INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The paramyxomyxovirus C proteins were first found in Sendai virus (SeV)-infected cells and, since they were apparently absent in virions, were termed nonstructural proteins (reviewed in reference 22). Subsequent work with specific antisera showed that the C proteins could be detected in virions, associated with nucleocapsids. They are, however, greatly underrepresented in virions relative to their intracellular amounts (22, 27, 38). Furthermore, unlike the other virion proteins, which are translated from separate transcriptional units, the C proteins are translated from the P (phosphoprotein) gene mRNA, from an alternate open reading frame (ORF) which overlaps the N-terminal end of the P protein ORF (14) (Fig. 1).

View larger version (31K): [in this window] [in a new window]   FIG. 1.   ORF organization and expression of the SeV P gene. The three ORFs expressed as proteins (P, C, and V) are shown as horizontal boxes, drawn roughly to scale (top). A blow-up of the 5' end of the mRNA is shown in the middle, and the five ribosomal start sites in this region are indicated. The mutations used to eliminate expression from the Cprime and C protein start codons are shown. The effects of these mutations on C protein expression in BHK cells infected with a 1:5 dilution of the wt (SeVZ), [C-minus], [Cprime-minus], and double-mutant virus stocks (see the text) were examined by immunoblotting with a polyclonal rabbit anti-C antiserum (bottom). aa, amino acids.

The expression of the paramyxovirus C proteins can also be complex. SeV is a member of the newly renamed Respirovirus genus (formerly the Paramyxovirus genus) of the subfamily Paramyxovirinae. This subfamily also includes the Morbillivirus (e.g., measles virus) and the Rubulavirus (e.g., mumps virus and simian virus 5) genera. The more distantly related pneumoviruses, such as respiratory syncytial virus, are now classified in a separate subfamily (Pneumovirinae) of the Paramyxoviridae (24). The members of the Paramyxovirinae have been recently reclassified based on their P gene organizations, including the presence of a C ORF. In SeV-infected cells, the C proteins are relatively abundant and are the major products of this P gene on a molar basis. There are four independently initiated SeV C proteins (Fig. 1) expressed via a non-AUG start site (C' or C^prime), and there are both scanning-dependent (C) and scanning-independent (Y1/Y2) ribosomal initiation (reviewed in reference 7). However, morbilliviruses such as measles virus express only a single C protein, and in significantly smaller amounts than that of P (1, 3). Furthermore rubulaviruses do not contain a C ORF at all (32). The optional nature of this ORF is also apparent in other mononegaviruses. For the pneumoviruses, an overlapping ORF is absent in the P gene of respiratory syncytial virus, but the pneumonia virus of mice P gene does contain a C-like ORF, which expresses two small basic proteins (2). Similarly, for the more distantly related Rhabdoviridae, vesiculoviruses like vesicular stomatitis virus express two C proteins whereas lyssaviruses like rabies virus do not contain a C ORF of any size (30). We note that the term "C protein" refers only to the origin of its coding sequence with reference to that of SeV; it is not as yet clear whether the different proteins carry out similar functions.

The SeV C proteins (C^prime, C, Y1, and Y2) are relatively small basic proteins (175 to 215 residues), whose intracellular localization shows little or no specificity (27, 38). Only very small amounts of the C proteins are found bound to nucleocapsids intracellularly or in reconstitution experiments with transfected cell extracts (unpublished data); this presumably accounts for their vast underrepresentation in virions. Nevertheless, consistent with their presence on nucleocapsids, the C proteins appear to play a role in viral RNA synthesis. When mRNA synthesis is studied with transfected cell extracts containing the P and L proteins, elimination of C protein coexpression (by placing a stop codon just downstream of AUG^201/Y2 [P/C^stop]) was found to strongly increase mRNA synthesis in vitro. Moreover, its reexpression from a separate plasmid eliminated this increase (8). The ability of C to negatively affect viral RNA synthesis depended on its coexpression with P and L, and C may therefore act during formation of the P₃-L polymerase complex. Similarly, when SeV minigenome replication is reconstituted in transfected cells by using P genes which do (P/C^wt) and do not (P/C^stop) express C proteins, C protein coexpression was also found to be strongly inhibitory, in a promoter-specific manner. The genomic promoter (which also contains the start site for N mRNA synthesis) was particularly sensitive to inhibition by C protein, in contrast to genome synthesis from the antigenomic promoter (6). The C proteins also have the remarkable property of preferentially inhibiting the replication of minigenomes containing mutant promoters (26) or minigenomes which were not of hexamer length (31). In this respect, the C proteins appeared to function as viral fidelity (or stringency) factors. In infectious-virus recoveries from DNA, C protein coexpression (from the P gene support plasmid needed to initiate viral RNA synthesis) must also be repressed for a successful outcome (6, 12). It is possibly this fidelity function of C which needs to be repressed. The temporary lack of stringency of the viral polymerase may be beneficial here, since our full-length clone expresses an antigenome transcript with three extra G residues at its 5' end and is therefore not of hexamer length.

More recently, the SeV C proteins have been found to play a positive role in infections of mice (a natural host). These conclusions were based on studies of SeV^MVC, an avirulent cell culture-derived mutant of the highly virulent SeV^M, which differs from SeV^M by only two substitutions; F170S in the protein encoded by the C gene (silent in the P ORF) and E2050A in the protein encoded by the L gene (18, 37). To examine which amino acid substitutions were responsible for the avirulence, rSeV^Z carrying the C gene of either the virulent M strain (rSeV^Z-C^M) or the avirulent MVC strain (rSeV^Z-C^MVC) within the virulent Z strain background were prepared. Only the chimeric rSeV^Z carrying the C^MVC gene was found to be avirulent, suggesting that the normal function of the C proteins was required for virulence in mice, i.e., for multiple cycles of virus replication (13). These experiments offered the first evidence that the SeV C proteins provide a function required for virus replication, at least in mice.

This paper reports the preparation and characterization of rSeV strains which specifically do not express the C^prime, C, or C^prime plus C proteins, by mutating their ribosomal start codons. Our results suggest that (i) the functions of the C proteins are complex, in that they display both positive and negative effects on viral RNA synthesis during the course of the infection, and (ii) the functions of the individual C proteins must be partly different.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell cultures and viruses. LLC-MK2 and BHK cell lines were grown in minimal essential medium supplemented with 5% fetal calf serum. Wild-type (wt) and rSeV stocks were prepared in the allantoic cavity of 9-day-old embryonated chicken eggs and harvested after 3 days at 33°C. Titer determination and plaque isolation of rSeV were carried out for 3 to 9 days at 33°C in LLC-MK2 cells covered with a 0.3% agarose overlay containing 1.2 µg of acetyl-trypsin per ml. The 50% egg infective dose (EID₅₀) of each viral stock was determined by injecting serial dilutions into the allantoic cavity of hen eggs as above. The presence of virus in the allantoic fluids was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (10% polyacrylamide) and staining with Coomassie brilliant blue on material pelleted at 10,000 × g for 30 min through a TNE (10 mM Tris-Cl, pH 7.4, 100 mM NaCl, 1 mM EDTA)-25% glycerol cushion in an Eppendorf centrifuge. To determine the 50% tissue culture infective dose (TCID₅₀), 4-cm petri dishes of BHK cells were infected in duplicate with serial dilutions of the virus stocks, and the cytopathic effect was monitored visually until 40 h postinfection (p.i.). At this time, cytoplasmic extracts were prepared and the levels of N protein contained in three different amounts of the extracts were determined by Western blotting with monoclonal anti-N⁸⁷⁷ antibody (25).

Generation of mutant SeV (DNA) genomes. pGem-P/C carrying TCT^81/C' and GCG^114/C mutations in the initiation codons of the P/C mRNA have been described previously (8). pGem-P/C combining these mutations was constructed by fusion PCR on pGem P/C-TCT^81/C' DNA. The product was digested with SacI (in the upstream polylinker) and EagI (position 1005 on the P/C mRNA) and then subcloned into pGem P/C-TCT^/C'. Sendai viruses containing the various mutations in the C protein start sites were recovered from DNA by taking advantage of the recombinogenic nature of the vaccinia virus-based recovery system, as previously described (12, 13). Briefly, pFL3 (full-length DNA clone) was modified to contain a deletion encompassing most of the N and P genes (pFL3-1), such that a rSeV could not be generated simply by the recombination of pFL3-1 DNA with that of the individual pGEM-N and pGEM-P support plasmids. A shuttle vector (pN/P^Xho/M) containing the deleted region plus flanking sequences into which the mutated P/C genes were cloned was then constructed as follows. PCR was performed with a positive-sense primer carrying an XhoI site at position 69 (of the untranslated region of the P/C mRNA) and a negative-sense primer complementary to position 1178. The products were digested with XhoI and EagI and inserted into the same sites of pN/P^XhoI/M. The various mutated P/C genes were then recovered in viruses via recombination of the various pN/P^Xho/M vectors with pFL3-1. In all cases, the relevant regions were verified by reverse transcription-PCR and DNA sequencing.

Generation of rSeV with an alternate expression of C proteins. The recovery system was essentially as described previously (12, 13). Briefly, one 9-cm dish of HeLa cells (ca. 80% confluent) was infected with 3 PFU of vaccinia virus TF7-3 per cell (11). The cultures were transfected 1 h later with 1.5 µg of pGEM-L, 2.5 µg of pGEM-N, 4.5 µg of pGEM-^HAP/C^off (which do not express C proteins), 15 µg of pFL3-1, and 5 µg of one of the pN/P^Xho/M shuttle vectors carrying the desired mutation. At 24 h later, 1--D-arabinofuranosylcytosine (100 µg/ml) was added to inhibit vaccinia virus replication, and 24 h after that, the cells were scraped into their medium and injected into two or three 10-day-old embryonated chicken eggs. At 3 days later, the allantoic fluids were harvested and reinjected undiluted into eggs. For further passages, the stocks were generally diluted 1/1,000 before injection. The presence of viruses in the allantoic fluids was determined as above.

Primer extension analysis of viral RNAs. Total RNA was extracted from virus-infected cells with Trizol (Gibco-BRL) and analyzed by primer extension using with Moloney murine leukemia virus reverse transcriptase (Gibco-BRL) and 200,000 cpm of each of the following ³²P-5'- primers, which had been purified on sequencing gels: NP126 (5'-¹²⁶CGGCCATCGTGAACTTTGGC¹⁰⁷-3' [numbers refer to map coordinates of the entire genome]) for estimation of antigenomes and N mRNAs and L15,270 (5'-^15,270GAAGCTCCGCGGTACC^15,285-3') for genomes.

Immunoblotting. C protein levels were determined by Western blotting of cytoplasmic extracts of BHK cells infected with the various viral stocks, using a rabbit anti-C polyclonal antiserum with the CSPD light detection system (Boehringer). Immunoblotting was also carried out on viruses released from cell cultures, purified by being pelleted through a TNE-25% glycerol cushion from the medium of infected BHK cells, with monoclonal anti-N877 antibody (25).

Determination of mouse pathogenicity. Specific-pathogen-free 7-week-old (female) C57BL/6 mice (Iffo Credo) were infected intranasally with 10⁶ PFU of the various viruses. The mice were observed for symptoms and weighed daily. Groups of three mice were sacrificed on days 2, 5, 7, and 8 or 9, and the number of PFU present in 10% lung homogenates was determined on LLC-MK2 cells. The lung homogenates were also injected into 9-day-old embryonated hen eggs and passaged twice. BHK cells were infected with this allantoic fluid, and viral protein expression was monitored by immunoblotting for the N and C proteins. Total RNA was also extracted from the lung homogenates with Trizol (Gibco-BRL). cDNA was prepared with Moloney murine leukemia virus reverse transcriptase (Gibco) and amplified with the Expand high-fidelity PCR system (Boehringer) with the primers mentioned above. The PCR product DNA was sequenced.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Failure to recover a totally C-minus Sendai virus. Measles virus (MeV) expresses a single C protein from its P gene, and Radecke and Billeter (28) have recovered a rMeV which cannot express its C protein due to the introduction of a stop codon into this ORF. This rMeV-C^minus strain replicated normally in cell culture, indicating that the MeV C protein does not play an essential role in the cellular replication of this paramyxovirus, at least in cell culture. A similar situation has also been found for the more distantly related rhabdovirus vesicular stomatitis virus (21). We have also attempted to recover a rSeV which cannot express any of its four C proteins, by placing a stop codon in the C ORF just downstream of the Y2 start codon (ATG^201/Y2) to terminate all four C proteins (8). A full-length SeV genome expression plasmid containing the C^stop mutation was constructed by exchanging the SalI fragment of pFL3 (which includes the N-terminal region of the P/C gene) with one containing the C^stop mutation (Fig. 2). As a positive control, the SalI fragment was always exchanged with itself at the same time. Virus recoveries from pFL3 and pFL3-C^stop were carried out in parallel (by using the pGEM-P/C^stop support plasmid, so that the C^stop mutation of pFL3-C^stop could not be lost by recombination), and the resulting viruses were amplified in hen eggs. In four separate attempts, sufficient virus was present in the pFL3 recoveries that 60 µl of the second-passage allantoic fluid showed a strong pattern of SeV structural proteins when analyzed by SDS-PAGE and Coomassie blue staining. This corresponded to reference stocks containing ca. 10⁹ to 10¹⁰ EID₅₀/ml (Fig. 3). However, we were unable to detect virus from the pFL3-C^stop transfections by this test, even when 200 µl of sixth-passage allantoic fluid was examined.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Inability to recovery a SeV strain which cannot express any of its C proteins. The reconstruction of full-length SeV DNA in which the C ORF remains open (C-wt) or is closed just after AUG201/Y2 (C-stop) by exchanging SalI fragments is shown schematically at the top. The three pGEM support plasmids expressing N, P, and L, needed to recover infectious virus from the full-length DNA, are shown just below; "off" in the box representing the C ORF indicates that C either is not expressed or is not completely functional (see the text). A summary of the results of the virus recoveries obtained with both SalI fragments and three different pGEM-P support plasmids (stop, HA, and MVC [see the text]) is shown at the bottom. When virus was recovered from full-length DNA containing the C-stop mutation, this virus was used to infect BHK cells, and the presence or absence of C expression was examined by immunoblotting (bottom right). Y1 or Y2 protein expression was not detected in this infection.

View larger version (47K): [in this window] [in a new window]   FIG. 3.   Titer determination of the various SeV stocks. Egg allantoic fluid stocks of the various mutant SeV strains and wt virus (SeVZ) were subjected to titer determination by (i) plaque formation on MK2 cells with stocks prepared in Kobe and Geneva (Ge), (ii) TCID50 determination on BHK cells, and (iii) EID50 determination in hen eggs (top). All values listed are per milliliter. Relative amounts of virus proteins present in 600, 200, and 60 µl of allantoic fluid (AF) were determined by SDS-PAGE and staining with Coomassie brilliant blue (bottom). dm1 and dm2 are independently recovered double-mutant viruses (dm2 was subjected to titer determination [above]). The N proteins of the mutant viruses migrate slightly slower than that of SeVZ, as expected, because these viruses were prepared in the rSeVZ-NH background to mark them as recombinants.

rSeV expressing two or three C proteins. To eliminate the expression of individual C proteins, we mutated their start codons individually or in combination, as detailed in Fig. 1. The mutations were first constructed in the pGEM-P/C support plasmids, so that these genes could be controlled for their ability to express the various C proteins. Mutation of ACG^81/C' to TCT and ATG^114/C to GCG eliminated expression of the C^prime and C proteins respectively, as expected (data not shown [cf. Fig. 1]). Mutation of ATG^114/C to GCG introduces an Asp4Gly substitution in the P protein, but the only suitable P-silent mutation (ATG^114/C to ACG) is unreliable in this case. The Asp4Gly substitution had no noticeable effect on the ability of P to support minigenome RNA synthesis in transfected cells. Our attempts to eliminate Y1 and Y2 expression by simultaneously mutating ATG^183/Y1 and ATG^201/Y2 to ACG (both silent in the P ORF) were unsuccessful. These mutations had only a minor effect in reducing Y1 and Y2 expression. Further, their simultaneous mutation to GCG more strongly reduced but did not eliminate Y1 and Y2 expression (data not shown). The Y proteins appear to be initiated by an unusual mechanism, in which the codon-anticodon interactions at the ribosomal start site are considerably less important than at standard start sites (23). Hence, we have so far been unable to prepare rSeV strains which express C^prime and C but not Y1 and Y2.

Titers of the rSeV stocks. Allantoic fluid stocks suitable for high-multiplicity infections of cell cultures (5 × 10⁸ to 20 × 10⁸ PFU/ml) were routinely prepared in recoveries of [C^prime-minus] and [C-minus], as in the parallel rSeV^Z or rSeV^Z-N^H wt controls [Ge (MK2), Fig. 3]. These virus stocks also contained roughly similar amounts of virus proteins (data not shown; see Materials and Methods). [C^prime/C-minus] (the double mutant) was recovered twice (dm1 and dm2), and although fourth- passage stocks of these viruses contained considerable amounts of viral proteins (only ca. 10-fold less than the wt stock [Fig. 3]), we were unable to determine the titers of the double-mutant stocks in Geneva. In contrast to the three other rSeV strains, which formed clear plaques 1 to 2 mm in diameter by day 4, the double mutant did not form visible plaques by day 5, when our serum-free monolayers began to disintegrate. A fresh stock of dm2, however, did form very small plaques by day 7 in Kobe (Fig. 4), albeit with a titer of only 4 × 10⁶ PFU/ml, i.e., 3 log units lower than that for the wt control (6 × 10⁹ PFU/ml) (Fig. 3).

View larger version (86K): [in this window] [in a new window]   FIG. 4.   Plaque formation of recombinant wt and double-mutant virus. Serial dilutions of wt (rSeVZ-NH) and dm2Kobe stocks were allowed to form plaques for 9 days at 33°C on LLC-MK2 cells in the presence of trysin and the absence of serum. An example of the plaques formed by each virus and its dilution is shown.

Infection of cells in culture. Parallel 4-cm-diameter BHK cultures (ca. 10⁷ cells) were infected with 2 (MK2) PFU of wt, [C^prime-minus], or [C-minus] per cell and 2 (BHK) TCID₅₀ of dm2^Ge per cell [2 PFU of wt virus per cell is equivalent to 10 to 20 (BHK) TCID₅₀/cell (Fig. 3)]. Infected cultures were harvested at various times after infection, and the relative amounts of viral genomes, antigenomes, and N mRNAs present in equal samples of the total RNA (one-quarter of each dish) were estimated by duplex primer extension (Fig. 5A and B). Relative to the wt infection, the [C^prime-minus] and [C-minus] infections led to the accumulation of significantly more viral RNA, and this was most apparent in the [C^prime-minus] infections, where the accumulation occurred earlier and to higher levels. The double-mutant infection, in contrast, led to the accumulation of relatively little viral RNA by 20 h p.i. From 20 to 30 h p.i., however, a large increase in the amounts of the three viral RNAs occurred in this infection. In particular, N mRNAs had now accumulated to higher levels than in the wt and [C-minus] infections (but remained below those of the [C^prime-minus] infection) and the genome and antigenome levels approached those of the wt infection by 30 h p.i. N mRNA levels continued to increase until 40 h p.i. in the double-mutant infection and reached levels equivalent to the peak of the [C^prime-minus] infection, whereas the N mRNA levels decreased from 30 to 40 h p.i. infection in the three other infections. Similarly, genomes continued to accumulate from 30 to 40 h p.i. in the double-mutant infection, such that their abundance was equal to those in the [C^prime-minus] and [C-minus] infections at this time, which were all fourfold higher than in the wt infection at 40 h p.i. Independent isolates of all three mutant viruses (e.g., dm1 and dm2) behaved similarly in cell culture infections (data not shown).

View larger version (34K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   View larger version (23K): [in this window] [in a new window]   FIG. 5.   Accumulation of viral RNAs in the mutant and wt virus-infected BHK cells. Parallel 4-cm-diameter BHK cultures (ca. 107 cells) were infected with 2 (MK2) PFU or 20 (BHK) TCID50 of wt, [Cprime-minus], or [C-minus] per cell and 2 (BHK) TCID50 of dm2Ge per cell. Infected cultures were harvested at various times after infection, as indicated above the lanes. The relative amounts of viral genomes, antigenomes, and N mRNAs present in equal samples of the total RNA (from one-quarter of each dish) were estimated by duplex primer extension with NP126, whose 5' end is 126 and 70 nucleotides from that of the antigenome and N mRNA, respectively, and L15,270, whose 5' end is 114 nucleotides from that of the genome. (A) The extended primers were separated on an 8% sequencing gel. N* mRNA is presumably a degradation product of the N mRNA. (B) The intensities of the various bands were determined in a PhosphorImager, and their values are plotted in terms of genomes and N and N* mRNA. (C) Accumulation of genomes and antigenomes in cultures infected individually with 2 (BHK) TCID50 of each virus per cell (or 0.2 PFU of wt, [Cprime-minus] or [C-minus] per cell), as above. hpi, hours p.i.

View larger version (24K): [in this window] [in a new window]   FIG. 6.   Accumulation of viral N protein in the culture supernatants. The supernatants of the infected cultures in Fig. 5A and B were harvested at the times indicated. Virus was pelleted through a glycerol cushion (see Materials and Methods), dissolved in sample buffer, separated by SDS-PAGE (10% polyacrylamide), and immunoblotted with an anti-N monoclonal antibody. The intensities of the various bands were determined in a densitometer, and their values are plotted below the gels. hpi, hours p.i.

Infection of mice. Groups of 7-week-old, specific-pathogen-free C57BL/6 mice were inoculated intranasally with 10⁶ PFU of [C^prime-minus] or [C-minus], as well as SeV^Z as a virulent wt virus control. As an avirulent virus control, the same mice were also infected with SeV^MVC, which is avirulent for 3-week-old (outbred) ICR mice (18, 37). All the mice were weighed daily as a general indicator of their health (Fig. 7A), since loss of body weight is a useful indicator of SeV pathogenicity in these animals (20). Groups of three mice were sacrificed at various times p.i., and the virus titers (PFU on MK2 cells) in their lungs homogenates were determined.

View larger version (17K): [in this window] [in a new window]   FIG. 7.   Infections of C57BL/6 mice. Groups of 7-week-old C57BL/6 mice (weighing ca. 20 g) were infected intranasally with 106 PFU of either SeVZ (wt), SeVMVC, or the recombinant [C-minus] or [Cprime-minus]. Phosphate-buffered saline was used for the mock-infected mice. (A) The body weights of the mice were measured daily and are given as weight loss relative to the weight on day 0. Three mice from each of the infected groups were sacrificed at the times indicated, and the PFU present in 10% lung homogenates were determined on LLC-MK2 cells. (B) The titers for each of the three mice are shown individually. Triangles (below the limit of detection by this method [102 PFU/lobe]) indicate that injecting these homogenates into eggs and passaging twice does lead to detectable virus; ovals indicate that virus could not be detected by this procedure.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

It has been suspected for some time that the paramyxovirus C proteins play an important role in natural infections. In contrast to the V proteins (which are coded for in part by the other ORF that overlaps the P ORF [Fig. 1]) (32, 34), C proteins are expressed in all viruses of the Respirovirus and Morbillivirus genera. Furthermore, in long-term persistent MeV infections of the human brain (subacute sclerosing parencephalitis), where expression of the M protein and the cytoplasmic tail of the F protein are often lost by point mutations which close these ORFs, the C protein ORF has always remained intact (4). The paramyxovirus C proteins have nevertheless remained a riddle, since it has never been clear why some viruses express four C proteins and some express only one. More recently, characterization of the C gene of an avirulent mutant virus has provided evidence that these proteins are required for multiple cycles of replication in mice but dispensable for growth in hen eggs or cell culture (13). This latter property, however, is unlikely to represent the pressure maintaining the integrity of this overlapping ORF, especially in reference strains that have been passaged in eggs for countless generations. This paper provides more direct evidence that the SeV C proteins play an important role in virus replication at the cellular level, as well as hinting at why this virus expresses a nested set of four C proteins. We have been unable to prepare a totally C-minus virus which grows to detectable levels in eggs, even though we have recovered rSeV strains which cannot express their V and/or W proteins by the same approach (9, 10). We have, however, been able to prepare rSeV strains that cannot express their C^prime, C, or C^prime plus C proteins, by mutating their ribosomal start codons. The characterization of their infections of cell culture (and eggs) has added insight into the possible function(s) of these proteins.

In one respect, our results agree with previous findings that coexpression of either C^prime or C individually (along with the P and L proteins) leads to an inhibition of RNA synthesis in transfected cell extracts whereas coexpression of Y1 and Y2 was not inhibitory (8). These results suggested that one function of C^prime or C is to negatively affect viral mRNA synthesis. In this view, viral transcription is less inhibited in the [C-minus] infection than in the wt infection due to the absence of C, and the overaccumulation of Y1/Y2 cannot compensate for this absence. Similarly, transcription has apparently been even less inhibited in the [C^prime-minus] infections, presumably because C^prime is the more inhibitory of the two. However, in contrast to [C^prime-minus] infections, where the viral RNAs overaccumulate precociously, the simultaneous absence of both C^prime and C in the double-mutant infections leads to a delay in viral RNA accumulation. This suggests that C^prime and C also provide a positive function without which the onset of viral RNA accumulation is delayed. We presume that either the C^prime or C proteins (but not the Y proteins) can supply this required function in natural infections. Viruses unable to express either protein individually would have this function provided by the remaining C^prime or C protein and would not undergo a delay in viral RNA accumulation.

Although viral RNA synthesis is delayed in the double-mutant infection, once started, it quickly leads to the overaccumulation of the viral RNAs as seen in the [C^prime-minus] infection, consistent with the absence of the inhibitory C^prime (and C) protein. The remaining Y1 and Y2 proteins apparently cannot compensate for the absence of C^prime and C in this delay, but this may be misleading. It is presumably the presence of Y1 and Y2 expression in the C^prime and C-minus infection that accounts for the relative viability of the double-mutant virus with respect to rSeV-C^stop. Y1 and Y2 may to some extent also provide the required function, but less efficiently, accounting for the delay in the onset of viral RNA accumulation. Furthermore, we note that in contrast to transfected-cell systems, where (ectopic) C protein expression inhibits minigenome replication in a promoter-specific manner, we have seen no evidence for this in natural virus infections. Although mutation in the C gene (13, 18) or elimination of C^prime and/or C expression (Fig. 5) leads to significantly higher levels of mRNA, the ratios of antigenomes to genomes remain unaltered. The inhibitory effects of C may thus act primarily on mRNA transcription in natural virus infections, as opposed to acting on antigenome synthesis. We also note that the simultaneous ablation of all C protein expression (C^stop) does not adversely affect genome amplification in transfected cells (6, 31), as it does during natural virus infections. We presume that the positive function of C required early in natural virus infection is somehow bypassed in the transfected-cell system, where SeV protein expression is no longer under the control of the viral genome.

The view that C^prime and C provide a positive function for virus replication is supported by the specific infectivities of these virus stocks prepared in eggs. Viruses in which C^prime or C have been individually ablated grow only slightly less well in eggs than wt virus does (i.e., the titers of their stocks are only 4- and 10-fold lower by PFU and EID₅₀, respectively [Fig. 3]). The double-mutant virus stock, however, contains 3 log units less (MK2) PFU than does the wt virus and 6 log units less EID₅₀ (even though the amount of virus protein in these allantoic fluids is reduced by only 1 log unit). As before, either C^prime or C (but not the Y proteins) can presumably supply this required function (for growth in eggs), and so only the double mutant displays the highly debilitated phenotype. One explanation for the exaggerated effect on EID₅₀ is that amplification of the infectious unit is more exponential here and may require more cycles of replication to be detected. Infections that liberate fewer virus particles (Fig. 6) and with lower specific infectivities (Fig. 3) might be expected to be associated with a particularly low EID₅₀. Nevertheless, the double-mutant stocks would contain almost as many virions as the wt virus stocks would when eggs are inoculated with high doses, since fewer cycles of replication are needed to reach these levels.

In spite of the similar abilities of [C^prime-minus] and [C-minus] to be amplified in eggs, [C^prime-minus] accumulated to high levels in the lungs of these infected mice and remained highly virulent for this animal host (like SeV^Z) whereas infectious virus did not accumulate in the lungs of [C-minus]-infected mice, and this virus infection was strongly attenuated. These results further suggest that the C^prime and C proteins cannot be functionally equivalent in all respects, since C can apparently replace C^prime for virulence in mice whereas C^prime (or Y1 and Y2) cannot replace C for this property. This latter result is somewhat surprising, since C^prime contains all the sequences of C plus an 11-amino-acid N-terminal extension. This difference in virulence may be due to less [C-minus] than [C^prime-minus] being liberated during the initial infection of the mouse respiratory epithelium, similar to what is seen in BHK cells (Fig. 6), allowing the host antiviral defense mechanisms enough time to operate. Some of these antiviral defense mechanisms (e.g., natural killer cells) would presumably be absent from hen eggs.

The view of the SeV C proteins that emerges from this study is that their expression is complex because the different C proteins carry out nonoverlapping functions to some extent. For example, C^prime and C (but not Y1 and Y2) act as inhibitors of viral RNA synthesis, and C (but not C^prime, Y1, or Y2) is specifically required for virulence (multiple cycles of replication) in an animal host. Furthermore, some (but not all) of these proteins are also apparently required for an essential function(s) in virus replication at the cellular level, which remains to be elucidated. This functional complexity, due to a nested set of C proteins, is made possible by the translational gymnastics used by this mRNA to direct ribosomal initiation at the various start codons (7). C^prime is initiated at a non-ATG codon (ACG⁸¹), where the bases at positions +5 and +6, in addition to those at positions 3 and +4, play an important role (5, 15). P, V, and W (all ATG¹⁰⁴) and C (ATG¹¹⁴) are initiated by leaky scanning ribosomes, since mutating ACG⁸¹ to ATG ablates initiation from ATG^104/P and ATG^114/C. Initiation from ATG^183/Y1 and ATG^201/Y2, in contrast, does not appear to be due to a leaky scanning mechanism, since the translation of Y1 and Y2 in these constructs is either unaffected or enhanced (7). Rather, initiation from ATG^183/Y1 and ATG^201/Y2 appears to be due to a shunting mechanism, in which ribosomes are translocated from a region upstream of ACG⁸¹ directly to an acceptor site, which includes ATG^183/Y1 and ATG^201/Y2 (23). The tripartite leader of adenovirus late mRNAs mediates a similar shunting mechanism, and in this case shunting is promoted by heat shock (39). Translation of the four SeV C proteins does not respond to heat shock (data not shown). However, the relative amounts of the various P gene products sometimes differ widely in cell culture infections (e.g., Y1 and Y2 are sometimes undetectable [Fig. 2] and the ratio of C^prime to C varies severalfold). Translation of the four SeV C proteins might respond to other stress-related modifications of translational capacity, such as those induced by interferon (35). Different mechanisms of translational initiation (non-ATG codons, leaky scanning, and ribosomal shunting) may have different requirements for initiation factors. If so, the relative amounts of the various P gene products may be determined in part by the stress-related modifications of the host cell translational apparatus. These products would then be well placed to act as sensors of the host antiviral defense mechanisms, particularly those of the innate immune system, and to thwart this defense accordingly.

Our results suggest that C^prime and C provide a positive function for viral RNA accumulation early in intracellular virus replication but that they act as inhibitors at later times. Although these two properties of the C proteins act in opposite directions, it is possible that they are, in fact, the same property. Whereas the relative ratios of P, L, and the N:RNA template are constant throughout the virus life cycle, the C proteins are very abundant at the later stages of infection but are largely excluded from virions. Intracellular viral RNA synthesis thus begins with relatively little C in proportion to P, L, and N:RNA, and this ratio increases as the infection proceeds. The opposite effects of C early and late in infection could then be due to the different ratios of C to the other viral components involved in RNA synthesis. Exactly how C acts in modulating RNA synthesis is far from clear. The C proteins (as glutathione S-transferase fusion proteins) bind the L protein (17) but apparently act through P, since their inhibitory effects (on viral RNA synthesis in transfected cells) are relieved by overexpressing P and not by overexpressing L (8, 31). We have suggested that high concentrations of C act as viral fidelity factors by somehow causing the viral polymerase to be more selective in its choice of promoters and that this results in a less active enzyme (31). It is possible that the low concentrations of C found intracellularly at the start of the replicative cycle interact with the P-L polymerase in a similar manner but that these limited interactions have a positive (as opposed to a negative) effect on polymerase activity. If so, the trace amounts of these proteins found in virions could play an important role in the initiation of viral RNA accumulation in infected cells.

It is also possible that the C proteins exert their positive effects indirectly, by helping to form active polymerase complexes intracellularly prior to virion release. The vif protein (viral infectivity factor) is a late-gene product of most lentiviruses and acts during virus assembly to allow the formation of particles competent for the early steps of infection (reviewed in reference 33). The effect of vif is probably indirect, since (like the SeV C proteins) there is very little vif protein in virions. Human immunodeficiency virus type 1 strains with mutations in the vif gene efficiently produce virus particles, but these fail to carry out proviral DNA synthesis due to improperly packed nucleoprotein cores (16, 36). C^prime/C-minus SeV prepared in eggs have higher particle/PFU ratios than do wt or single-mutant virions (Fig. 3). The delay in viral RNA accumulation seen at the start of the double-mutant infections may then be due not only to the paucity of C proteins at this time but also to improperly assembled viral cores [P + L + N:RNA] contained in these virions.