INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The etiological agent of mumps was first shown reproducibly to be a virus by Johnson and Goodpasture in 1935 (17). Since then, propagation in tissue culture has facilitated virus classification and studies on the biological properties of mumps virus (MUV). Originally classified with influenza viruses in the myxovirus family, mumps virus has since been reassigned to the Paramyxoviridae family, subfamily Paramyxovirinae, genus Rubulavirus, based on nucleocapsid morphology, genome organization, and biological properties of the proteins. Other members of the Rubulavirus genus include simian virus 5 (SV5), human parainfluenza virus type 2 and type 4, and Newcastle disease virus (18). Like all viruses of the Paramyxoviridae, MUV is pleomorphic in shape, comprising a ribonucleoprotein core surrounded by a host cell-derived lipid membrane; the nucleocapsid core forms a helical structure composed of the 15,384-nucleotide (nt) nonsegmented negative-sense RNA genome closely associated with virus nucleocapsid protein (NP). The genetic organization of the MUV genome has been determined to be 3'-NP-P-M (matrix)-F-SH (small hydrophobic)-HN-L-5' (10). Each gene encodes a single protein except for the P cistron, from which three unique mRNAs are transcribed. One is a faithful copy of the P gene, encoding the V protein. The two other mRNAs contain two and four nontemplated G residues inserted during transcription by a RNA editing mechanism; they encode the P and I proteins, respectively (26). The role of the V and I proteins in virus growth is not yet clear, but there is evidence that V is a structural protein and is associated with virus nucleocapsid (27). It is believed that the P and L proteins in association with nucleocapsid form the functional RNA polymerase complex of MUV. F and HN, integral membrane proteins which project from the surface of the virion, are involved in virus attachment and entry of cells. The SH and M proteins are also membrane associated (18, 38).

The MUV replicative cycle initiates with release of virus nucleocapsid into the host cell cytoplasm, brought about by fusion of virus envelope with host cell plasma membrane. Primary transcription ensues, resulting in the production of all virus proteins; a switch to replication of the virus genome occurs later followed by assembly of virus components to form new virus particles which bud from the host cell plasma membrane. Only the intact nucleocapsid structure can act as template for RNA transcription, replication, and subsequent virus amplification; therein lies the difficulty in genetic manipulation of MUV and other negative-strand RNA viruses. Unlike the positive-strand RNA viruses, where naked genomic RNA is infectious and infectious virus can be recovered from a cDNA copy of the genome in the absence of additional viral factors (31, 39), the naked genome of nonsegmented negative-strand RNA viruses is not infectious, and rescue of virus from cDNA requires at least intracellular coexpression of viral NP, P, and L proteins, along with a full-length positive-sense genome RNA transcript, all under control of a bacteriophage T7 RNA polymerase promoter (2, 3, 6, 9, 14, 15, 16, 19, 29, 32, 33, 40). In all of the rescue systems described so far for the nonsegmented RNA viruses, T7 RNA polymerase was supplied either by a coinfecting recombinant vaccinia virus (12, 41) or by endogenous expression in a transformed cell line (32). Recently influenza virus, whose genome comprises eight separate negative-sense RNA segments, was rescued from cDNA; where essential virus proteins and genomic RNAs were coexpressed intracellularly under control of either cellular RNA polymerase promoters (23) or a combination of the human RNA polymerase I promoter and the adenovirus type 2 major late promoter (11). The ability to rescue negative-strand RNA viruses from cDNA has provided an entirely new approach for the study of virus growth and may provide the basis for a range of novel virus expression vectors which may be used for the prevention and treatment of disease. Here we report the recovery of infectious MUV from a cloned cDNA of the virus genome.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and viruses. Primary chick embryo fibroblast (CEF) cells were obtained from SPAFAS Inc. (Preston, Conn.) and cultured in Eagle's basal medium supplemented with 5% fetal calf serum. HEp-2, 293, A549, and Vero cells were obtained from the American Type Culture Collection (Manassas, Va.) and grown in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum. The Jeryl Lynn strain of mumps virus was obtained directly from a vial of Mumpsvax (Merck and Co., Inc., Westpoint, Pa.). Recombinant vaccinia virus Ankara (MVA-T7), expressing bacteriophage T7 RNA polymerase, was obtained from B. Moss (National Institutes of Health, Bethesda, Md.).

Construction of expression plasmids for MUV NP, P, and L proteins. Expression plasmids for the MUV NP, P, and L proteins (pMUVNP, pMUVP, and pMUVL) were constructed by positioning cDNA for each open reading frame (ORF) between the T7 RNA polymerase promoter and the T7 RNA polymerase transcription termination sequence of a plasmid vector which contained the cap-independent translation enhancer of encephalomyocarditis virus. The primers used for reverse transcription (RT)-PCR amplification of the MUV NP protein ORF, from total MUV infected-cell (CEF) RNA, were 5'ATCATTCGTCTCCCATGTTGTCTGTGCTCAAAGC and 5'ATCATTCTCGAGTTGCGATTGGGGTTAGTTTG; the resulting cDNA fragment was gel purified, trimmed with BsmBI and XhoI (endonuclease sites are indicated in boldface), and then cloned into NcoI/XhoI-cut pEMC, such that the AUG of the NP protein ORF was adjacent to the cap-independent translation enhancer. Primers for amplification of the MUV P ORF were 5'TTCCGGGCAAGCCATGGATC and 5'ATCATTCTCGAGAGGGAATCATTGTGGCTCTC. The P ORF cDNA was also cloned into the NcoI/XhoI sites of pEMC and subsequently modified by site-directed mutagenesis to include the two G nucleotides not present in the virus genome (26). Because of its large size, the L protein ORF was assembled in two steps. Primers 5'ATCATTCGTCTCCCATGGCGGGCCTAAATGAGATACTC and 5'CTTCGTTCATCTGTTTTGGATCCG were used in the first step to produce a cDNA fragment which was trimmed with BsmBI and BamHI and then cloned into the NcoI/BamHI sites of pEMC. In the second step, primers 5'CAGAGTACCTTATATCGGATCC and 5'ATCATTCTGCAGGAATTTGGATGTTAGTTCGGCAC were used to amplify a cDNA fragment which was cloned into the BamHI/PstI sites of the plasmid from step 1 above, to complete the L protein ORF. Four cDNA clones for each of the three ORFs were sequenced, and the ORF with the highest level of homology to the Jeryl Lynn consensus nucleotide/amino acid sequence was chosen in each case for use in rescue experiments.

Construction of a synthetic MUV minireplicon. A synthetic MUV minireplicon (MUVCAT) was assembled from cDNA representing a modified MUV genome, where all coding and intercistronic regions were replaced by the bacterial chloramphenicol acetyltransferase (CAT) ORF; cDNA for the MUV 3' 145-nt and 5' 161-nt ends was amplified by RT-PCR from total infected-cell (CEF) RNA, using primer pairs 5'ACCAAGGGGAGAATGAATATGGG-5'ATCATTCGTCTCTTTTCCAGGTAGTGTCAAAATGCC and 5'ACCAAGGG GAGAAAGTAAAATC-5'ATCATTCGTCTCTATCGAATAAAGACTCTTC TGGC, respectively. In a second round of PCR amplification, nested primers were used for addition of the T7 RNA polymerase promoter and the NarI-containing portion of the hepatitis delta virus (HDV) ribozyme sequence to the MUV 5' and 3' ends, respectively; these primer pairs were 5'AAGCTCGGCGGCCG CTTGTAATACGACTCACTATAACCAAGGGGAGAAAGTAAAATC-5'ATCATTC GTCTCTATCGAATAAAGACTCTTCTGGC for addition of the T7 RNA polymerase promoter (which is shown in smaller print) and 5'ATCATTGGCGCCAG CGAGGAGGCTGGGACCATGCCGGCCACCAAGGGGAGAATGAATATGGG-5'AT CATTCGTCTCTTTTCCAGGTAGTGTCAAAATGCC for addition of the ribozyme component, which is shown in smaller print (30). The CAT ORF cDNA was amplified using primers 5'ATCATTCGTCTCGGAAAATGGAGAAAAAAATCACTGGATATACC and 5'ATCATTCGTCTCTCGATTTACGCCCCGCCCTGCCACTC. All three cDNA components were gel purified, trimmed with BsmBI, joined together in a four-way ligation, and cloned into the NotI/NarI sites of modified pBluescript KS(+) (35) to produce the complete minireplicon plasmid, pMUVCAT (Fig. 1).

View larger version (11K): [in this window] [in a new window]   FIG. 1.   Diagram (not to scale) showing the organization of the MUVCAT minireplicon DNA construct and T7 RNA polymerase-transcribed minireplicon antisense RNA genome. Key restriction endonuclease sites used in the assembly of the DNA construct are shown. The T7 RNA polymerase promoter sequence was designed to start transcription with the exact MUV 5'-terminal nucleotide, and an HDV ribozyme (Rib.) sequence was positioned to generate the precise MUV 3'-terminal nucleotide in minireplicon RNA transcripts. Duplicate T7 RNA polymerase termination signals were included after the HDV ribozyme sequence. The CAT ORF replaces all of the coding and intercistronic sequence of the MUV genome; the remaining essential MUV-specific sequence comprises the 3' MUV leader (55 nt) with adjacent 90-nt NP gene untranslated region (UTR) and the 5' MUV trailer (24 nt) adjacent to the 137-nt L gene untranslated region.

Construction of a full-length genome cDNA for MUV. To enrich for a clonal population of virus for the construction of a full-length cDNA clone of the Jeryl Lynn strain of MUV, a well-isolated virus plaque from the vaccine preparation was picked and used to prepare working stocks of virus. The full-length genome cDNA (pMUVFL) was assembled 5' end to 3' end (negative-sense genome) by the successive cloning of contiguous cDNA fragments into the same plasmid backbone that was used for the construction of pMUVCAT (Fig. 2). Each cDNA fragment was amplified from total infected-cell RNA by RT-PCR using primer pairs which contained unique restriction sites; in each case the positive sense primer contained a 5'-proximal NotI site in addition to the virus-specific endonuclease site, to facilitate the stepwise cloning strategy. Prior to addition to the growing full-length clone, the cDNA fragment spanning the virus 3' end to the BssHII site was assembled separately in pBluescript II SK(+) (Stratagene, La Jolla, Calif.); in the first step, the BssHII/ClaI cDNA fragment was cloned into the ClaI/XhoI sites of pBluescript II SK(+), using a 5'-extended primer to generate an XhoI site adjacent to the virus-specific BssHII site. In the second step, the virus 3' end-to-ClaI cDNA fragment was cloned into the NotI/ClaI sites of plasmid from the first step to complete the virus 3'-end-to-BssHII sequence. The T7 RNA polymerase promoter sequence was added to the virus 3' end by incorporation into the plus-sense RT-PCR primer used to generate the virus 3'-end-ClaI fragment. The 5'-terminal fragment (BamHI/NarI) of the genome cDNA was separately modified in a second round of PCR amplification to add the virus 5' end to NarI-containing portion of the HDV ribozyme sequence. A total of four cloning cycles were required for assembly of pMUVFL; after each round, four clones were sequenced in the region of newly added cDNA and compared to the MUV consensus sequence. The cDNA clone containing the least number of mutations was then selected for addition of the next cDNA fragment. The fully assembled cDNA clone was resequenced to verify stability during bacterial amplification. Electrocompetent SURE cells (Stratagene) and DH5-alpha cells (GibcoBRL, Rockville, Md.) were used as bacterial hosts for cloning of MUV cDNA.

View larger version (7K): [in this window] [in a new window]   FIG. 2.   Schematic representation (not to scale) of the MUV full-length genome cDNA construct showing the genetic organization of the MUV genome including the nontranscribed leader (Le) and trailer (Tr) and the proteins expressed from each gene. The subgenomic cDNA fragments and restriction endonuclease sites used in the assembly process are delineated by the horizontal solid lines and the vertical dotted lines, respectively. The T7 RNA polymerase promoter (T7-P) and the HDV ribozyme (Rib) sequence were positioned to initiate transcription with the exact 5'-terminal nucleotide and generate the precise 3'-terminal nucleotide of the MUV antisense genome, respectively. Tandem T7 RNA polymerase termination sequences were placed adjacent to the HDV ribozyme.

Rescue of CAT activity from transfected cells. For rescue of CAT activity, cells were either infected with MUV and transfected with in vitro-transcribed MUVCAT minireplicon RNA or infected with MVA-T7 and transfected with pMUVCAT along with expression plasmids pMUVNP, pMUVP, and pMUVL. In vitro transcriptions were carried out with 4 µg of pMUVCAT as template for T7 RNA polymerase in a 20-µl final volume as specified by the manufacturer (Promega, Madison, Wis.); template DNA was then destroyed by digestion with RQ-1 DNase for 15 min at 37°C. Overnight cultures of 293 cells grown to ~80% confluence in six-well dishes were infected with MUV at a multiplicity of infection (MOI) of 1 to 2; at 1 h postinfection (hpi), a mixture containing 5 to 10 µl of in vitro transcription reaction (approximately 5 to 10 µg of RNA) and 10 to 12 µl of LipofectACE (GibcoBRL) was added to each well according to the supplier's protocol. At 48 hpi, cells were scraped into suspension, collected by centrifugation, resuspended in 100 µl of 0.25 M Tris buffer (pH 7.8), and subjected to three rounds of freeze-thaw. The clarified cell extracts were then assayed for CAT activity using either C-14-labeled chloramphenicol or fluorescein-labeled chloramphenicol (Molecular Probes, Eugene, Oreg.), followed by analysis of reaction products by thin-layer chromatography.

Recovery of infectious MUV from transfected cells. For rescue of infectious MUV from cDNA, A549 cells grown overnight to ~90% confluence in six-well dishes were infected with MVA-T7 at an MOI of 4; at 1 hpi, cells were transfected with a mixture containing 3 to 7 µg of pMUVFL, 300 ng of pMUVNP, 50 ng of pMUVP, 200 ng of pMUVL, and 14 µl of LipofectACE. At 24 hpi, the transfection mixture was replaced with growth medium (Dulbecco's modified Eagle medium containing 10% fetal calf serum), and cells were incubated at 37°C for a further 48 h; either supernatants (P1) or total transfected cell monolayers scraped into suspension were then transferred directly onto confluent A549 cell monolayers, which were incubated at 37°C for 4 days and then prepared for whole-cell enzyme-linked immunosorbent assay (ELISA) (see below) in order to detect MUV infectious foci. Supernatants (P2) from the A549 indicator cells were further passaged onto confluent Vero cell monolayers and incubated at 37°C for 3 to 4 days to observe MUV-induced syncytia.

Identification and authentication of rMUV. Initial identification of rescued MUV (rMUV) was carried out using a whole-cell ELISA; A549 cells infected with P1 transfection supernatants (see above) were fixed with 10% formaldehyde in 1× phosphate-buffered saline (PBS) for 30 min at room temperature; cells were then rinsed once with PBS and once with blocking solution (5% [wt/vol] dried milk in 1× PBS), followed by incubation overnight at 4°C in blocking solution. The overnight blocking solution was then removed, and cells were incubated at room temperature for 2 to 3 h with MUV polyclonal rabbit antiserum (Access Biomedical, San Diego, Calif.) diluted 1/400 in fresh blocking solution. The polyclonal antiserum was then removed; cells were rinsed five times with blocking solution and then incubated at room temperature for 2 to 3 h with horseradish peroxidase-conjugated goat anti-rabbit serum (DAKO Corporation, Carpinteria, Calif.) diluted 1/1,000 in blocking solution. The goat serum was then removed; cells were washed five times with blocking solution and once with PBS, followed by addition of enough 3-amino-9-ethylcarbazole substrate (DAKO) to cover cell monolayers, which were then incubated at 37°C for 15 to 20 min to facilitate stain development.

Nucleotide sequence accession number. The consensus nucleotide sequence for the complete genome of the Jeryl Lynn strain of MUV and the amino acid sequences of the virus proteins encoded therein have been submitted to Genbank (accession no. AF201473).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Rescue of reporter gene activity from transfected cells. To help define conditions which would permit the rescue of infectious MUV from cDNA, an MUV minireplicon containing the CAT reporter gene was assembled. The construct was designed to allow synthesis of a RNA minigenome of negative polarity under control of the T7 RNA polymerase promoter. The three terminal G residues of the T7 promoter were omitted during construction of the minireplicon to provide a transcriptional start site which began with the precise 5' nucleotide of the MUV genome. Inclusion of the HDV ribozyme in the minireplicon construct permitted cleavage of the T7 RNA polymerase transcript to produce the authentic MUV specific 3' end (30). The total number of nucleotides (966) in the MUVCAT minireplicon RNA was divisible by 6, in agreement with the "rule of six" proposed for Sendai virus, where it has been suggested that each NP molecule interacts with six nucleotides along the RNA genome, and so genome lengths containing multiples of six nucleotides would be more efficient in replication and rescue (4, 22). Expression of the CAT gene was under control of an MUV-specific promoter and could occur only if minireplicon RNA became encapsidated with NP and that ribonucleocapsid template then interacted with functional MUV-specific RNA polymerase protein(s) to transcribe CAT mRNA.

View larger version (58K): [in this window] [in a new window]   FIG. 3.   (A) Thin-layer chromatograms showing CAT activity in 293 cells following infection with MUV and transfection with RNA transcribed in vitro from pMUVCAT. Panels A1, A2, and A3 show the results from three separate rescue experiments. (A1) Lane 1, CAT activity in MUV-infected cells transfected without in vitro-transcribed pMUVCAT RNA; lane 2, CAT activity in extracts of MUV-infected cells transfected with RNA transcribed in vitro from pMUVCAT; lane 3, CAT activity in MUV-infected cells transfected with RNA transcribed in vitro from pMUVCAT-GG; lane 4, CAT activity in uninfected cells transfected with RNA transcribed in vitro from pMUVCAT. Each CAT assay was carried out at 37°C for 3 to 4 h with 20% of the extract from ~106 transfected cells. (A2) Lane 1, MUV-infected cells transfected with RNA transcribed in vitro from pMUVCAT; lane 2, uninfected cells transfected with RNA transcribed in vitro from pMUVCAT. Each CAT assay was carried out at 37°C for 5 h using 50% of the extract from ~106 transfected cells. (A3) Lane 1, MUV-infected cells transfected with RNA transcribed in vitro from pMUVCAT; lane 2, MUV-infected cells transfected without in vitro-transcribed pMUVCAT RNA; lane 3, uninfected cells transfected with in vitro-transcribed RNA from pMUVCAT. Each CAT assay shown in panel A3 was carried out at 37°C for 4 h using 50% of the extract from ~106 transfected cells. (B) Thin-layer chromatograms showing CAT activity in extracts of MVA-T7-infected HEp-2 and A549 cells following transfection with pMUVCAT and plasmids expressing MUV NP, P, and L proteins. The level of pMUVNP expression plasmid was titrated in both cell lines. Lanes 1 to 4, CAT activity following transfection with mixtures containing 200 ng of pMUVCAT, 50 ng of pMUVP, and 200 ng of pMUVL each, and 300, 450, 600, and 750 ng of pMUVNP, respectively; lane 5, CAT activity produced when pMUVL was omitted from the transfection mixture. Each CAT assay was performed at 37°C for 3 h using 20% of the cell extract from each well of transfected cells (~106 cells/well of a six-well dish).

Recovery of MUV from transfected cells. In line with the strategy first used by Schnell et al. (33) for the rescue of rabies virus, the full-length MUV cDNA was assembled to permit the synthesis of a precise 15,384-nt positive-sense RNA copy of the virus genome under control of the T7 RNA polymerase promoter, precluding intracellular annealing with mRNA transcribed from NP, P, and L expression plasmids during rescue experiments. As for the MUVCAT minireplicon, the T7 RNA polymerase promoter sequence was modified to omit the three terminal G residues, providing a transcriptional start site beginning at the exact MUV 5'-terminal nucleotide. The HDV ribozyme was used to generate the exact MUV 3'-terminal nucleotide of the positive-sense genome transcripts.

View larger version (199K): [in this window] [in a new window]   FIG. 4.   (A) Photographs showing rMUV-induced syncytia on Vero cell monolayers. Supernatants from MVA-T7-infected A549 cells transfected with pMUVFL, pMUVNP, pMUVP, and pMUVL were passaged onto A549 cells, incubated at 37°C for 4 days, then transferred onto Vero cell monolayers, and incubated for 3 more days at 37°C prior to photography (A1). (A2) Representative portion of Vero cell monolayer following transfer of supernatant from transfected A549 cells as described for panel A1 except that pMUVL was omitted from the transfection mixture; (A3) syncytia produced on Vero cells following infection with Jeryl Lynn vaccine virus. (B) Photographs showing rMUV-induced plaques on Vero cell monolayers stained by whole-cell ELISA. Supernatant from transfected cells was passed onto A549 indicator cells and incubated for 3 days at 37°C; supernatant from these cells was then passed onto Vero cell monolayers. One of the resulting syncytia was picked and used to infect fresh Vero cell monolayers. Virus-induced plaques were then stained by whole-cell ELISA 4 days postinfection (B1) and compared to plaques induced by Jeryl Lynn vaccine virus (B3). Panel B2 shows Vero cells infected with cell supernatants as for cells in panel B1 except that the L expression plasmid was omitted from the starting transfection mixture.

Identification of rMUV. It was important to demonstrate that rMUV was derived from cDNA (pMUVFL). This was made possible by the presence of three nucleotide tags in pMUVFL, introduced by RT-PCR misincorporation during assembly of the full-length genome cDNA. These tags differentiated pMUVFL from the consensus sequence of the Jeryl Lynn vaccine virus and from that of a passaged plaque isolate of the Jeryl Lynn vaccine preparation from which pMUVFL was derived. Two of the tags represented silent changes at nt 6081 (T to C) and 11731 (A to G) in the F and L genes, respectively; a third tag (nt 8502, A to G) resulted in a Lys-to-Arg substitution at amino acid 22 of the L protein of pMUVFL. To show that rMUV was generated from pMUVFL and not from either of the other two MUV populations grown in the laboratory, RT-PCR products prepared from rMUV-infected-cell RNA, spanning each of the three nucleotide tags, were sequenced at the relevant position(s). To demonstrate that carryover of transfecting plasmid DNA was not contributing significantly to these RT-PCR products, one reaction was carried out with rMUV-infected-cell RNA as the template for PCR amplification without prior RT. Results from the RT-PCR amplifications (Fig. 5), and subsequent sequence analysis of RT-PCR products (Fig. 6), clearly demonstrated that the rescued virus was derived from pMUVFL. Interestingly, the Lys-to-Arg mutation at position 22 of the L protein of rMUV did not prevent rescue from occurring even though an alignment of L proteins indicates that Lys is highly conserved at the same relative position among members of the Paramyxovirinae.

View larger version (73K): [in this window] [in a new window]   FIG. 5.   Gel analysis of RT-PCR products used to identify rMUV. Total RNA was prepared from Vero cell monolayers infected with P2 rMUV virus from transfected cells. RT-PCRs were set up to generate cDNA products spanning the three separate nucleotide tag sites present only in pMUVFL and rMUV. Lane 1, marker 1-kb ladder (Gibco/BRL); lanes 2 to 4, RT-PCR products spanning nucleotide tag positions 6081, 8502, and 11731, respectively. To demonstrate the absence of contaminating plasmid DNA, a reaction identical to that used for generation of the cDNA shown in lane 4 was performed without RT; the product(s) of this reaction is shown in lane 5. To demonstrate that no rMUV could be recovered when pMUVL was omitted from transfection mixtures, RT-PCR identical to that used to generate the cDNA products shown in lane 4 was set up using Vero cell RNA derived from transfections carried out without pMUVL; products from this reaction are shown in lane 6.

View larger version (28K): [in this window] [in a new window]   FIG. 6.   Electropherograms showing nucleotide sequence across identifying tag sites in rMUV. RT-PCR products (Fig. 5) were sequenced across each of the three tag sites. The nucleotide sequence at each tag site obtained for rMUV is compared with consensus sequence for the plaque isolate of MUV used to derive pMUVFL.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Infectious MUV has been generated from a DNA copy of the virus genome following cotransfection of MVA-T7-infected A549 cells with plasmids encoding MUV NP, P, and L proteins, along with a plasmid containing the complete genome cDNA of MUV. The success of this process necessarily was preceded by the development of a consensus sequence for the entire MUV genome (Jeryl Lynn strain) and the development of an MUV minireplicon rescue system, which had not been previously reported.

Prior work had shown that the Jeryl Lynn vaccine strain contained a mixture of two distinct virus populations (1). Therefore, to minimize the potential for suboptimal protein-protein interactions (by splicing together cDNA fragments derived from the different virus populations into the genome cDNA) during the rescue process, a well-isolated plaque from the Jeryl Lynn vaccine preparation was selected and amplified for the derivation of the full-length genome cDNA and the NP, P, and L expression plasmids.

To find conditions suitable for the rescue of infectious virus from a genome cDNA, a synthetic MUV minireplicon similar to those described for influenza virus and members of the Paramyxoviridae and Rhabdoviridae was assembled so as to contain the CAT reporter gene (5, 7, 8, 20, 24, 28, 35). Initially CAT activity was rescued by transfection of MUV-infected 293 cells with RNA transcribed from this construct in vitro. These results demonstrated that the minireplicon genome contained all MUV-specific cis-acting sequences (nt 1 to 145 for the 3' leader region; nt 15223 to 15384 for the 5' trailer region) necessary for encapsidation, transcription, and presumably replication of the minireplicon, and they define a basic rescue system by which these important regulatory sequences can be further dissected. Interestingly, CAT rescue was not detected when MUV-infected 293 cells were transfected with RNA transcribed from a minireplicon construct which contained two additional G residues at the 3'-proximal end of the T7 RNA polymerase promoter; it is possible that the additional one or two G residues present at the 5' end of the resulting negative-sense minireplicon RNA transcript violated the rule of six proposed for members of the Paramyxovirinae (4).

MUV minireplicon CAT rescue was then achieved in MVA-T7-infected 293, HEp-2, and A549 cells transfected with pMUVCAT and plasmids expressing MUV NP, P, and L proteins in the absence of MUV helper virus. This demonstrated that the expressed NP, P, and L proteins were sufficient to substitute for helper virus in CAT rescue experiments, thus indicating a role for these proteins in genome encapsidation and RNA polymerase activity. The ratio of expression plasmids was optimized in the minireplicon rescue system for later use in the effort to rescue full-length infectious MUV. It was notable that the P expression plasmid was required at approximately one-sixth the molar ratio of the NP plasmid. This observation was in line with optimum plasmid ratios required for the rescue of another rubulavirus, SV5 (14), and mirrors the anticipated lower level of the P protein mRNA transcribed in infected cells, which is a result of the RNA editing mechanism that occurs during transcription of the P gene (26). Initially plasmid-based rescue of CAT activity was carried out in 293 cells; however, these cells were very susceptible to MVA-T7-induced cytopathology and did not support vigorous MUV growth. Although HEp-2 cells showed increased efficiency and reproducibility of CAT rescue relative to 293 cells, further improvements in CAT rescue efficiency were made in A549 cells which had been used successfully for the rescue of SV5 (14). This cell line appeared to be more resistant to MVA-T7-induced cytopathology than HEp-2 cells, was readily transfectable, and did indeed support MUV growth. For these reasons, A549 cells were selected for attempts to rescue infectious MUV from cDNA.

Rescue of other nonsegmented negative-strand RNA viruses succeeded only when Conzelmann and Schnell (7) and Lawson et al. (19) recognized that intermolecular annealing of N, P, and L mRNA transcripts with full-length negative-sense genome RNA transcripts in transfected cells might greatly constrain nucleocapsid formation, which is a prerequisite for rescue of infectious virus. Virus rescue was achieved when they altered the design of plasmids containing full-length virus genome cDNA such that a RNA transcript of antigenome polarity would be produced in transfected cells. Similarly, to prevent intermolecular annealing of RNA transcripts from the MUV NP, P, and L expression plasmids with the full-length MUV genome RNA transcript, the T7 RNA polymerase promoter was positioned to generate a positive-sense RNA copy of the virus genome. Like the MUVCAT minireplicon, the full-length construct was modified to allow synthesis of a genome transcript beginning with the exact MUV 5'-terminal nucleotide and ending with the exact MUV 3'-terminal nucleotide. The conditions used for rescue of infectious MUV from A549 cells were similar to those optimized for the MUVCAT minireplicon; however, it was not clear how much full-length genome cDNA should be used in rescue experiments, since the large size of the genome cDNA plasmid could affect efficiency of uptake during transfection, preventing direct correlation with levels of the much smaller MUVCAT plasmid DNA used in CAT rescue experiments. A titration of transfection components showed that rescue of infectious MUV was very reproducible when 3 to 7 µg of the genome cDNA plasmid was present with the NP, P, and L expression plasmids in transfection mixtures. The number of infectious particles released during rescue experiments was not precisely measured; however, the number of infectious foci observed by whole-cell ELISA on A549 indicator cells ranged from 3 to 20 for each well of transfected cells. It is possible that further improvements in the efficiency of infectious MUV rescue can be made by adjusting the MOI of MVA-T7 on A549 cells and by using the process of heat shock described recently for the rescue of measles virus from cDNA (25). One possible impediment to rescue, a Lys-to-Arg substitution at amino acid 22 of the L protein in the full-length construct, was not realized even though Lys is highly conserved at the same relative position in other members of the Paramyxovirinae. It would be interesting to see if that position reverts to Lys on continued passage and what effect the mutation has on the relative fitness of rescued virus.

Now that MUV can be rescued from cDNA, it may be possible to engineer the virus genome of express foreign genes, as described for a number of other negative-sense RNA viruses (13, 14, 16, 21, 34, 36). The ability to do this may provide an ideal means to deliver prophylactic and therapeutic agents for the prevention and treatment of diseases other than mumps. Some properties of the Jeryl Lynn vaccine strain which should facilitate its acceptance as an expression vector in humans include a highly favorable attenuation phenotype, an apparent absence of recombination, an impressive safety record for the >100 million doses administered, and the ability to induce long-lasting immunity with a single inoculation.