INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Rinderpest virus (RPV) belongs to the Morbillivirus genus of the family Paramyxoviridae and is thus related to Measles virus (MV), Canine and Phocid (seal) distemper viruses, the cetacean morbillivirus, and Peste des petits ruminants virus. The disease it causes (rinderpest) has for decades been one of the most widespread and economically important diseases affecting cattle in Africa, the Middle East, and the Indian subcontinent. It has been eliminated from most of these areas in recent years, largely through the efforts of the European Union- and Food and Agriculture Organization-backed EMPRES program, with the aim of global eradication by 2010 (22). Like all of the morbilliviruses, there is only one serotype of RPV, yet wide variation can be found in the pathogenicity of field isolates, from those causing essentially 100% mortality to others in which infection causes barely detectable clinical signs (58, 64). The molecular basis of this variation is unknown.

All of the Paramyxoviridae have single-stranded RNA genomes of negative sense, with six viral genes, which are transcribed in order from a single promoter at the 3' end of the genome. From the second of those genes (the P gene), through utilization of more than one translation initiation codon and/or the introduction of one or more nontemplated residues to allow access to alternate reading frames, more than one protein is always produced, though the exact number and type of extra proteins vary both between and within genera. The expression of other proteins from overlapping reading frames was first shown in Sendai virus (SeV) (26, 55). SeV (15, 29), and Human parainfluenza virus type 1 (hPIV1) (8) express a set of four carboxy-coterminal proteins (C', C, Y₁, and Y₂), whereas the morbilliviruses express only a single C protein (6), and the rubulaviruses (e.g., Mumps virus and Simian virus 5 [SV5]), with the possible exception of La-Piedad-Michoacan-Mexico virus (7), do not express a C protein at all. The C proteins of MV (6) and SeV (67) have been reported to associate with the N and P proteins in infected cells, and the SeV C is found in purified virions (67). Other reports have suggested that neither C nor V of MV associates with other viral proteins (45). In vitro studies suggested that the SeV C protein specifically decreases transcription from the genome promoter (i.e., mRNA and antigenome synthesis) (9, 57), possibly through interaction with the L protein (33). Recombinant SeVs lacking expression of either C' or C are viable, grow as well as the wild type, and show the expected increase in viral mRNA levels (41); however, the double mutant grows more slowly (39, 41), and abrogation of expression of all four C/Y proteins results in a virus that is very severely disabled (39). An SeV mutation in the C protein has been reported to abolish pathogenicity in mice (25). MV lacking its one C protein grows normally in tissue culture lines (53) but not in peripheral blood leukocytes (21). The expression of the P protein and the V protein from viral mRNAs differing only by insertion of nontemplated bases was first shown for SV5 (59); in this group of viruses the genome codes for the V protein, while insertion of two G's is required to produce a mRNA from which the P protein is translated. Similar editing was subsequently shown in MV (10) and SeV (61) and shown to be a virus-encoded activity (61), possibly resulting from polymerase stuttering on the genome template during mRNA transcription (62). In SeV and the morbilliviruses, the P gene encodes the P protein directly, and insertion of a single extra G is required to produce a V-encoding mRNA. The V protein always shares the amino-terminal half of the P protein. In V, this is followed by a highly conserved motif containing seven cysteines which has been shown to bind zinc (44, 51) and also to be required for interaction of the V protein with damage-specific DNA binding protein (43). The V proteins of mumps and SV5 are found in virions (51, 56), but those of MV (27) and SeV (14) are not. This difference may be due to differences in the P proteins (which are not conserved as to sequence between these two groups of viruses); the N-terminal domain common to P and V proteins has been shown, in SeV (34) and SV5 (54), to interact with free N protein, i.e., protein that has not been incorporated into nucleocapsids, while SV5 V protein has also been shown to bind to nucleocapsids (50, 59). Both immunofluorescence (65) and biochemical (27) studies of MV V protein have shown it to be distributed throughout the cytoplasm, as is that of SeV (14), whereas the V proteins of hPIV2 (66) and SV5 (51, 52) contain nuclear localization signals.

Studies in vitro have shown that the SeV V protein inhibits genome replication but not mRNA transcription (13, 17, 33); recombinant SeVs that do not express the V protein show higher rates of genome synthesis, though only in some cell types (18, 19, 35, 36). In MV, on the other hand, the major effects of abrogating V expression were an increase in viral mRNA transcription and protein synthesis (60). In all cases, viruses lacking V protein appear to grow well in tissue culture but are defective in growth in animals (35, 60).

We have recently developed techniques to create recombinant RPV (rRPV) based on the most commonly used vaccine strain (1) and have used these techniques to study the utility of recombinant morbilliviruses as immunizing agents (2, 63), as well as creating chimeric viruses containing elements of RPV and other morbilliviruses (S. C. Das, M. D. Baron, and T. Barrett, unpublished data). Since the C and V proteins clearly have important roles in regulating viral replication and pathogenicity, we have also investigated the roles of the C and V proteins in RPV replication by creating viruses in which the expression of one or both of these proteins has been abolished. We report here the effects of these changes on the growth of the resultant mutant RPVs in tissue culture.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and virus. Cells were maintained and virus stocks were grown and titered as described elsewhere (1). Rescue of rRPV from cDNA was performed as previously described (1) except that transfection was performed using FuGENE6 (Roche Biologicals) or Transfast (Promega), as described by the manufacturer, and using a ratio of 6 µl of reagent per µg of plasmid DNA for both reagents. For measuring viral growth rate, B95a cells (2 × 10⁶ per 35-mm-diameter well), Vero cells, or primary bovine skin fibroblasts (5 × 10⁵ per well) were infected with rRPV at a multiplicity of infection (MOI) of 0.005 to 0.04 (depending on the experiment) for 1 h. The inoculum was removed, the cells were washed twice with medium, and 2 ml of medium was added to each well. Samples were taken immediately and at various times thereafter and stored at 70°C. After thawing and clarifying at 2,500 rpm for 10 min, the 50% tissue culture infective dose (TCID₅₀) of released virus was quantitated by standard methods. All results were normalized to an initial infection with 10⁴ TCID₅₀ per well.

Molecular biology and assays. Except where indicated, all DNA manipulation was by standard methods. Plasmids were cloned and grown in Escherichia coli JM109 or DH5 and purified on CsCl gradients or using Qiagen columns. The change of the C open reading frame (ORF) initiation codon was introduced into pKSP (1) by using unique site elimination mutagenesis (20). The XmaI-XmaI fragment containing the mutation was swapped with the corresponding fragment from pMDBNP (1) to give pMDBNPCE, to which were added the M and F genes from pKSMF (1). The ClaI-SunI fragment containing the whole N, P, and M genes and most of the F gene was then swapped for the corresponding segment from the full genome cDNA clone in pMDBRPV. The change to editing site was performed by using the Quick-Change mutagenesis protocol (Stratagene) on pMDBNP; the sequenced ApaI-KpnI fragment was then used to replace the same section in pMDBNP (to give pMDBNPVE) or pMDBNPCE (to give pMDBNPVCE), and the single and double mutations were assembled into the full-length genome as described above. The introduction of stop codons in the C ORF was done by using two-stage overlap PCR (33), modified by using the proofreading DNA polymerase Pfu (Stratagene) instead of Taq DNA polymerase. The template was either pMDBNPCE or pMDBNPVE, and the XhoI-KpnI fragments containing the mutations were removed from these plasmids and inserted into pMDBNP-PacSbf, a version of pMDBNP where PacI and SbfI sites have been introduced at the ends of the N and P ORFs, respectively. The ClaI and SbfI sites at the beginning of the N gene and end of the P gene were then used to transfer the mutations into pMDBRPV2C, a version of pMDBRPV containing the same restriction sites (which have been found to have no effect on virus growth in vitro).

Antibodies and immunoprecipitation. Rabbit antisera recognizing the RPV C protein were raised against bacterially expressed glutathione S-transferase fusion proteins, using the pGEX vector. The rabbit anti-V antiserum was the kind gift of D. Briedis, Department of Microbiology and Immunology, McGill University, Montreal, Quebec, Canada. Mouse monoclonal antibody 2-1 recognizing the RPV P protein was the kind gift of M. Sugiyama, Department of Veterinary Public Health, Faculty of Agriculture, Gifu University, Gifu, Japan. For immunoprecipitation studies, cells in six-well plates were infected with rRPV for 2 days, incubated for 1 h in Eagle's medium without methionine or cysteine, and then incubated for 2 h in 0.5 ml of the same medium containing 5 to 10 µCi of PROMIX (Amersham Pharmacia Biotech), a mixture of amino acids containing [³⁵S]methionine and [³⁵S]cysteine. The medium was removed, and the cells were lysed in 0.5 ml of lysis buffer (1% [vol/vol], Nonidet P-40, 50 mM Tris-Cl [pH 7.5], 0.5 M NaCl, 5 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 2.5 mM iodoacetamide, 100 ng each of leupeptin, pepstatin, antipain, and chymostatin per ml). The lysate was incubated first with nonimmune rabbit serum and protein A-Sepharose (Amersham-Pharmacia) for 1 to 2 h at 4°C, centrifuged to remove the protein A-Sepharose, and then incubated with the specific antiserum and protein A-Sepharose overnight at 4°C. Immunoprecipitation of P protein was routinely from one-fifth the amount of lysate used from which to immunoprecipitate V or C. Where the primary antibody was a mouse monoclonal antibody, 0.5 µl of rabbit anti-mouse immunoglobulin G (Dakopatts) was added in addition to bind the mouse antibody more efficiently to the protein A. Sepharose pellets were washed, and samples were prepared for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) as described elsewhere (3). SDS-PAGE was performed using the buffer system of Laemmli (40); gels were fluorographed using sodium salicylate (12).

Digital image processing. Original gel pictures (autoradiographs or fluorographs) or photomicroscopy transparencies were photographed using a Kodak DCS420 digital camera or scanned with a Linotype-Hell Saphir, transferred to a PowerMac 7500, converted to grey scale and clipped in Adobe Photoshop, laid out and labeled in QuarkXPress, and printed on a Kodak XLS 8600 printer. Gel images from the Chemi Doc system were exported as TIFF files, cropped to the appropriate region, and imported to ClarisDraw for labeling and printing as described above.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Introduction of mutations to abolish the expression of C or V protein. The three P gene ORFs that are utilized by the virus are illustrated in Fig. 1a. The AUG initiator codon for the P and V proteins has a suboptimal context (C at 3, rather than the preferred A or G [38]); the C ORF starts at the next AUG, 20 bases downstream (Fig. 1b). To abolish C expression, we first changed this latter AUG codon to ACG, this being the only change to that codon that would be silent in the P protein (Fig. 1b). Although the rescued viruses containing this mutation [RPV(CE) and RPV(VCE)] did not express C protein (see below), we were concerned that a reversion could occur through a single point mutation, and RNA viruses have a relatively high mutation rate, making such a reversion possible. In addition, ACG is used as the C' protein start codon in SeV (15, 29), and so it was possible that some C protein might be expressed from this ACG, even though the next base after the ACG codon in our mutant (C) is the least conducive to such usage (28). We therefore introduced two stop codons into the C ORF (again, with no change to the P protein), in addition to the change in the initiation codon (Fig. 1b), giving rise to recombinant viruses RPV C and RPV VC. In fact, no difference was seen in the behavior of viruses with the single change or with both changes to the C ORF, and data on their growth rates in tissue culture were pooled.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Mutations introduced into the P gene of RPV. (a) ORFs used in the RPV P gene. Positions of the ORFs for the P and C proteins are shown, as well as the V protein-specific (Vs) motif accessed by the cotranscriptional insertion of a nontemplated G residue at the editing site (marked). (b) Elimination of expression of the C protein. Sequences of P gene transcripts at the start of the ORFs for the P and V proteins (P/V) and the C protein (C) are shown, along with sequences of the start of the P/V and C proteins. The highlighted bases were changed to the base above, as described in the text. (c) Elimination of cotranscriptional editing and hence expression of the V protein. Sequences of P gene transcripts at the editing site are shown; all four highlighted bases were changed to the base above.

Expression of C and V proteins by mutant viruses. The effectiveness of the introduced mutations in preventing the expression of the C or V protein was assessed by immunoprecipitation of ³⁵S-labeled proteins from lysates of labeled infected cells. As can be seen in Fig. 2, all three P-gene-derived proteins were made in cells infected with ordinary RPV. No C protein was found in cells infected with RPV CE, VCE, C, or VC, and no V protein was found in cells infected with RPV V, VCE, or VC. To determine if the mutations that we had introduced into the editing site of the P gene had abolished nontemplated base insertion completely, we used a modification of the primer extension technique that we had used previously to examine P gene editing in different strains of the virus (30), in which a labeled primer is extended on isolated mRNA, or PCR products derived from mRNA, until the first A residue is incorporated. As the only source of adenine is ddATP, this terminates extension after only 10 to 12 bases. This technique should show one band for the normal unmodified P gene transcript, with an additional band for transcripts containing an additional base (Fig. 3a). Further bands will be seen for transcripts containing two or more nontemplated bases: such editing events are normally seen, to various extents, in morbilliviruses (5, 10, 30). Due to the mutations introduced into the P gene in order to abolish editing, the predicted size of the primer extension product from the unedited P(V) gene transcript is one base longer than that from the P gene of the unmutated virus (Fig. 3b). Although editing of transcripts from the P gene of normal RPV could easily be seen (including +2 and +3 bases), there was clearly no editing in viruses containing the V mutation (Fig. 3c), the single primer extension product being 29 bases long, the same length as the product from the V mRNA cDNA clone.

View larger version (48K): [in this window] [in a new window]   FIG. 2.   Immunoprecipitation of P, V, and C proteins from B95a cells infected with RPVs. Lysates of labeled infected cells were prepared as described in Materials and Methods from cells infected as indicated at the bottom. Samples of lysates were immunoextracted with mouse monoclonal anti-P protein (lanes 1, 4, 7, 10, 13, 16, and 19), rabbit polyclonal anti-C protein (lanes 2, 5, 8, 11, 14, 17, and 20), or rabbit polyclonal anti-V protein (lanes 3, 6, 9, 12, 15, 18, and 21). Immunoprecipitated proteins were analyzed by SDS-PAGE on 12% gels, and fluorographs of the gels are shown.

View larger version (51K): [in this window] [in a new window]   FIG. 3.   Primer extension analysis of editing events in P gene mRNA transcripts. (a) Illustration of how the primer is extended on transcripts from a normal P gene, which will contain three G residues for an exact copy or four (or more) G's if an editing event takes place. P* shows the 5' label (32P) on the primer. The incorporated ddA base that terminates primer extension is highlighted in bold. (b) As for panel a, but for transcripts from the mutated (V) P gene. Note that because of a TC change in the mutated mRNA sequence, the first A is incorporated one base later, so the minimum product from unedited mRNAs is 29 bases. (c) Results from experimental primer extension. Templates used: lane 1, cDNA clone from P type mRNA; lane 2, cDNA clone from V type mRNA; lane 3, RT-PCR product from RPV mRNA; lanes 4 and 5, RT-PCR products from two separate isolates of RPV V.

Growth of mutant viruses in tissue culture cells. We examined the replication rates of all four virus types in tissue culture using B95a cells, the monkey kidney-derived Vero cell line, or primary bovine skin fibroblasts. Although unmodified RPV could be grown to titers of 10⁶ to 10⁷ TCID₅₀/ml after only a few passages, we found that none of the mutants could be grown in large-scale culture to greater than 10⁵ TCID₅₀/ml, and the double mutant could barely be grown to 10⁴ TCID₅₀/ml. All growth curves therefore had to be multistep, with an initial infection at an MOI of approximately 0.01 to enable all viruses to be used at the same (or nearly the same) infectious dose. In B95as, we found that the initial growth rate of the virus was unaffected by the absence of the V protein, but there was a decrease in the final titer reached (Fig. 4a). In contrast, the initial growth rate of viruses lacking the C protein (RPV C or RPV VC) was significantly lower (Fig. 4a). Since the final titer reached is dependent on the relative rates of viral replication and decay (RPV is not particularly stable at 37°C, and newly made virus particles lose infectivity during subsequent culture of the infected cells), it is not surprising that the more slowly growing RPV C mutants also grew to a lower final titer. Identical results were seen in Vero cells (not shown). In bovine skin fibroblasts, the main difference observed (Fig. 4b) was that the double mutant grew more slowly than that lacking only the C protein, while the RPV V mutant continued to grow almost identically to the normal RPV. In these cells, it appears that the absence of the V protein is somehow more deleterious to viral replication in the absence of the C protein than in its presence.

View larger version (22K): [in this window] [in a new window]   FIG. 4.   Growth of mutant viruses in tissue culture. Growth rates of mutant viruses were determined under multistep growth conditions (MOI of ~0.01) for normal rRPV and the three mutant virus types in B95a cells (a) or primary bovine skin fibroblasts (b). The results from four (a) or two (b) experiments, each with two separate isolates of each virus type, are plotted.

View larger version (154K): [in this window] [in a new window]   FIG. 5.   Plaque morphology. B95a cells infected with normal RPV or each of the mutant virus types were grown under carboxymethyl cellulose and stained as described in Materials and Methods; individual foci of infection were photographed. Two representative foci of each virus type are shown. Bar = 200 µm.

Effects of defects in V or C expression on synthesis of viral RNAs. Although the V mutation had no detectable effect on viral growth rate in tissue culture, it was clear from the above results that this mutation did change the way the virus replicated, inasmuch as the balance between virus-induced cytopathic effect and viral protein-induced cell-cell fusion was altered in viruses carrying this mutation. As we were unable to grow virus to high titer and hence unable to infect at high MOI, we were unable to generate sufficiently strong signals in Northern blots to study the synthesis of viral RNAs at early time points (0 to 24 h postinfection [hpi]) when viral replication is still effectively in single-step mode (growth curves show that new virus is released from cells only at 20 to 24 hpi). We therefore used semiquantitative RT-PCR (35) to compare the four virus types with each other. Initial experiments showed that the amount of PCR product was proportional to the input RNA over a range of about 50-fold, though the optimum number of cycles depended on the maximum and minimum of the range of input RNA concentrations. RNA was isolated from cells infected at an MOI of approximately 0.005 at various times postinfection, and RT reactions were performed with primers specific for mRNA [oligo(dT)], genome sense RNA, or antigenome sense RNA. Primer pairs for PCR amplification of antigenome sequences included one primer from a promoter (non-mRNA) region. The results of these studies are shown in Fig. 6. We used the amplification of actin mRNA as an internal control to show that RNA from the same number of cells was included in each reaction mixture. The actin signal was in the linear range after 15 or 16 cycles, similar to the results found for cells infected with SeV at high MOI (35). To detect the RPV genome, however, 20 to 22 cycles were required, while antigenome amplification required about 28 cycles (Fig. 6). We consistently found less antigenome than genome in infected cells (compare the amounts of PCR product found for genome and antigenome at 22 cycles in Fig. 6a). The number of cycles required for detecting viral mRNAs depended on the gene, since RPV, like all members of the order Mononegavirales, shows a transcription gradient along its genome, with most transcription taking place from the N gene (the nearest to the genome promoter), followed in decreasing amounts by the P, M, F, H, and L genes, in order along the genome (3' to 5') (11). This arises, it is thought, because primary interaction of the viral polymerase with the genome can occur only at the genome promoter, and the polymerase has only a finite probability of initiating transcription from any one gene after it has finished transcribing and polyadenylating the previous one. Transcription from either the P gene or the H gene was clearly depressed in the absence of the C protein (compare P or H gene mRNA from RPV with RPV C and RPV V with RPV VC in Fig. 6), suggesting that the initiation or extension of viral mRNA transcripts was decreased in these mutants and that the C protein of RPV has a role in these processes.

View larger version (67K): [in this window] [in a new window]   FIG. 6.   Semiquantitative RT-PCR of viral RNAs. RNA was purified from B95a cells infected with 103.5 TCID50 of normal RPV or the three mutant virus types (C, V, and VC) for 0, 8, 16, and 24 h. Semiquantitative RT-PCR was carried out as described in Materials and Methods for 16 cycles (actin mRNA), 20 cycles (P mRNA), 22 cycles (genome and antigenome RNAs and H mRNA), or 28 cycles (antigenome RNA). The PCR products were analyzed on 1.3% agarose gels, and the amount of PCR product for the viral RNAs was determined by imaging the gels on a Bio-Rad Chemi Doc system. (a) Gel images; (b) quantitation of PCR product for genome, antigenome (from the 28-cycle data), P mRNA, and H mRNA from the images in panel a. The area under a notional plot of fluorescence signal was calculated for each PCR product band after removal of the lane background, and this value was plotted for each time point.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The V-specific domain of the V protein is conserved among almost all of the paramyxoviruses, although it is not universally expressed, being absent from hPIV1 and hPIV3 (24, 46). This domain has been shown to bind zinc in MV and SV5 (and hence probably in other V proteins) (44, 51) and is responsible for interaction of the MV V protein with several host cell proteins (45); the V proteins of SV5, MV, and hPIV2 (but not SeV) interact with damage-specific DNA binding protein and require this Zn binding domain to be intact in order to do so (43). The P protein itself, on the other hand, especially its amino-terminal half, is poorly conserved (5). Nevertheless, the amino-terminal part of the P protein has been shown to interact specifically with monomeric nucleocapsid protein in SV5 (54), SeV (17, 34), and MV (31), and this property is shared by the V protein. In SeV and the morbilliviruses, insertion of two G's at the editing site gives rise to a third protein called W or R, which is the P protein amino-terminal domain without the V protein zinc binding motif and which also is able to bind monomeric N protein. As can be seen in Fig. 3, a small but detectable amount of R-encoding mRNA is normally produced in RPV. In vitro studies have shown that V protein (or W/R) inhibits viral replication, but not mRNA transcription (13, 17, 34), at least in SeV. The binding of these proteins to N appears to prevent its interaction with the P protein (34), making the N protein unavailable for encapsidation; if one thinks of replication as being divisible into transcription plus encapsidation (13), an increase in available N should increase replication of the genome and antigenome or bring forward the point in the virus replication cycle where sufficient N protein accumulates in the infected cell to allow replicative rather than mRNA transcription to occur. As predicted from these observations, SeV that does not edit its P gene transcripts shows increased genome and antigenome synthesis (35), though this appears to occur only in some cell lines (19, 35). Similarly, increased antigenome accumulation is observed in glioblastoma cells infected with MV V (though genome levels were not determined in those experiments) (60), and we have shown here that RPV V (with or without the C protein) shows increased production of both genome and antigenome, i.e., increased transcription of the encapsidated viral RNAs. Although the increased replicative transcription will decrease the transcription of mRNA from any one genome, this may be offset, or more than offset, by the increase in the number of genome templates available. Exactly what final effect there will be on viral mRNA levels and hence on viral protein levels will depend on how efficiently the virus can utilize these templates, which will depend in turn on the general replication machinery of that strain of virus, and also possibly on the availability of host cell proteins necessary for viral transcription, and so on the host cell itself. Hence, a normally virulent strain of SeV shows a large increase in mRNA levels in the absence of V protein (35), while modified versions of the Edmonston vaccine strain of MV (60) and the RBOK vaccine strain of RPV (work presented here) show only a slight increase in mRNA levels when V protein is not expressed. However, given the increased level of genomes observed in our own studies, there is probably a decrease in mRNA production per genome template in RPV V and RPV VC. MV V (in glioblastoma cells) and RPV V (in lymphoblastoid cells) both show an increase in syncytium formation. This is unlikely to be due simply to a rise in expression of viral glycoproteins due to an increase in viral mRNAs, since the increases seen are small in both cases, but may reflect an additional change in the total cytopathic effect caused by the absence of the V protein which alters the balance among viral assembly, cell death, and cell-cell fusion. This balance is also cell type dependent, as we see syncytium formation easily in RPV-infected B95a cells but rarely in Vero cells.

Both SeV V and MV V have been shown to be defective in replication in infected animals (35, 60). This effect has been seen most markedly in the SeV studies (35), since the parental SeV is virulent in mice but the V form is avirulent, being rapidly cleared from the lungs of infected mice. The RPV strain that is the parent for all of our recombinants is itself avirulent, and so no major effect of the V mutation would be apparent unless the virus became so attenuated as to no longer form an effective vaccine. However, cattle infected with RPV V are effectively immunized against lethal challenge (unpublished observations), indicating that any change in in vivo phenotype caused by the absence of V protein expression is masked by the preexisting attenuation of the virus.

The C proteins of the Sendai group of paramyxoviruses and those of the morbilliviruses are not conserved as to sequence and therefore cannot be assumed to have the same function in the viral life cycle. SeV C proteins have been shown to suppress mRNA transcription (16), and in general transcription from the genome promoter (9), possibly through the interaction of the C protein with the viral L protein (33). As predicted, SeV defective in expression of either of the first C proteins (C' or C) showed increased viral mRNA levels (41); however, abolition of expression of both C' and C leads to a slow-growing virus with decreased viral RNA levels (39, 41). SeV that expresses no proteins from the C ORF (no C', C, Y₁, or Y₂) is almost nonviable (39).

Much less is known about the morbillivirus C protein. Like the SeV C proteins (67), it colocalizes with viral N, P, and L proteins in infected cells (6; M. D. Baron, unpublished data), suggesting that it is involved in some way with viral RNA transcription. MV and RPV which express no C protein are reasonably viable in most cell lines tested (references 21 and 53 and work presented here), unlike SeV. MV C grows normally in Vero cells but very poorly in peripheral blood leukocytes (21, 53); in the latter, levels of MV viral genome and N mRNA are both reduced in the absence of the C protein. RPV C is defective in growth in B95a, Vero, and primary bovine skin fibroblasts; genome and antigenome accumulation are like that of the parental virus, but there is a clear effect of the absence of the C protein on mRNA transcription, suggesting that the C protein may play a role in initiation, extension, or termination of mRNA transcription.

A noticeable feature of the data obtained with some of the new recombinant paramyxoviruses is how dependent the phenotype of mutations is on the host cell involved. In primary bovine skin fibroblasts, we found that the double mutant (VC) grew more slowly than RPV C, although both virus types grew at the same rate in Vero and B95a cells. The fibroblasts are slightly poorer hosts for RPV than Vero cells, as judged by the titer that RPV grows to in these cells (Fig. 4), and it may be that they lack some necessary host cell protein or have an excess of some inhibitory protein that particularly exacerbates the problems caused by the great overproduction of viral genomes relative to viral envelope proteins in the VC virus type. The role of host cell proteins in morbillivirus replication remains relatively unexplored. In vitro, tubulin has been shown to be required for MV transcription (47), and a number of host cell proteins have been shown to bind to the MV V and C proteins (45), including the binding of damage-specific DNA binding protein to the V protein (43). One of the heat shock proteins (hsp72) appears to be a necessary part of the replication complex in canine distemper virus (48), and yet other host cell proteins have been shown to bind specifically to the promoter regions of MV (42). Clearly further information as to the nature of these host cell proteins and their distribution and availability in different cell types will be required in order to interpret the cell-type-specific phenotypes of various virus mutations.