovered by the addition of protein A-Sepharose beads. The beads were pelleted and washed three times with RM salts containing 1 mM DTT before being resuspended in 100 μl of transcription buffer. RNA synthesis was produced by mixing these P-L antibody complexes (Ab-P/L) with nondefective core N:RNAs in either (i) extracts from mock-transfected cells supplemented with no trimer (lane 1, -ve ctrl) or 25, 50, 100, and 140 μl of either the purified ^HisP homotrimer (lanes 2 to 5, respectively) or the ^HisP^1-445-Pwt mixed trimer (lanes 6 to 9, respectively) or (ii) vTF7-3-infected cells mock transfected (lane 10, vacT7) or transfected with ^HisP (lane 11) or ^HisP^1-445 (lane 12) in the presence of [³²P]GTP (these combinations are represented schematically in the upper diagram). Reactions products were pelleted through 5.7 M CsCl and resolved on a 1.5% agarose–HCHO gel. (B) The gel was quantitated on a phosphorimager and then plotted graphically. (C) The purified His-tagged proteins were also tested in an N:RNA binding assay as outlined in Fig. 3. Bound P protein was detected by immunoblotting with an anti-P polyclonal antibody (P^SDS) with a light detection system, and the proteins in the blots were quantitated with a densitometer. The results were then plotted graphically (bound protein in the mixed trimer is the sum of Pwt and ^HisP^1-445).

Immunoblotting and antibodies.

Anti-HA was obtained from the Berkeley Antibody Co. The monoclonal antibody N877 (29), which recognizes an epitope near the C terminus of N, was kindly provided by Claes Örvell, Stockholm, Sweden. The monoclonal antibody to a C-terminal peptide of the L protein (11), referred to as anti-L, was kindly provided by my laboratory. The anti-P polyclonal antibody (P^SDS) has already been described (4). Proteins were detected by immunoblotting with an enhanced chemiluminescence system (Amersham).

RESULTS

The bacterially expressed SeV P protein is trimeric.

^HisP was expressed in E. coli BL21. The majority of the protein was found to be insoluble (<10% of the protein remained soluble when the bacteria were lysed by sonication in the presence of 1% NP-40). The bacterial pellets were therefore resuspended in 8 M urea, and after purification (see Materials and Methods) the protein was renatured by a gradual removal of the denaturant by dialysis in the presence of 1% NP-40 (0.5 M steps). Failure to include nonionic detergent during renaturation resulted in the precipitation of the protein in the dialysis sac. Both the soluble and renatured ^HisP proteins were then analyzed by sedimentation, in parallel with cytoplasmic extracts from HeLa cells transfected with the same plasmid clone. As shown in Fig. 2, all proteins gave similar sedimentation profiles with a single peak in fraction 5. The bacterially expressed SeV P protein is thus also probably trimeric, even after denaturation and renaturation. The trimer appears to be the most energetically favored form of the protein under normal salt conditions and neutral pH.

FIG. 2.

Sedimentation analysis of bacterially expressed ^HisP protein. E. coli BL21 transformed with pT7-7 ^HISP was grown in L broth supplemented with 0.3% glucose at 37°C to an optical density at 600 nm of 0.7. Gene expression was then induced by the addition of 1 mM IPTG, followed 90 min later by the addition of rifampin (200 μg/ml). Incubation was then continued for a further 4 h. Bacteria were pelleted and resuspended in lysis buffer (20 mM Tris-HCl [pH 8.0], 100 mM NaCl) containing either 1% NP-40 (native conditions) or 8 M urea (denaturing conditions). ^HisP was purified and renatured as described in Materials and Methods. These proteins were centrifuged on linear 5 to 20% glycerol gradients (40,000 rpm, 22 h, 4°C in a SW60 rotor; see reference 9) along with a cytoplasmic extract prepared from vTF7-3-infected HeLa cells transfected with the same plasmid clone. Gradients were fractionated, and aliquots from each fraction were analyzed by immunoblotting with the anti-P1.180 monoclonal antibody with an enhanced chemiluminescence light detection system. The amount of protein in each fraction was determined by densitometry and plotted (sedimentation was from right to left). P1 refers to the position of ^HAP^Δ344-411 (a monomeric P protein [9]) sedimented on a parallel gradient.

Interaction with the N:RNA.

Ryan and coworkers (32) reported that two discontinuous regions of the P protein, namely BoxA (aa 344 to 411) and BoxC (aa 479 to 568) (Fig. 1), were important for N:RNA binding. These experiments were performed with proteins expressed in reticulocyte lysates; therefore, I decided to test their observations with proteins produced in the vac-T7 expression system. In addition to the deleted proteins examined previously (ΔA and ΔC), I also tested constructs in which the N-terminal chaperone domain for N° (aa 33 to 41) had been deleted (P^78-568 and P^145-568). Apart from the monomeric ^HAP^Δ344-411 (ΔA), all the constructs tested formed stable trimers, as estimated from their sedimentation on linear glycerol gradients (9). Figure 3 confirms that both the BoxA and BoxC regions are required for stable interaction with the N:RNA, whereas the N-terminal chaperone region of P is dispensable for this property. However, the region defined as BoxA also contains the coiled-coil motif required for P protein oligomerization (9). Therefore, rather than regions BoxA and BoxC folding to form a single N:RNA binding surface in a monomeric P protein (as suggested by Ryan and coworkers [32]), trimerization may be required either to permit contacts between multiple BoxC regions and the N:RNA or to induce a conformational change in the BoxC region necessary for stable P interaction with the N:RNA. I therefore decided to examine whether P₃ interaction with the N:RNA required three BoxC regions. For this examination, I expressed independently ^HAP, ^HAP^1-445, and ^HAP^1-538 and coexpressed ^HAP with ^HAP^1-445, ^HAP with ^HAP^1-538, and ^HAP^1-445 with ^HAP^1-538. N:RNA binding assays were then performed, and the proteins present in the pellet were visualized by immunoblotting with a combination of anti-N and anti-HA monoclonal antibodies. Neither of the C-terminal deletion mutants bound to N:RNA, either when they were expressed alone or when they were coexpressed. However, both deleted proteins were detected in the pellet when each was coexpressed with ^HAP (Fig. 4B). Both these deletion mutants form homotrimers and can cooligomerize with ^HAP (7, 9), suggesting that mixed trimers containing only one or two BoxC regions can still bind to N:RNA. Quantification of the blots showed that the pellets contained ca. three times as much ^HAP as ^HAP^1-538 and ca. twice as much ^HAP as ^HAP^1-445, even though the levels of the proteins in the cytoplasmic extracts were either equal (as with ^HAP and ^HAP^1-568) or the deleted protein was in slight excess (as with ^HAP and ^HAP^1-445) (Fig. 4A). If the two proteins are expressed at similar levels and one assumes that there is a random assortment of the monomers, then the intracellular ratio of the trimeric forms (α₃ to α₂β₁ to α₁β₂ to β₃, where α is ^HAP and β is ^HAP^Δ) is 1:3:3:1. If one BoxC region within a heterotrimer is sufficient for stable N:RNA binding (stable being defined as the ability of the protein to remain bound after traversing a glycerol cushion), then the ratio of ^HAP to ^HAP^Δ in the pellet would be 4:3, whereas if two BoxC regions are required, this ratio would be 3:1. Therefore, the ratio of proteins observed suggests that two BoxC regions are required for stable binding of P₃ to the N:RNA.

FIG. 3.

Binding of P to the N:RNA requires the BoxA and BoxC regions but not the N° chaperone domain. A549 cells (5-cm-diameter petri dish) infected with vTF7-3 were transfected with 2.5 μg of either ^HAPwt (aa 1 to 568), ^HAP^78-568, ^HAP^145-568, ^HAP^Δ344-411 (ΔA), or ^HAP^1-538 (these mutants are depicted schematically above panel A). Cytoplasmic extracts were prepared in 250 μl of RM salts containing 1 mM DTT. (A) Cytoplasmic extract (50 μl) was mixed with either 1 μg (5 μl) of core N:RNA (+) or 5 μl of TE (−) and incubated on ice for 60 min. N:RNAs were recovered by pelleting through 50% glycerol–TNE, and the presence of tagged P protein in the pellet was confirmed by immunoblotting with an anti-HA monoclonal antibody. (B) The steady-state levels of the tagged P proteins present in the extracts were also evaluated by immunoblotting with the anti-HA monoclonal antibody. The slower-migrating forms represent a small fraction of the P trimer that remain intact under the conditions of the sodium dodecyl sulfate-polyacrylamide gel (indicated as P₃). Note that in the mutant ^HAP^Δ344-411 (ΔA), no P₃ band is visible, consistent with this mutant being monomeric (9). wt, wild type.

FIG. 4.

Mixed P trimers can bind to the N:RNA. (A) A549 cells were transfected with ^HAPwt (568), ^HAP^1-538 (538), or ^HAP^1-445 (445) or doubly transfected with the plasmid combinations indicated above the figure (these mutants are depicted schematically above panel A). (A) Cytoplasmic extracts were prepared in 250 μl of RM salts containing 1 mM DTT, and the steady-state levels of the proteins expressed were analyzed by immunoblotting with the anti-HA monoclonal antibody. (B) Cytoplasmic extract (50 μl) was mixed with either 1 μg (5 μl) of core N:RNA (+) or 5 μl of TE (−) and incubated on ice for 60 min. N:RNAs were recovered by pelleting through 50% glycerol–TNE, and the presence of N protein and tagged P protein in the pellet was confirmed by immunoblotting with a combination of anti-N and anti-HA monoclonal antibodies.

Binding of mixed trimers is not sufficient for function.

To determine whether a P₃ protein containing only two C-terminal regions (BoxC) was active in RNA synthesis, I tested whether mixed P trimers could substitute for Pwt in its supplemental function during transcription. This is technically more feasible than isolating P-L complexes containing mixed P trimers. A His₆ tag was fused to the C terminus of P^1-445 (^HisP^1-445). The ^HisP^1-445 construct was coexpressed with Pwt (nontagged) under conditions in which the Pwt protein was in steady-state excess over the level of the truncated form, so that the majority of the mixed trimers generated contained a single ^HisP^1-445. As a control, a His₆ tag was also fused to the N terminus of Pwt (^HisP). Tagged proteins were then selected from transfected cells with a talon affinity column (see Materials and Methods), and aliquots of the proteins eluted were analyzed by immunoblotting with a polyclonal anti-P antiserum. The amounts of ^HisP homotrimer and Pwt-^HisP^1-445 heterotrimer recovered from the column were estimated by densitometric scanning of the immunoblot. The nontagged Pwt protein bound to the talon column when it was coexpressed with ^HisP^1-445 but not when it was expressed alone (data not shown), consistent with the formation of mixed trimers. The relative ratio of the Pwt and ^HisP^1-445 proteins recovered from the doubly transfected cell extract ± the standard deviation was ca. 3.5:1 ± 0.4 (n = 4).

RNA synthesis was reconstituted by mixing nondefective N:RNA templates and the immobilized P-L polymerase complex (purified from free P protein by immunoselection with protein A-Sepharose beads coated with an anti-L monoclonal antibody; see Materials and Methods and reference 3) with increasing amounts of either the purified homo- or heterotrimer. In each reaction, 100 μl of mock-transfected cell extract was also added, as this markedly stimulates viral RNA transcription. De novo RNA synthesis was followed by the incorporation of [α-³²P]GTP, and the products were resolved on a HCHO-agarose gel (Fig. 5A). As controls, the immobilized polymerase complex and N:RNA template were mixed with cell extracts from mock-, ^HisP-, and ^HisP^1-445-transfected cells (lanes 10 to 12, respectively). This analysis confirmed that the ^HisP protein was active and that the ^HisP^1-445 homotrimer was inactive in this supplemental function of P. Titration of the purified ^HisP homotrimer progressively stimulated RNA synthesis (no plateau was observed), with the activity at the highest concentration corresponding to 60% of the activity observed in the ^HisP-transfected cell extract (Fig. 5B). In contrast, only a weak stimulation of transcription was observed with the purified heterotrimer (RNA synthesis was sixfold lower than that observed with ^HisP at the highest concentration of protein tested).

A trivial explanation for the failure of the His-tagged heterotrimer to stimulate transcription is that the presence of the histidine stretch at the C terminus of P^1-445 impeded interaction with the template. To test this, I repeated the N:RNA binding assay. Both the tagged homo- and heterotrimers bound the N:RNA template with similar efficiencies (Fig. 5C). These experiments indicate that although P trimers carrying less than three BoxC regions can interact stably with the N:RNA, all three C-terminal domains are required at least for the supplemental function of P in RNA synthesis.

DISCUSSION

In this paper, I have examined the role of SeV P trimerization in the protein’s interaction with the N:RNA template. Interaction with the template required both the trimerization domain (BoxA) and the C-terminal domain (BoxC). Ryan and coworkers (32) postulated that these two discontinuous regions of the P protein form a single surface that interacts with the N:RNA, thereby looping out the intervening BoxB region, which contains the L binding site. However, stable interaction with the N:RNA requires that a minimum of two BoxC regions interact simultaneously with the exposed C-terminal tail of the assembled N protein (1, 5, 18, 28), and this simultaneous interaction is achieved through a trimeric P protein. The N-terminal region of P involved in chaperoning N° is dispensable for N:RNA binding, which is consistent with the demonstration that the entire N-terminal half of P is not required for RNA synthesis (3).

Like the SeV P protein, the vesicular stomatitis virus (VSV) P protein is found as an oligomer, possibly a trimer (14), although there are clearly differences in the ways these structures form. The VSV P protein has only a weak coiled-coil prediction within the N-terminal 30 aa, and it appears that phosphorylation at Ser62 and Thr64 facilitates protein multimerization (15, 34). Only this multimeric form of the P protein is active in RNA synthesis, since it is in this form that the protein interacts both with L and with the N:RNA template (16, 34). The SeV P protein, on the other hand, is found exclusively as a trimer, whether it is expressed in mammalian, insect, or bacterial expression systems, which apparently precludes a role for phosphorylation in the association of the P monomers. This lack of phosphorylation is further supported by the demonstration that N-terminal P deletion mutants lacking all the known sites for P protein phosphorylation are also trimeric and active in RNA synthesis (3). Even when made in reticulocyte lysates, conditions in which the concentration of the expressed protein is very low, SeV P is found only as a trimer. The inherent high stability of the trimer is further highlighted by the fact that the insertion of Pro (a helix-destabilizing amino acid) into the middle of the predicted coiled-coil A2 region (after aa 400) failed to perturb either P trimer formation or P function in RNA synthesis (data not shown). Furthermore, I have been unable to demonstrate subunit exchange within the trimer under the conditions of RNA synthesis (conditions under which exchange rates are very high for the VSV multimer [15]). Therefore, the association constant for SeV P trimerization is clearly much lower than that observed for VSV, and once formed, the P₃ appears to be stable.

Why a trimer? At least in part, an explanation can be found by considering P binding to the N:RNA. Monomeric P proteins containing the BoxC region do not bind, whereas trimers containing two such regions appear to interact stably with the N:RNA (Fig. 4B). The three BoxC regions may form a binding site that is altered when one region is deleted, leading in turn to an incorrect (and hence inactive) interaction with the N:RNA. Alternatively, as already suggested for the VSV P protein (16), each BoxC region may interact weakly with the tail of a single N monomer, this interaction then being stabilized by the multivalency of the P protein. Assuming that only two BoxC regions are required for binding, what then is the function of the third leg of the trimer? One possibility is that it interacts with a factor(s) important for RNA synthesis, e.g., a host cell factor, the polymerase L protein, or one of the viral nonstructural proteins that are known to modulate genome expression (the V and C proteins). Alternatively, the nonbound third leg may be involved in the processivity of the P protein on the template. In such a model, P is envisaged to “walk” on the template via the simultaneous breaking and reforming of subunit-template contacts (which is easily achieved, as each subunit interaction is weak; therefore, the energy barrier is quite low and the net reaction is isoenergetic) such that two “feet” of the trimer continuously engage the template, somewhat like a cartwheel (Fig. 6), to ensure processivity. The model explains why mixed trimers carrying only two BoxC regions appear to bind normally to N:RNA but do not function in RNA synthesis. Although in this study I have examined only the activity of the supplemental P protein in RNA synthesis, I assume that the P protein of the polymerase complex interacts with, and functions on the N:RNA, in a similar fashion. Indeed, all mutations that affect the supplemental function of P also affect its function in the polymerase complex, suggesting that the two activities involve similar protein domains (e.g., there may be an exchange of the template-bound P and the P of the polymerase complex during RNA synthesis [3]). Thus, the trimer permits the formation of multiple weak contacts, which in turn facilitates the modulation of protein-protein interactions. Such a characteristic is ideally suited for a multifunctional protein such as P, which must continuously modulate its interaction with other viral proteins.

FIG. 6.

A model for P trimer function on the N:RNA template. Stable interaction between P₃ and N:RNA involves contacts between two BoxC regions (depicted as ellipses at the ends of trimeric P proteins [bent lines]) and the exposed C-terminal tails of two assembled N proteins (depicted as rectangles on circles, the circles representing the structural core of N). The third leg of P₃ remains unbound and free. The trimer is visualized as moving along the N:RNA (indicated by arrows) by simultaneously making and breaking P-N contacts such that two feet of the trimer continuously engage the template to ensure processivity by cartwheeling. The model also proposes that binding of P (and possibly also P-L) opens the N:RNA structure so that the polymerase can interact with the bases (22).