Rescue of Defective Poliovirus RNA Replication by 3AB-Containing Precursor Polyproteins

Abstract

This study demonstrates the in vitro complementation of an RNA replication-defective lesion in poliovirus RNA by providing a replicase/polymerase precursor polypeptide [P3(wt) {wild type}] in trans. The replication-defective mutation was a phenylalanine-to-histidine change (F69H) in the hydrophobic domain of the membrane-associated viral protein 3AB. RNAs encoding wild-type forms of protein 3AB or the P3 precursor polypeptide were cotranslated with full-length poliovirus RNAs containing the F69H mutation in a HeLa cell-free translation/replication assay in an attempt to trans complement the RNA replication defect exhibited by the 3AB(F69H) lesion. Unexpectedly, generation of 3AB(wt) in trans was not able to efficiently complement the defective replication complex; however, cotranslation of the large P3(wt) precursor protein allowed rescue of RNA replication. Furthermore, P3 proteins harboring mutations that resulted in either an inactive polymerase or an inactive proteinase domain displayed differential abilities to trans complement the RNA replication defect. Our results indicate that replication proteins like 3AB may need to be delivered to the poliovirus replication complex in the form of a larger 3AB-containing protein precursor prior to complex assembly rather than as the mature viral cleavage product.

Poliovirus (PV), the prototypic picornavirus, replicates its genomic RNA via membranous replication complexes within the cytoplasm of an infected cell (11, 12). These complexes appear as rosette-like structures (8) and are thought to provide an environment for increased local concentrations of poliovirus proteins, presumably limiting diffusion within the replication complex (22). Highly specific interactions between cellular and viral proteins associated with virion RNA (vRNA) and cellular membranes result in the formation of PV replication complexes. These interactions include tight membrane-protein associations by PV proteins 3AB (17, 40, 45, 47, 52) and 2C (15, 20, 48) combined with specific RNA-protein interactions between the PV 5′ noncoding region (5′NCR) and the viral polypeptide 3CD (2, 3). In addition, the presence of the 5′ terminal ∼100 nucleotides (nt) of PV RNA mediates an in vitro interaction between 3CD and the cellular protein poly(rC) binding protein 2 (21, 36). The viral polypeptide 3AB (24, 54) and the cellular protein EF-1α (24) also appear to complex with 3CD and the 5′ end of the PV genome. Interactions between the viral proteins 3D and 3AB have been documented (25, 29, 37), as have RNA-protein interactions involving the 3′ ends of positive- and negative-strand picornavirus RNAs (4, 39, 50). While many of these studies have focused on individual molecular interactions, little is known about the viral polyprotein subunits necessary for the initial assembly of the vRNA replication complex. For PV, like all picornaviruses, mature gene products are specifically processed from viral precursor polyproteins (42). The efficient and highly regulated protein processing cascade produces cleavage products with functions distinct from those of their precursor proteins. Due to the short half-lives of large precursor proteins, it has been difficult to determine the composition of assembly intermediates.

An additional control of RNA replication relies on the translation of PV RNA; that is, translation of a particular genome is a prerequisite for that genome to be competent for replication due to the requirement for either a cis-acting viral protein(s) or ribosomal passage over the RNA template (34). A relationship between ribosomal passage and RNA replication stems from the observation that all naturally occurring defective interfering particles contain deletions in the capsid region (P1 [Fig. 1]) that maintain the reading frame (28). The requirement for translation prior to RNA replication was later confirmed by using genetically engineered replicons in which replication occurred only if deletions were in frame and the P2-P3 region was left intact (except for the N terminus of viral protein 2A) (16). The apparent cis dominance of translation over replication does not in theory preclude the utilization of some of the viral nonstructural proteins in trans. For example, passage of the ribosome could transiently expose a cis-acting replication determinant present on the RNA template, while the active replication proteins in the complex could consist of mixed viral proteins, some of which were synthesized from other viral mRNAs.

FIG. 1.

Schematic representation of full-length and subgenomic PV RNAs. (A) Schematic representation of the PV genomic structure and viral protein organization where the viral structural proteins are derived from the P1 portion of the genome and the nonstructural proteins are derived from the P2 and P3 portions of the genome. Contained within the PV1 5′NCR are the sequences that comprise the IRES. Note that the region of the genome designated VPg is also referred to as 3B. (B) Representation of the full-length and genetically engineered subgenomic RNAs used in this study. The PV1(wt) and PV1(F69H) RNAs contain a nonviral guanosine residue at the 5′ end of SP6-transcribed RNAs. The coding sequences for any of the P3 or 3AB proteins (derived from subgenomic RNAs) are preceded immediately by the PV1 IRES (modified as described in Materials and Methods). An × denotes the approximate location of the point mutation. Note that each P3-encoding RNA contains 18 nt of the 3′NCR immediately downstream of the authentic stop codon in 3D, a result of the 3′NCR primer used in the original amplification of the P3 sequences prior to cloning.

Complementation of site-specific lesions in trans has been assayed by providing the wild-type gene product(s) through a helper virus (6, 7, 14, 18, 22, 26, 32, 49) and through novel approaches using dicistronic RNAs or amber-suppressing cell lines (13, 34). The collective results of such studies indicate that each viral protein likely has multiple functions, some of which can be complemented in trans and some of which cannot, depending on whether a mutation exerts its effect at the level of the precursor protein or at the level of the mature cleavage product.

Previous experimental approaches often utilized a helper virus or helper RNA to provide the complementing gene products in infected or transfected cells. vRNAs produced in cells from such approaches most likely generate individual RNA replication complexes that are physically separated from each other in the cytoplasm. By the time viral proteins are generated to levels sufficient for diffusion and effective complementation, polypeptides in the replication complexes may be processed or assembled such that they are no longer competent for subunit exchange. In this study, we attempted to circumvent the problem of inaccessibility to the PV replication complex by using a HeLa cell-free replication assay to produce potential complementing viral proteins during the initial stages of complex formation. We addressed four questions. (i) What polyprotein subunit delivers the RNA replication function of viral protein 3AB during assembly of the replication complex? (ii) Can this 3AB-containing assembly intermediate by provided in trans? (iii) Does this assembly subunit need to be proteolytically active? (iv) Does 3AB need to be physically linked to the active viral RNA polymerase during complex assembly? Our results show that a 3AB mutation causing a severe RNA replication defect can be efficiently complemented in trans by providing the large replicase precursor (P3) but not the mature 3AB polypeptide. Furthermore, the rescuing trans-P3 protein must contain an active 3C proteolytic domain but not an active polymerase domain. Possible mechanisms for complementation are discussed.

MATERIALS AND METHODS

Plasmids and cloning.

Construction of plasmid pSP6-PV1 was the result of a five-fragment ligation in which pT7-PV1 (23) was first digested with StuI and BglI, yielding three fragments: 1142-nt StuI to BglI (pBR322 nt 3480), BglI (pBR322 nt 3480) to BglI (PV nt 5318), and BglI (PV nt 5318) to BglI (PV nt 35). These three fragments were then gel purified and joined by T4 DNA ligase in the presence of two synthetic oligonucleotide cassettes (cassette 1 contained the oligonucleotide 5′-CCTATTTAGGTGACACTATAGTTAAAACAGCT-3′ annealed to the oligonucleotide 5′-GTTTTAACTATAGTGTCACCTAAATAGG-3′; cassette 2 contained the oligonucleotide 5′-CTGGGGTTGTACCCACCCCAGAGGCCCACG-3′ annealed to the oligonucleotide 5′-GGGCCTCTGGGGTGGGTACAACCCCAGAGCT-3′) which together contain the first 35 nt of poliovirus type 1 (PV1) placed immediately downstream of the SP6 promoter sequences such that a single guanosine residue separates the SP6 promoter and the first nucleotide of the PV1 sequence.

To produce significant levels of either 3AB or precursor polypeptide P3 in trans, sequences encoding these proteins were cloned into a phage T7-based transcription plasmid immediately downstream of a modified PV1 internal ribosome entry site (IRES) in which sequences present in the variable region (nt 641 to 722) not essential for efficient translation initiation (1, 27) were removed and the ATG at nt 743 was placed within an NcoI site to facilitate cloning. To generate such plasmids [pT7-5′NCR-P3wt, pT7-5′NCR-P3(F69H), pT7-5′NCR-3AB(wt), and pT7-5′NCR-3AB(F69H)], plasmid pT7-5′NCR (23) was digested with StuI and EaeI (PV nt 627), and the ∼650-nt fragment containing the T7 promoter and the first 627 nt of the PV 5′NCR was gel purified. In parallel, vector sequences were prepared by digesting pT7-5′NCR with StuI and EcoRI. Half of the digested vector was incubated with the Klenow fragment of Escherichia coli DNA polymerase (generating a blunt end at the EcoRI site), and the resulting 2.3-kb fragments were gel purified. The DNAs encoding both 3AB (wt [wild type]), P3(wt), and P3(F69H) sequences were amplified by PCR from pT7-PV1(wt) or pT7-PV1(F69H) DNA, using the oligonucleotide 5′-AGATCCATGGGACCACTCCAGTATAAA-3′ (which encodes an NcoI site surrounding the start codon [underlined]) along with either the oligonucleotide 5′-GACTGAATTCTATTGTACCTTTGCTG-3′ (for 3AB), which contains an EcoRI site immediately following the stop codon (underlined), or the oligonucleotide 5′-TGTACTCGAGGACTGAGGTAGGGTTACT-3′ (for P3), which contains an XhoI site 18 nt downstream of the natural stop codon in 3D^pol (underlined). To clone the 3AB sequences into plasmid pT7-5′NCR, the 3AB-encoding PCR fragment was then digested with both EcoRI and NcoI, gel purified, and incubated in the presence of T4 DNA ligase along with the ∼650-nt StuI-to-EaeI fragment, the ∼2.3-kb StuI-to-EcoRI (not end filled) fragment, and an oligonucleotide cassette consisting of the oligonucleotide 5′-GGCCATCCGGTGAAATCAGACAATTGTATCAC-3′ annealed to the oligonucleotide 5′-CATGGTGATACAATTGTCTGATTTCACCGGAT-3′. The oligonucleotide cassette contained the PV1 5′NCR sequences from PV1 nt 627 to the ATG at PV1 nt 743 (placed in the context of an NcoI restriction site) with PV1 nt 641 to 722 removed. To clone the sequences for P3(wt) downstream of the modified PV1 IRES, the P3-encoding PCR fragment was treated like that for the 3AB DNA except that the DNA was digested with XhoI (blunt ended with Klenow enzyme) instead of EcoRI and the ∼2.3-kb vector fragment was blunt ended as well at the EcoRI site prior to ligation with T4 DNA ligase. To clone the K61L mutation in 3D^pol from the pExcalibur plasmid described by Richards et al. (38), pExcalibur was digested with BglII and PvuII, releasing a 1,452-nt fragment (corresponding to the sequences encoding the C-terminal half of 3C plus the N-terminal half of 3D), which was then gel purified. This 1,452-nt fragment was incubated in the presence of T4 DNA ligase with gel-purified pT7-5′NCR-P3(wt) that had been digested with PvuII and BglII and gel purified. The cloning of the C147A mutation in 3C to generate pT7-5′NCR-P3(C147A) was performed by digesting a 3CD expression vector (35) with BglII and AccI followed by the purification of the 617-nt DNA fragment encoding the 3C mutation. This 617-nt fragment was then cloned into pT7-5′NCR-P3(wt) that had been digested with AccI and BglII and gel purified.

RNA transcriptions.

Prior to transcription, the full-length pSP6-PV1 clones were linearized with EcoRI, while the pT75′-NCR-P3 and pT75′-NCR-3AB clones were linearized with AatII, followed by end-fill repair using Klenow enzyme. Individual RNA transcription reactions were performed as described by Charini et al. (14), with the modifications described in Towner et al. (52).

Coupled in vitro translation/replication assays.

Assays were performed essentially by the method described by Todd et al. (51) in which the HeLa initiation factor was prepared as described by Brown and Ehrenfeld (9) and the HeLa S10 extract was prepared as described by Barton et al. (5). Minor differences or new modifications in either the extract preparation or the translation/replication assay were as follows: (i) HeLa cells were resuspended in hypotonic buffer in a volume equal to 120% (vol/vol) of that of the cell pellet volume; (ii) each replication reaction mixture consisted of 51% (vol/vol) HeLa S10, 18% (vol/vol) initiation factor preparation, 21% (vol/vol) diethylpyrocarbonate-treated H₂O, and 10% (vol/vol) 10× mix containing (at 10×) 10 mM ATP, 2.5 mM GTP, 2.5 mM UTP (no CTP was added), 600 mM potassium acetate, 300 mM creatine phosphate (Boehringer Mannheim), 4 mg of creatine kinase (Boehringer Mannheim) per ml, and 155 mM HEPES-KOH (pH 7.4) (the composition of the 10× mix was previously described by Barton et al. [5]); and (iii) reaction mixtures (50 μl) were programmed with in vitro-transcribed RNA at a concentrations of 7.7 nM for full-length RNAs and 3.6 nM for P3 and 3AB mRNAs, unless otherwise specified. Each 50 μl of reaction mixture was divided into portions containing 10 μl and 40 μl. The 10-μl portions (containing an additional 10.5 μCi of [³⁵S]methionine [Amersham] at >1,000 Ci/mmol) were used for translation analysis, while each 40-μl portion was used for analysis of RNA synthesis. All reactions were incubated for 6 h at 30°C, at which time the translation reaction mixtures were diluted in Laemmli sample buffer, boiled, and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis through a 12.5% gel. The replication reaction mixtures were subjected to centrifugation at 15,000 × g for 15 min at 4°C and subsequently resuspended in 5 μl of buffer (50 mM HEPES [pH 8.0], 3 mM MgCl₂, 10 mM dithiothreitol, 0.5 mM each ATP, GTP, and UTP) containing 25 μCi of [α-³²P]CTP and incubated for an additional hour at 37°C. Total RNA was then extracted from each sample; the RNA was ethanol precipitated and, following centrifugation, resuspended in diethylpyrocarbonate-treated H₂O. Total RNA was then subjected to gel electrophoresis on a 1.1% agarose Tris-borate gel containing ethidium bromide. In an adjacent lane, ∼500 ng of vRNA was loaded as a marker to visualize the mobility of PV single-stranded RNA (ssRNA). For experiments in which guanidine HCl was used to inhibit RNA replication to synchronize RNA synthesis (method in reference 5), each replication reaction mixture was resuspended in a total volume of 25 μl of replication extract mixed together as described above (item ii) containing 25 μCi of [α-³²P]CTP.

RESULTS

Rationale for plasmid constructions and mutations.

To examine the mechanism of delivery of PV protein 3AB to the RNA replication complex, we focused on complementation of a replication-defective lesion in the putative VPg precursor protein, 3AB. The replication-defective lesion chosen for study codes for a phenylalanine-to-histidine change at amino acid 69 (F69H) in the hydrophobic domain of PV protein 3AB. The F69H mutation was originally designed to disrupt 3AB membrane association by introducing a charged residue in a putative amphipathic helix. The region of the hydrophobic domain containing the F69H lesion was later found to be nonessential for membrane association (52). However, yeast two-hybrid studies suggested that this lesion disrupted protein-protein interactions between 3AB and itself (homodimer interactions) as well as with the polymerase 3D^pol (55), suggesting that the mutation may disrupt subunit interactions during replication complex formation. When full-length RNA harboring the F69H lesion (a two-nucleotide change of UUC to CAC) was transfected into HeLa cell monolayers and incubated at either 33 or 37°C, 100- to 1,000-fold more RNA was required to see delayed but comparable levels of cytopathic effects relative to those seen for wild-type PV1 RNA. Sequence analysis of RNA isolated from mutant virus recovered following transfections of PV1(F69H) RNA revealed that the adjacent amino acid at position 70 was changed to either valine or threonine and that the original F69H mutation in the absence of a second-site reversion was never recovered (data not shown). This result indicates that the F69H lesion in 3AB causes a debilitating replication defect that results in a quasi-infectious RNA. Furthermore, in vitro translation of PV1(F69H) RNA showed no obvious defects in translation efficiency or proteolytic processing compared to that of the wild type. Our data suggested that the F69H mutation in 3AB caused a severe defect in viral replication at the level of RNA synthesis (see Fig. 3), making this lesion a good candidate for trans-complementation studies.

FIG. 3.

Inability of 3AB(wt) to rescue PV1(F69H). The overall format is like that described in the legend to Fig. 2 except that the RNAs used to program each parallel translation and replication reaction are PV1(wt), PV1(F69H), 3AB(wt) (36 nM), and 3AB(F69H) (36 nM). The mock controls are shown in lanes 7 and 8 of panels A and B, respectively, while reactions programmed with vRNA(wt) are shown in lanes 8 and 9 of panels A and B, respectively.

To maximize the cell-free replication efficiency of in vitro-transcribed RNA, the PV1 cDNA [including the full-length cDNA encoding the 3AB(F69H) mutation] was recloned (as described in Materials and Methods) under the control of the bacteriophage SP6 promoter instead of the bacteriophage T7 promoter. Our rationale for this recloning was that the T7 RNA polymerase requires two guanosine residues downstream of the phage T7 promoter element for efficient transcription initiation, resulting in the placement of two nontemplated guanosine residues 5′ of the authentic PV sequence. The SP6 RNA polymerase requires only a single guanosine residue for efficient transcription initiation leading to a more authentic in vitro transcript. Side-by-side comparisons using wild-type T7- or SP6-transcribed PV1 RNA revealed that the use of SP6-derived RNA results in a 10-fold increase in in vitro production of infectious virus in the HeLa cell-free viral replication assay (data not shown). It is noteworthy that in vitro translation/replication reactions programmed with SP6-transcribed RNA yielded approximately 1 log less infectious virus per ml of extract (data not shown) than those programmed with wild-type purified vRNA despite displaying approximately equal levels of translation. For a comparison of translation and RNA replication levels directly, see Fig. 3A (lanes 3 and 8) and 4B (lanes 4 and 9). The decrease in RNA replication in the SP6-PV1 programmed reaction (Fig. 3B, lane 4) relative to that programmed with vRNA(wt) (Figure 3B, lane 9) is likely due to the single nontemplated guanosine at the 5′ end of the transcript.

Finally, to gain insights into the possible mechanisms of complementation, a number of mutated versions of the P3 polyprotein were studied. Two mutations were used in the complementation analysis. (i) C147A is a mutation in which the active site cysteine of 3C has been mutated to an alanine, abolishing the proteinase activity of 3C (30). This 3C lesion in P3 is referred to as P3(3C*). (ii) K61L (also referred to as μ61) is a mutation in 3D which renders the RNA-dependent RNA polymerase completely inactive for strand elongation (38). This 3D mutation in P3 is referred to as P3(3D*). All of the plasmid constructs and mutations are outlined in Fig. 1.

Inability of 3AB(F69H) to exhibit trans-dominant effects.

We reasoned that if 3AB is a soluble component of the PV RNA replication complex, then cotranslation of a replication-defective form of 3AB might be able to inhibit the replicative abilities of wild-type RNA replication complexes. When either 3AB- or 3AB(F69H)-encoding RNA was cotranslated with PV1 vRNA(wt) (Fig. 2A, lanes 1 and 2), the overall levels of translation were decreased significantly but equally with respect to vRNA(wt) alone (Fig. 2A, lane 3). Reduced translation levels may be due to the increased concentrations of PV IRESs present when the two mRNAs are mixed together, suggesting that limiting levels of translation factor(s) are present in the extract. For reasons that are unclear, the mRNAs encoding 3AB consistently translated less efficiently than equimolar amounts of either PV1 RNA or P3-encoding RNA (data not shown). In the corresponding replication assays, no differences in the levels of ³²P-labeled single-stranded or replicative intermediate/replicative form (RI/RF) vRNA were observed when PV1 vRNA(wt) was incubated in the presence of either 3AB(wt) or 3AB(F69H) mRNA (Fig. 2B, lanes 3 and 4).

FIG. 2.

Inability of 3AB(F69H) to exhibit trans-dominant effects. (A) [³⁵S]methionine-labeled in vitro translations that were programmed either singly or doubly with 7.7 nM vRNA(wt), 36 nM 3AB(wt), or 36 nM 3AB(F69H) RNA. The sample shown in lane 9 shows the mock reaction that was not programmed with any RNA; lanes 7 and 8 show the translation of each of the 3AB RNAs during the 30-min preincubation. Note that the F69H mutation in 3A causes a distinct electrophoretic mobility difference which allows for the identification of 3AB(wt), if present, in the same reaction. Lanes 1 to 3 show reactions in which the mRNAs were cotranslated, while those in lanes 4 to 6 contained the presynthesized 3AB proteins. (B) The corresponding ²³P-labeled RNA replication reactions that were performed in parallel with the [³⁵S]methionine-labeled translations. Lane 2 was not programmed with any RNA. Lanes 3 to 5 show reactions in which the mRNAs were cotranslated, while those in lanes 6 to 8 contained the presynthesized 3AB proteins. Positions of RI/RF RNA and ssRNA are shown on the right. (C) Ethidium bromide staining of the agarose gel following electrophoresis of the replication reactions to demonstrate even loading of total RNA in each lane. The photograph was taken just before the agarose gel was dried and exposed to film. The mobilities of both 28S and 18S rRNAs are indicated along with the 7.5-kb vRNA marker (lane 1) used to show the mobility of full-length PV ssRNA.

Based on the experiments performed by Lundquist and Maizel (33), which showed that the functional RNA replication complex forms very early during the PV replication cycle, it is possible that there is a only a small kinetic window in which 3AB is able to undergo subunit exchange with a competent replication complex. To maximize the opportunity for 3AB exchange, we presynthesized either 3AB(wt) or 3AB(F69H) for 30 min prior to the addition of vRNA. The result of this experiment is shown in Fig. 2A, lanes 4 to 8; lanes 7 and 8 show the levels of 3AB generated during the 30-min preincubation period. The F69H mutation confers an obvious electrophoretic mobility difference from that of 3AB(wt). The translation and replication assays are shown in lanes 4 and 5 of Fig. 2A and lanes 6 and 7 of Fig. 2B, respectively. Equivalent RNA replication levels in the presence of 3AB(wt) or 3AB(F69H), either presynthesized or cotranslationally synthesized, demonstrate that 3AB(F69H) was not able to exert a trans-dominant effect.

3AB(wt) is unable to rescue the replication defect of PV1(F69H).

Since we were unable to detect the ability of 3AB(F69H) to effect a decrease in wild-type RNA replication, we attempted to measure 3AB complementation by assaying for 3AB(wt) trans rescue of the defective replication phenotype of the full-length PV1 transcript encoding the 3AB(F69H) mutation. The results of such an experiment in which either 3AB(wt) or 3AB(F69H) was cotranslated with PV1(wt) or PV1(F69H) RNA are shown in Fig. 3. The levels of [³⁵S]methionine-labeled viral proteins synthesized from either of the full-length PV1 transcripts were equivalently decreased when cotranslated with either of the 3AB mRNAs (Fig. 3A; compare lanes 1 and 2 to lane 3 and compare lanes 4 and 5 to lane 6); significant levels of 3AB proteins were produced in all of the translation reactions. Figure 3B shows the levels of ³²P-labeled virus-specific RNA synthesized in parallel replication reactions to those shown in panel A. Comparison of lane 5 to lane 6 reveals no detectable virus-specific RNA synthesis, indicating that the presence of 3AB(wt) in trans was unable to rescue the replication defect in 3A, a result documented even when long exposures of the autoradiographs were examined (data not shown). This result is consistent with the inability of 3AB(F69H) to exhibit a trans-dominant effect on PV1(wt) RNA replication. As seen in each RNA replication assay and the mock-treated reaction (Fig. 3B, lane 8), the abundant 28S and 18S rRNAs (migrating below the ssRNA species in Fig. 3B) are ³²P labeled, the likely result of an endogenous terminal transferase activity present in the HeLa cell extracts. Our data suggested that the F69H lesion may exert its effects at the level of RNA secondary structure or that protein 3AB may be unable to enter the replication complex unless provided as part of a larger precursor protein that is a normal component of complex assembly.

Complementation by P3(wt) but not 3AB(wt).

To test the hypothesis that the F69H lesion in 3AB exerts its effects in a precursor polyprotein larger than 3AB, we tested the ability of the 3AB-containing precursor protein, P3, to trans complement the replication defect of PV1(F69H). Figure 4A (lanes 3 to 8) shows that the levels of [³⁵S]methionine-labeled proteins synthesized from the full-length RNAs are all approximately equal (for reference, compare levels of protein 2A) when translated alone or in the presence of 3AB- or P3-encoding mRNA. Translations programmed with equimolar amounts of RNA for either 3AB or P3 in the absence of any full-length PV1 mRNA are shown in Fig. 4A, lanes 9 and 10. Production of 3AB (lane 9) can be seen in the long exposure in Fig. 4A. It is important to note that the P3 polyprotein is proteolytically active when translated alone (lane 10) or when cotranslated with PV1 mRNA (lanes 5 and 8), generating the cleavage products 3BCD, 3CD, 3AB, and 3A.

FIG. 4.

Efficient complementation by P3(wt) but not 3AB(wt). The overall format is like that described in the legend to Fig. 2 except that the RNAs used to program each parallel translation and replication reaction are PV1(wt), PV1(F69H), 3AB(wt), and P3(wt). The mock controls are shown in lanes 2 and 3 of panels A and B, respectively. A long exposure of the polyacrylamide gel is shown in the bottom portion of panel A to show the production of 3AB(wt) in lane 9. Lane 10 of panel A shows the translation of P3(wt) in the absence of any full-length PV1 RNA. In addition to the mock controls shown, PV1(wt) was translated and replicated in the presence of 2 mM guanidine HCl (A, lane 1; B, lane 2) to demonstrate that the radiolabeled bands marked “ss vRNA” and “RI/RF” are indeed specific products of PV RNA replication. The asterisks denote reactions carried out in the presence of 2 mM guanidine HCl.

Parallel ³²P-labeled RNA synthesis reactions are shown in Fig. 4B. As shown in this experiment, no inhibition or enhancement of RNA synthesis is seen when PV1(wt) mRNA is cotranslated with P3(wt) mRNA, and only a slight inhibition is seen when wild-type PV1 is cotranslated with 3AB(wt) (compare lane 4 to lanes 6 and 5). However, when PV1(F69H) mRNA is cotranslated with P3(wt) mRNA, a six- to sevenfold enhancement of ssRNA accumulation (quantitated by PhosphorImager [Molecular Dynamics] analysis) can be seen over that for PV1(F69H) alone (compare lane 7 to lane 9), demonstrating that P3(wt) is able to partially rescue the replication defect. In contrast, 3AB(wt) causes inhibition of RNA replication of PV1(F69H), which agrees with what was observed in the experiment shown in Fig. 3B. The reason for the general inhibition of RNA replication in the presence of 3AB, despite equal levels of translation, is not fully understood but has been observed at multiple 3AB mRNA concentrations (data not shown). As an additional control to demonstrate that the ³²P-labeled RNAs were indeed authentic products of in vitro PV RNA replication, we carried out a parallel reaction using PV1(wt) vRNA in which 2 mM guanidine HCl was added (lane 1 in Fig. 4A and lane 2 in Fig. 4B). This concentration of guanidine HCl is known to specifically inhibit PV RNA replication (10) without affecting translation.

Dose-dependent rescue by P3(wt).

Based on the ability of P3 to trans rescue a 3AB replication defect (Fig. 4), it was hypothesized that if intact or proteolytically processed P3 was able to undergo successful subunit exchange with the replication complex, then such rescue effects should be dose dependent. Therefore, increasing amounts of P3(wt) mRNAs were cotranslated with a constant amount of either PV1(wt) or PV1(F69H) mRNA (Fig. 5). The quantities of P3 mRNAs ranged from 0.9 to 7.2 nM, based on our unpublished observations that PV1 IRES concentrations of 10 to 15 nM decrease the overall levels of translation. As mentioned above, this may be due to the limiting levels of translation factors present in the cell extract. As shown in Fig. 5A, the levels of PV proteins generated from full-length PV1 RNAs were approximately equal over the entire range of cotranslated P3 mRNAs tested (lanes 3 to 10). However, the level of P3-specific products did increase with increasing amounts of P3 mRNA (compare 3CD, 3BCD, and P3 levels in lanes 3 to 6 and 7 to 10 of Fig. 5A). The results of the parallel RNA replication reactions (Fig. 5B) demonstrate a dose-dependent rescue of the PV1(F69H) RNA replication defect by P3(wt) proteins (lanes 8 to 11). A maximum level of ∼5-fold rescue of ssRNA was observed at 3.6 nM P3 mRNA (lane 10), a result consistent with that shown in Fig. 4 in which the same amount of P3 mRNA was used. These results indicate not only that P3 is able to complement the replication defect in trans but also that the primary defect of the F69H lesion in 3AB is at the protein level and not at the level of RNA secondary structure in the mutated genome.

FIG. 5.

Dose-dependent rescue of PV1(F69H) by P3(wt). The overall format is like that described in the legend to Fig. 2 except that the RNAs used to program each parallel translation and replication reaction are PV1(wt), PV1(F69H), and P3(wt). The mock controls are shown in lanes 2 and 3 of panels A and B, respectively; the guanidine HCl control reaction (described in the legend to Fig. 4) is shown in the lanes marked with asterisks. The amount of P3(wt) RNA added increased in range from 0.9 to 7.2 nM of message (0.9 nM for lanes 4 and 8, 3.6 nM for lanes 5 and 9, and 7.2 nM for lanes 6 and 10 of panel A; 0.9 nM for lanes 5 and 9, 3.6 nM for lanes 6 and 10, and 7.2 nM for lanes 7 and 11 of panel B). The reactions shown in lanes 1 to 3 and 7 of panel A and lanes 2 to 4 and 8 of panel B did not receive any RNA encoding P3(wt) mRNA.

trans dominant inhibition of RNA replication by P3(F69H).

If P3(wt) is able to trans rescue a replication-defective complex, then a logical prediction is that a mutated version of P3 will exert a trans-dominant effect on a wild-type RNA replication complex. To test this hypothesis, increasing amounts of either P3(wt)- or P3(F69H)-encoding mRNA were added to replication reactions programmed with a constant amount of PV1(wt) mRNA. The concentrations of P3 RNAs (1.8 to 7.2 nM) were similar to those chosen for the previous dose-dependent rescue experiment (Fig. 5). Since these cell-free replication reactions are capable of translating genomic PV RNA that is synthesized during the in vitro reaction (53), the primary 6-h incubations were carried out in the presence of 2 mM guanidine HCl to inhibit RNA replication. This guanidine reversal method (5) thus eliminates any amplification effects that might result from a trans-dominant phenotype of P3(F69H). It also greatly diminishes the possibility of RNA recombination which has recently been shown to take place in these cell extracts (19, 46).

The results of the trans-poisoning experiment are shown in Fig. 6. The data displayed in Fig. 6A show that the overall levels of translation from the PV1(wt) mRNAs are similar throughout the range of P3 mRNA concentrations tested (lanes 3 to 9). In Fig. 6B, however, the results indicate that increasing amounts of the P3(wt) mRNA have no effect on the overall levels of PV RNA replication (lanes 5 to 7), while a dose-dependent inhibition of RNA replication can be seen with increasing amounts of P3(F69H) mRNA (lanes 8 to 10). The dose-dependent poisoning of RNA synthesis by P3(F69H) ranges from 2.7-fold at 1.8 nM to 19-fold at 7.2 nM. The 19-fold decrease could be a slight overestimate due to the small decrease in translation seen in lanes 8 and 9 of Fig. 6A. These results are consistent with those shown in Fig. 4 and 5, and, taken together, indicate that P3 is likely to be a major subunit in the initial assembly of the RNA replication complex. Given the sensitivity of PV1(wt) replication to trans poisoning by P3(F69H) but not by 3AB(F69H), these data further indicate that the exchangeable 3AB-containing subunit is P3 and not 3AB. In addition, the data demonstrate that P3(wt) protein is able to complement replication-defective lesions in regions of 3AB previously thought to be cis acting (22).

FIG. 6.

trans-dominant inhibition of RNA replication by P3(F69H). The overall format is like that described in the legend to Fig. 2 except that the RNAs used to program each parallel translation and replication reaction are PV1(wt), P3(wt), and P3(F69H). The mock controls are shown in lanes 2 and 3 of panels A and B, respectively; the guanidine HCl control reaction (described in the legend to Fig. 4) is shown in the lanes denoted by asterisks. The amounts of P3 mRNA added to each reaction increased in range from 1.8 to 7.2 nM of message (1.8 nM for lanes 4 and 7, 3.6 nM for lanes 5 and 8, and 7.2 nM for lanes 6 and 9 of panel A; 1.8 nM for lanes 5 and 8, 3.6 nM for lanes 6 and 9, and 7.2 nM for lanes 7 and 10 of panel B). The reactions shown in lanes 1 to 3 of panel A and lanes 2 to 4 of panel B did not receive any P3(wt) or P3(F69H) mRNA.

Complementation by mutated versions of P3.

In an effort to further define the mechanism of P3(wt) rescue, mutated forms of P3 containing lethal mutations in the putative nucleoside triphosphate binding site of the PV RNA polymerase 3D (3D*) or in the proteolytic active site of the proteinase 3C (3C*) were tested. Both of these mutations (K61L in 3D and C147A in 3C) render these proteins completely inactive for their polymerase (elongation) (38) and proteinase (30) activities. The requirement for a physical linkage between the active viral polymerase and the putative carrier of VPg during RNA replication complex formation was tested by programming in vitro translation/replication reactions with PV1(wt) and PV1(F69H) mRNAs in the presence and absence of P3(3D*) or P3(wt) (Fig. 7A, lanes 3 to 5 and 7 to 9). A clear increase in some P3-derived proteins (e.g., 3CD) can be seen in lanes 4, 5, 8, and 9, but otherwise all levels of protein production appear equivalent. Translation of all three forms of P3 in the absence of any PV1 RNA can be seen in Fig. 7A, lanes 11 to 13. The corresponding RNA replication reactions are shown in Fig. 7B, lanes 4 to 6 and 8 to 10. In this experiment, a slight decrease in RNA replication can be seen when P3(3D*) is present in the wild-type RNA replication reaction (compare lanes 5 and 6). When P3(3D*) is cotranslated with PV1(F69H), we observe a significant rescue of RNA replication (Fig. 7B; compare lane 8 to lane 10) that is slightly less than the level of rescue effected by P3(wt) (Fig. 7B; compare lane 8 to lane 9). Overall, these results suggest that 3AB and the functional RNA polymerase do not need to enter the complex as part of the same polyprotein molecule given that the active polymerase was derived from the PV1(F69H) RNA. Furthermore, the ability of P3(3D*) to partially poison the PV1(wt) replication complex and diminish the rescue of the PV1(F69H) replication defect is consistent with previous studies suggesting that 3D^pol can be provided in trans to the viral RNA replication machinery (14, 34).

FIG. 7.

Differential complementation by mutated versions of P3. The overall format is like that described in the legend to Fig. 2 except that the RNAs used to program each parallel translation and replication reaction are PV1(wt), PV1(F69H), P3(wt), P3(3D*), and P3(3C*). P3(3D*) contains a K61L mutation that renders the polymerase inactive for chain elongation, and P3(3C*) contains a C147A mutation that renders the 3C proteinase domain inactive. The mock controls are shown in lanes 2 and 3 of panels A and B, respectively; the guanidine HCl control reaction (described in the legend to Fig. 4) is shown in the lanes denoted by asterisks. The amounts of P3 RNAs added were all at the same concentration of 3.6 nM (lanes 4 to 6 and 8 to 13 of panel A; lanes 5 to 7 and 9 to 11 of panel B). The reactions shown in lanes 1 to 3 and 7 of panel A and lanes 2 to 4 and 8 of panel B did not receive any P3-encoding RNA. Note that P3(3C*) (panel A, lane 13) shows no proteolytic cleavage, unlike P3(wt) and P3(3D*) (panel A, lanes 11 and 12).

To determine if intact P3 provides a replication function distinct from that of cleaved P3, the ability of a noncleaving P3 molecule [P3(3C*)] to rescue PV1(F69H) was assayed. The results of the translation assays are shown in Fig. 7A, lanes 6 and 10; those for the RNA synthesis reactions are shown in Fig. 7B, lanes 7 and 11. Comparison of lane 8 to lane 11 of Fig. 7B indicates that PV1(F69H) RNA replication was not rescued, suggesting that the F69H lesion does not act at the level of the entire P3 polyprotein. An interesting observation is that similar amounts of 3AB(wt) are generated from all three P3 molecules tested (Fig. 7A, lanes 8 to 10). Considering that P3(3C*) cannot cleave either in cis or in trans to generate 3AB, the generation of 3AB from the cotranslated P3 molecules must result from trans cleavage by proteinases derived from the full-length PV1 RNA.

DISCUSSION

The data presented in this study show that wild-type P3 polypeptide can trans complement a PV RNA replication defect caused by a single amino acid substitution (F69H) in the hydrophobic domain of protein 3AB. Surprisingly, no complementation was seen when we performed similar experiments in which 3AB(wt) was similarly provided in trans. The trans-rescue phenotype of P3(wt) is further corroborated by experiments in which P3 proteins containing the same F69H lesion exert substantial trans-dominant effects on a wild-type RNA replication complex. No such inhibition is observed when the same F69H mutation is present in the smaller P3 cleavage product 3AB and tested under similar conditions. In addition, the ability of P3(3D*) to successfully trans complement the defective replication complex indicates that the active RNA polymerase 3D^pol and 3AB (or 3A or 3ABC) need not be delivered to the replication complex from the same molecule. This result suggests that the mechanism of complementation is not as simple as providing more RNA polymerase to a functional but crippled replication complex since the extra RNA polymerase is itself nonfunctional.

Based on these results, the RNA replication defect in protein 3AB may be complemented because P3 provides a precursor molecule(s) to VPg. The precursor to VPg has been hypothesized to be 3AB (40, 43, 44), a lipophilic protein capable of tight membrane association (47, 52). Utilization of this mechanism would indicate that P3 supplies 3AB, or possibly an alternative VPg-containing precursor, and that the role of P3 in primary complex formation is to merely deliver the necessary VPg-containing P3 cleavage product. This latter hypothesis is consistent with all of the data presented in our study, including the observation that P3(3D*) and P3(wt) are able to rescue PV1(F69H).

It is possible that protein 3AB is not the true precursor of VPg. Rather, the precursor may be another VPg-containing polyprotein such as 3BCD or 3ABC (Fig. 1). If the VPg precursor is 3BCD, the data suggest that (i) 3BCD is not the precursor to the 3D RNA polymerase [due to the ability of P3(3D*) to complement] and (ii) the proposed anchoring function of 3A would be mediated through tight protein contacts and not via covalent attachment to 3B or other 3B-containing proteins. An observation that favors 3BCD (or 3BC) as the immediate precursor of VPg is that cleavage of in vitro-translated P3(wt) at the 3A-3B junction is dilution independent whereas cleavage of the 3B-3C junction is dilution dependent (53). These data are consistent with the inability of P3(3C*) to complement RNA synthesis of the 3A(F69H) lesion because this proteolytically inactive form of P3 would be unable to cleave in cis to generate 3BCD. However, it would be a suitable substrate for a dilution-dependent (trans) cleavage to generate the 3AB observed in the in vitro translation (Fig. 7A, lane 10).

Complementation of a lethal VPg mutation by 3AB(wt) was observed by Cao and Wimmer (13), who provided 3AB in trans by placing the 3AB coding sequence in the first cistron of a dicistronic RNA. The virus generated was genetically unstable, undergoing multiple genetic rearrangements. The results of this unique approach are difficult to compare directly to the results presented here; however, the fact that any virus was recovered at all is indicative of complementation. It should also be pointed out that the exact details of the defect conferred by the 3AB(F69H) mutation are not known, and therefore it is possible that the mechanism of VPg donation may not be perturbed at all, but that an alternative function of 3AB (or a larger 3A-containing polyprotein) is affected. The possibility that 3AB is involved in multiple functions is supported by the ability to complement an amino acid insertion in the N-terminal half of 3A (6) by using a helper virus, while a temperature-sensitive mutation in the hydrophobic domain of 3AB could not be similarly complemented (22).

There is precedent for precursor polypeptide functions distinct from those of mature cleavage products in the RNA replication activities of other positive-strand RNA viruses. Alphaviruses like Sindbis virus and Semliki Forest virus encode their replication proteins (including nsP4, the putative RNA polymerase) in the form of polyproteins encoded in the 5′ portion of their genomic RNAs. Data from cell culture studies using mutant viruses (or plasmids that encode RNAs with site-directed lesions) showed that the mature nsP4 polymerase and an uncleaved precursor polypeptide (P123) are required for synthesis of negative-strand intermediates in Sindbis virus-infected cells. However, virus-specific proteolytic processing of this P123 precursor polypeptide effects a switch from negative-strand RNA synthesis to positive-strand synthesis (31, 41). Based on these observations, one might speculate that the PV P3 replicase precursor is required for initiation of the synthesis of negative-strand intermediates and that mature 3AB polypeptides or other cleavage products function in the synthesis of positive-strand RNAs by using these newly synthesized intermediates as templates.

A model for replication complex assembly is depicted in Fig. 8. Figure 8A shows a scenario in which the PV proteins are fully processed prior to complex assembly. This model predicts that all of the freely interchangeable viral proteins capable of complementation are mature cleavage products. An alternative mechanism for complex assembly is shown in Fig. 8B. In this model, the RNA replication function disrupted by the 3AB(F69H) mutation is initially delivered to the replication complex by the P3 precursor protein. As a result, the mechanism of replication complex assembly would involve a multistep process in which the mature virus proteins are properly positioned only when delivered from defined precursor proteins in a prerequisite assembly step. This idea, together with the intricate protein processing cascade that generates the mature PV proteins, underscores the notion that the replication complex is an active and kinetically transient structure. The model in Fig. 8B does not rule out the ability of other functions of the same protein to interchange as mature viral cleavage products, an important consideration given that some of the nonstructural proteins have been shown to be both cis and trans acting.

FIG. 8.

Schematic representation of proposed mechanisms of replication complex assembly. (A) Following translation, viral proteins are co- and posttranslationally processed to generate mature virus proteins. The mature virus proteins then form the functional replication complex that is associated with cytoplasmic membrane structures (depicted as ladder-like forms). (B) Following translation, viral proteins undergo limited proteolytic cleavage, generating the larger precursor protein P3 and/or other precursor proteins such as 2BC-P3. These larger precursor proteins then interact to form a complex which subsequently undergoes further proteolytic cleavages to generate a functional RNA replication complex. As shown in this model, P3 is the precursor protein from which 3AB is derived.

In summary, this study provides important clues as to how the PV RNA replication complex may initially assemble. Much of the previous complementation data that suggested requirement for a viral protein acting in cis could be explained by the assembly mechanism outlined in Fig. 8B. The replication complex consists of a combination of mature and stable virus cleavage products mixed with viral protein precursors with short half-lives (in addition to template RNAs, possibly cellular proteins, and membranes). If a cis-acting lesion in a virus protein exerts its effects at the level of a transient precursor protein, then this model of replication complex assembly predicts that the kinetic window for genetic complementation will be limited due to the requirement of a series of ordered subunit interactions. In addition, the concentrations of such precursors will never reach (or sustain) the levels of the mature viral proteins, thereby limiting the complementation potential of a given precursor polypeptide. Based on this hypothesis, complementation might be enhanced by using PV mutants with retarded proteolytic processing kinetics where precursor assembly intermediates would have longer half-lives. Our model also predicts that when successful complementation does occur, the defective replication function must be rescued by an interchangeable protein unit. In the case of protein 3AB, the data suggest that the interchangeable protein unit is P3 and not the mature polypeptide.