INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Human parainfluenza virus 3 (PIV3) is the second leading cause of hospitalization of infants and young children for viral respiratory tract disease worldwide (3). Previous efforts at paramyxovirus vaccine development have suggested that live-attenuated, intranasally administered vaccine viruses represent the best strategy for the prevention of the severe lower respiratory tract disease that occurs in infants and children (8). A licensed PIV3 vaccine is not yet available, but attenuated candidate vaccine viruses, including a bovine PIV3 and a human cold-passaged (cp) PIV3, termed cp45, have been developed and are under clinical evaluation (18, 19). The PIV3 cp45 candidate vaccine was produced by passaging its parent, the JS wild-type (wt) strain of PIV3, 45 times in cell culture at progressively lower temperatures to a final temperature of 20°C (2). The cp45 virus was isolated as a biological clone that had the cold-adapted (ca), temperature-sensitive (ts), and attenuation (att) phenotypes. cp45 is attenuated for growth in the respiratory tract of hamsters, rhesus monkeys, chimpanzees, and humans and maintains its ts, ca, and att phenotypes following replication in vivo (6, 14, 15, 18, 19).

PIV3 is a single-stranded, negative-sense, enveloped RNA virus of 15,462 nucleotides (nt) (3). PIV3 encodes three nucleocapsid-associated proteins, the nucleocapsid protein (N), the phosphoprotein (P), and the major polymerase subunit (L). The N protein binds tightly to genomic RNA to form the nucleocapsid template, the P protein is a polymerase cofactor which also acts to bring soluble N and L proteins to the nucleocapsid, and the L protein contains conserved polymerase motifs that probably represent functional domains, including those that may be required for association with the P protein, RNA binding, RNA polyadenylation, RNA transcription, and RNA replication (3). PIV3 also encodes three envelope-associated proteins, the internal matrix protein (M), the fusion glycoprotein (F), and the hemagglutinin-neuraminidase glycoprotein (HN). The M protein is thought to mediate virion assembly, the HN protein mediates viral attachment as well as viral release through the action of its neuraminidase, and the F protein mediates viral penetration. The P mRNA also encodes the nonstructural C protein from an alternative translational open reading frame (ORF). As with many paramyxoviruses, RNA editing during the synthesis of the P mRNA results in the insertion of one or more G residues midway down the P-encoding ORF. In the case of PIV3, the insertion of two G residues shifts the reading frame to access an internal ORF and generate the chimeric D protein (11). The P mRNA also contains an internal ORF which has the potential to encode a cysteine-rich domain called V, but the presence of numerous translation stop codons between the editing site and the V-specific ORF seems to preclude its expression. The 3' and 5' ends of the viral genome contain extragenic leader and trailer regions, possessing promoters required for replication and transcription, and the PIV3 genetic map is 3' leader-N-(P/C/D)-M-F-HN-L-5' trailer. Transcription initiates at the 3' end and proceeds by a sequential stop-start mechanism that is guided by short conserved motifs found at the gene boundaries. Specifically, the upstream end of each gene contains a gene start (GS) signal, which directs initiation of its respective mRNA. The downstream terminus of each gene contains a gene end (GE) motif which directs polyadenylation and termination, and each gene is separated by a conserved intergenic trinucleotide which also is thought to play a role in transcription.

cp45 differs from its JS wt parent by 20 point mutations (31, 33a). Five of these changes are not considered to be important for the phenotypes of cp45, because they occur within ORFs but do not affect amino acid coding. The remaining 15 changes include four point mutations in the leader region, one point mutation in the N gene GS signal, and 10 amino acid substitutions distributed among six proteins. The development of a cDNA-based system for producing recombinant PIV3 (rPIV3) has made it possible to evaluate the genetic basis of the ts, att, and ca phenotypes by introducing mutations present in cp45 into an rPIV3 and determining the effect of the introduced mutation(s) on these phenotypes (9, 10, 29). Previously, we demonstrated that the three amino acid substitutions in the L protein of cp45 confer much of the ts and att phenotypes, but not the ca phenotype (29). In this study, we have produced a recombinant version of the cp45 virus (rcp45) and have examined the contributions of the mutations in the 3' leader, in the N GS signal, and in the N, C, M, F, and HN proteins to the ts, ca, and att phenotypes of this virus. We show here that this promising vaccine candidate contains multiple ts and non-ts attenuating mutations, which likely contribute to the high level of attenuation and phenotypic stability of cp45 following replication in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Viruses and cells. The rPIV3, PIV3 JS wt, and cp45 viruses were grown in simian LLC-MK2 cells (ATCC CCL 7.1) as described previously (9, 15, 29). The modified vaccinia virus Ankara (MVA-T7) (37), which expresses the T7 polymerase, was kindly provided by Linda Wyatt and Bernard Moss. HEp-2 (ATCC CCL 23) and LLC-MK2 cells were maintained in OptiMEM I (Life Technologies, Gaithersburg, Md.) supplemented with 2% fetal bovine serum (FBS) and gentamicin sulfate (50 µg/ml), or in Earle's MEM (EMEM) (Life Technologies) supplemented with 10% FBS, gentamicin sulfate (50 µg/ml), and 2 mM glutamine. A cold-tolerant cell line, L-132-cp2-7, was derived from L-132 cells (ATCC CCL 5) by selection of a cell population that survived long-term incubation at 20°C and also remained adherent to the growth surface (24). L132-cp2-7 cells were grown in EMEM supplemented with 10% FBS, 2 mM glutamine, 20 mM HEPES, 1 mM nonessential amino acids, and 100 U of streptomycin-neomycin/ml.

Construction of point mutations in JS rPIV3. Four subgenomic fragments of p3/7(131)2G+, the antigenomic cDNA clone of PIV3 JS wt previously used to recover infectious virus (9, 29)encompassing PIV3 nt 1 to 3903 (MluI-BamHI), nt 3903 to 5261 (BamHI-BstEII), nt 5261 to 7437 (BstEII-XhoI), and nt 7437 to 8195 (XhoI-NcoI)were cloned into pUC19 vectors modified to accept these fragments by standard molecular cloning techniques. Point mutations corresponding to mutations identified in cp45, as well as mutations creating or ablating silent restriction enzyme recognition sequences (see Table 1) were introduced with the Transformer mutagenesis kit (Clontech) as described previously (29). After mutagenesis, restriction endonuclease fragments were sequenced completely and were cloned into the pLeft2G+ or pRight+ plasmids and then into the full-length clone, p3/7(131)2G+, as XhoI-to-NgoMI fragments. The 3' leader and N mutations were amplified by reverse transcription (RT)-PCR directly from PIV3 cp45 virion RNA and were cloned into pLeft2G+ (9). Combinations of mutations were constructed by standard molecular cloning techniques. The full-length plasmid clone containing all 15 cp45 mutations, designated pFLCcp45, was completely sequenced, and it was confirmed that extraneous mutations had not been introduced during the cloning process.

Recovery of rPIV3. Each full-length antigenomic cDNA bearing cp45 mutations, together with the three support plasmids (9) pTM(N), pTM(P no C), and pTM(L), were transfected into HEp-2 cells on six-well plates (Costar, Cambridge, Mass.) with LipofectACE (Life Technologies) and MVA-T7 as described previously (9, 29). After incubation at 32°C for 4 days, the transfection harvest was passaged onto LLC-MK2 cells in T-25 flasks which were incubated at 32°C for 4 to 8 days. The clarified medium supernatant was called passage 1 and was subjected to three rounds of plaque purification on LLC-MK2 cells as described previously (9, 15, 29). Each biologically cloned recombinant virus was amplified twice in LLC-MK2 cells at 32°C to produce virus for further characterization. Virus was concentrated from clarified medium by polyethylene glycol precipitation (22), and viral RNA (vRNA) was extracted with Trizol reagent (Life Technologies). RT was performed on vRNA by using the Superscript II preamplification system (Life Technologies) with random hexamer primers. The Advantage cDNA PCR kit (Clontech) and sense and antisense primers specific for various portions of the PIV3 genome were used to amplify fragments for restriction endonuclease analysis. The PCR fragments were analyzed by digestion with each of the restriction enzymes whose recognition sites had been created or ablated during construction of the mutations (Table 1, data not shown).

Efficiency of plaque formation of rPIV3 bearing cp45 mutations at permissive and restrictive temperatures. The level of temperature sensitivity of plaque formation in vitro of control and recombinant viruses was determined at 32, 35, 36, 37, 38, 39, 40, and 41°C in LLC-MK2 monolayer cultures for 6 days as previously described (15). Plaques were enumerated by hemadsorption with guinea pig erythrocytes following removal of the methylcellulose overlay, or alternatively, the viral plaques present in the monolayer were identified by immunoperoxidase staining with a mixture of two PIV3-specific anti-HN murine monoclonal antibodies (MAbs) 101/1 and 454/11 diluted 1:250 (23, 34).

Evaluation of rPIV3 mutant viruses for ca phenotype. Growth of mutant and wt rPIV3 viruses was determined at 32 and 20°C on confluent L-132-cp2-7 cell monolayers prepared in 24-well tissue culture plates. Duplicate wells of each of two plates were inoculated with 0.2 ml of each mutant or wt rPIV3 virus at a multiplicity of infection of 0.01. After 1 h of adsorption at room temperature, the inoculum was aspirated and the monolayers were washed with 1 ml of phosphate-buffered saline (PBS) per well. The inoculated cultures were overlaid with 0.5 ml of EMEM supplemented with 10% FBS, 2 mM glutamine, 20 mM HEPES, 1 mM nonessential amino acids, and 100 U of streptomycin-neomycin/ml. One plate was sealed in a waterproof pouch (Kapak) and then submerged in a 20°C bath for 13 days. The duplicate plate was placed at 32°C in a CO₂ incubator for 3 days. At the end of the incubation period, virus was harvested by freeze-thawing. The titer of virus recovered from each well was determined by plaque assay in LLC-MK2 cells at 32°C using hemadsorption with guinea pig erythrocytes to visualize plaques. JS wt PIV3 and cp45 reference viruses were included as controls.

Evaluation of rPIV3 mutant viruses for att phenotype. Five-week-old Golden Syrian hamsters seronegative for PIV3 were inoculated intranasally with 0.1 ml of OptiMEM I containing 10^6.0 PFU of JS wt rPIV3, PIV3 cp45 virus, or one of the mutant rPIV3s. On day 4 postinfection, the hamsters were sacrificed, the lungs and nasal turbinates were harvested, and the virus was quantified as previously described (9, 29). The mean log₁₀ 50% tissue culture infective dose (TCID₅₀)/gram at 32°C was calculated for each group of hamsters.

Immunogenicity and efficacy of rcp45. Three groups of five hamsters were inoculated intranasally with 0.1 ml of (i) L15 medium (placebo), (ii) L15 medium containing 10⁶ TCID₅₀ rcp45, or (iii) L15 medium containing 10⁶ TCID₅₀ of biologically derived cp45. Fifty-seven days after infection, the hamsters were bled and serum titers of PIV3 antibody were determined and compared to those from preinfection bleeds, as described previously (34). On day 58 the hamsters were challenged by intranasal administration of 10⁶ TCID₅₀ JS wt rPIV3. Nasal turbinates and lungs were harvested 4 days later, and the titer of JS wt PIV3 was determined as described above.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Introduction of PIV3 cp45 mutations into JS wt rPIV3. The 15 mutations in the 3' leader, the N GS signal, and the N, C, M, F, HN, and L genes of cp45 (Table 1) were introduced into the complete PIV3 antigenomic cDNA by site-directed mutagenesis or by direct PCR amplification of a segment of cp45 cDNA bearing the desired mutations, and the following recombinant viruses were recovered from antigenomic cDNA: (i) rcp45 3'N, containing the four point mutations of the leader region, the point mutation in the N GS signal, and the two amino acid changes in the N protein; (ii) rcp45 C, containing the single amino acid change in C; (iii) rcp45 M, containing the single amino acid change in M; (iv) rcp45 F, containing the two amino acid changes in F; (v) rcp45 HN, containing the single amino acid change in HN; (vi) rcp45 L, containing the three amino acid changes in L, as described previously (29); (vii) rcp45 3'NL, containing the mutations i and vi described above; (viii) rcp45 3'NCMFHN, containing all of the mutations except for the three in L; and (ix) rcp45, containing all 15 cp45 mutations (Table 1; Fig. 1). In most cases, each cp45 change was engineered to be accompanied by one or more nearby silent changes which introduced or removed a restriction enzyme recognition site (Table 1). These served as markers to confirm the presence of the mutation in the cDNA and in the recovered virus. Also, two of the amino acid coding changes (mutations 10 and 15 in Table 1) were made by two nucleotide changes rather than the single change found in cp45, which should reduce the possibility of reversion to wt. The cp45 cDNA, which contains all 15 of the cp45 changes in Table 1, was assembled from the same mutagenized cDNA subclones that were used to introduce cp45 changes into the other antigenomic cDNAs; it was sequenced in its entirety and was found to possess the desired sequences and to lack other unwanted mutations. Thus, all of the mutagenized subclones also lacked unwanted adventitious mutations.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Schematic representation of cp45 mutations that were introduced into wt JS rPIV3 and summary of the phenotypes specified for each mutant virus. Each of the rPIV3s bearing cp45 mutations is displayed as a negative-sense RNA, 3' to 5'. The relative position of each cp45 point mutation (*) indicated in Table 1 is shown. The ts, ca, large plaque (lp), and att phenotypes are described in Results. A plus sign denotes a virus that possesses the indicated phenotype; a minus sign indicates that a virus has the wt phenotype.

Plaque morphology. Previously the plaque phenotype of cp45 on human L-132 cell monolayers was described as tiny plaque; that is, the diameter of cp45 plaques was found to be less than one-half that of wt PIV3 plaques at 32°C (2, 6). We examined the plaque phenotype of the rPIV3s on simian LLC-MK2 cells. Several of the recombinant viruses had distinctive plaque morphology when grown on these cells at 32°C for 6 days. JS wt rPIV3 plaques ranged in diameter from 1 to 2 mm and were indistinguishable in size from the biologically derived JS wt PIV3 (data not shown). Many of the recombinant viruses bearing individual or sets of cp45 mutations had a plaque size that was smaller than that of wt PIV3 at 32°C (Fig. 2). In contrast to the small plaque morphology described for cp45 grown on L-132 monolayers, plaques of the cp45 and rcp45 viruses on LLC-MK2 cells were at least two- to threefold larger than wt rPIV3, ranging in diameter from 3 to 6 mm, and were indistinguishable from each other. This demonstrated the comparability of the biologically derived and recombinant cp45 viruses for this phenotype, indicating that the distinctive large-plaque phenotype of cp45 observed on LLC-MK2 monolayers is a composite phenotype requiring multiple genetic elements. The basis of the differences between the plaque morphology described here for LLC-MK2 cells and that reported previously for L-132 cells (2, 6) is not understood but presumably involves host factors.

View larger version (62K): [in this window] [in a new window]   FIG. 2.   Plaque size of rPIV3s in LLC-MK2 cell monolayer culture. LLC-MK2 monolayers in 24-well plates were infected with JS wt rPIV3 and JS rPIV3s bearing cp45 mutations as indicated and were incubated for 6 days at 32°C. Plaques were visualized by immunoperoxidase staining using anti-PIV3 HN antibodies.

Efficiency of plaque formation of rPIV3s bearing the cp45 mutations in LLC-MK2 cells at permissive and restrictive temperatures. The biologically derived cp45 virus is ts with a shutoff temperature of 38°C (2). In this study the rPIV3s bearing the cp45 mutations were assayed for their ability to form plaques at permissive and restrictive temperatures ranging from 32 to 41°C (Table 2). A virus was defined as bearing the ts phenotype if its reduction in replication at 40°C, i.e., the titer at 32°C minus the titer at 40°C, was 100-fold greater than that of wt rPIV3. According to this definition, the rcp45 viruses bearing mutations in either the C, M, F, or HN proteins were not ts, and mutations in at least two regions of cp45 (3' N and L) were found to specify the ts phenotype. As shown in Table 2, rcp45, containing all of the cp45 mutations, had a shutoff temperature of 38°C, which was identical to that of the biologically derived cp45. This demonstrated that the ts phenotype of cp45 had been successfully reproduced in rcp45. This finding also supported the authenticity of the sequence analysis of cp45 and the subsequent reconstruction of the mutations into recombinant virus. Two subsets of cp45 mutations were found to specify a level of temperature sensitivity that was greater than that observed for rcp45, which contains the full set of mutations. The rcp45 3'NCMFHN virus, which is identical to rcp45 except that it lacks the three L mutations, and the rcp45 3'NL virus each had a shutoff temperature of 36°C. Since the L mutations are known to confer temperature sensitivity individually and in combination, it is paradoxical that rcp45 3'NCMFHN was more, rather than less, ts than rcp45. This finding implies that there is a complex interaction between the mutations within cp45 whereby mutations compensate for each other to give a level of temperature sensitivity which is less than the sum of the individual components.

ca phenotype of rPIV3s bearing cp45 mutations. The biologically derived cp45 has the ca phenotype, which is characterized by the ability to grow to a high titer at 20°C (2), whereas JS wt PIV3 grows poorly at that temperature (Table 3). The rPIV3s were analyzed to determine which genetic elements of cp45 specified the ca phenotype (Table 3) and to determine if the mutation(s) that specifies the ca phenotype also specifies the ts or att phenotypes of cp45. The biologically derived cp45 and rcp45 viruses exhibited comparable levels of ca, indicating that this phenotype, like the plaque size and ts phenotypes, was successfully reproduced in the recombinant version of cp45. It was previously observed that rcp45 L was ts and att but not ca (29). This indicated that the genetic elements specifying the greater part of the ca phenotype were located outside L, and this was partially confirmed in the present study. Three rPIV3s possessing the 3' leader and N mutations (rcp45 3'N; rcp45 3'NCMFHN, and rcp45 3'NL) were ca. However, each of these three viruses replicated approximately 100-fold less well at 20°C than either rcp45 or cp45, indicating that other regions of cp45 contribute to the ca phenotype, even though this was not apparent from analysis of the other regions individually. Since mutations in L plus those in the CMFHN region are needed in addition to the 3' N region, it is clear that the ca phenotype is a composite phenotype, requiring many of the cp45 mutations to achieve the full replicative ability at 20°C. Therefore, the ca phenotype resembles the ts and large plaque phenotypes, in that mutations that contribute to the overall phenotype interact in a complex way with each other to specify the level of cold adaptation, plaque morphology, and temperature sensitivity of cp45.

Growth of rcp45 mutant viruses in hamsters. The cp45 mutant is reduced in efficiency of replication in the upper and lower respiratory tract of hamsters. Previously we had shown that the mutations in the L gene of cp45 specify most of the attenuation phenotype of this virus (29). Replication of the rcp45 virus was reduced more than 60-fold in the nasal turbinates and more than 3,000-fold in the lungs and, thus, was as attenuated as the biologically derived cp45 virus (Table 4). This indicates that the attenuation phenotype of cp45 had been successfully reproduced in its recombinant version.

Immunogenicity and efficacy of rcp45. Intranasal administration of cp45 to hamsters was shown previously to protect against subsequent challenge with wt PIV3 (7). In the present study, hamsters immunized with biologically derived cp45 developed a geometric mean hemagglutinin inhibition antibody titer of 1:256, and those with rcp45 had a mean titer of 1:141, indicating that the levels of immunogenicity of the two viruses were similar (data not shown). Immunization with each virus induced a high level of resistance to wt virus replication in the upper and lower respiratory tract, as indicated by a 1,000- to 10,000-fold reduction in replication of the wt rPIV3 challenge virus at each site (data not shown).

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

The cp45 live-attenuated vaccine candidate was previously shown to be highly attenuated and phenotypically stable after replication in experimental animals and humans (6, 14, 15, 18, 19). Recent advances in the ability to generate infectious virus from paramyxovirus cDNAs have allowed us to begin to examine the genetic basis of the ts, ca, and att phenotypes of cp45 (1, 4, 9, 12, 16, 17, 20, 21, 25, 26, 35). By reverse genetics, 15 cp45 mutations were introduced into a JS strain PIV3 cDNA plasmid and rcp45 virus was recovered. The ts, ca, and att phenotypes, plaque morphology, immunogenicity, and efficacy of rcp45 were indistinguishable from those of the biologically derived cp45 virus, demonstrating that cp45 had been faithfully reproduced from cDNA. Our findings confirm that the 15 selected mutations indeed accounted for the full set of properties of cp45.

Having established that the set of 15 mutations in rcp45 was able to fully reproduce each property of cp45, we next sought to determine the relative contribution of individual mutations or sets of mutations to the ts, ca, and att phenotypes. Data presented here demonstrate that the ts phenotype is a composite phenotype with multiple genetic elements making a contribution. Each of the three mutations in L and one or more of the mutations in the 3' N region independently specify the ts phenotype, indicating that at least four individual mutations or groups of mutations make a separate contribution to the overall temperature sensitivity of cp45. This multicomponent contribution to the ts phenotype is a partial explanation for the high level of stability of the ts phenotype following virus replication in vivo and for the high level of restriction of replication of cp45 in vivo (15, 18). Three recombinant viruses, rcp45 3'NL, rcp45 3'NCMFHN, and the previously described rcp45 L 942/992 (29), were more ts than rcp45, indicating the complex and interdependent nature of the mutations that contribute to the ts phenotype of cp45. The interaction of the ts mutations in cp45, therefore, does not appear to be additive as has been demonstrated for other attenuated viruses (32) but is more complex. A similar complex interaction between the three amino acid substitutions in the cp45 L protein was demonstrated previously (29). In that case, the interacting mutations were all within the same protein, L, whereas in the present case they are in separate protein or cis-acting elements.

At 20°C cp45 grows to a titer >10,000-fold higher than that of wt PIV3. This ca phenotype, like the ts phenotype, was found to be a complex, composite phenotype. The only genetic element that was identifiable as an independent contributor to the ca phenotype is the 3' N region, which provided a 100-fold increase in the ability of a recombinant virus to replicate at 20°C compared to wt PIV3. When the cp45 L mutations or the cp45 CMFHN set of mutations were added independently to the cp45 3' N mutations, an increase in replicative ability at 20°C was not observed, but when both were added to the 3' N mutations the full capacity of cp45 to grow at 20°C was reconstituted. The ca phenotype, therefore, results from an interaction of separate mutations present in at least three distant regions of the cp45 genome. The ca phenotype is also stable following replication in vivo (19), but the genetic basis for this stability is not immediately obvious from the present data. It is possible that cp45 revertants arising in vivo that have lost the ca phenotype do not have a selective advantage over input virus and therefore are not readily identified in isolates. The genetic basis of the ca phenotype of the paramyxovirus cp45 mutant is very different from that of the orthomyxovirus influenza A/Ann Arbor/6/60 ca mutant in which one gene, the PA polymerase gene, is solely responsible for the ca phenotype (5, 30).

Our findings indicate that multiple genetic elements also contribute to the att phenotype of cp45. The three mutations in L (29), the mutation in C, and one or both of the mutations in F each make an independent contribution to attenuation. Each of the attenuating mutations in L is a ts mutation, whereas those in C and F are non-ts attenuating mutations. It was not surprising to identify non-ts attenuating mutations in cp45, since the presence of such mutations had been previously predicted from studies of the replication of cp45 and its derivatives in the upper and lower respiratory tract of rhesus monkeys (14, 15). The presence of five or six mutations in three proteins which directly contribute to the attenuation phenotype explains in part the stability of this phenotype during replication in vivo (14). The two recombinant viruses, which were more ts than cp45, rcp45 3'NCMFHN and rcp45 3'NL, were not found to be more attenuated in vivo, for reasons that are not clear. The observation that the rcp45 3'NCMFHN virus was more ts in vitro than rcp45 but was less attenuated, suggested that this virus was not ts in the respiratory tract of hamsters. Such temperature-dependent host range mutants, i.e., viruses that are ts in one host but not in another, have been described previously (27, 28), but the genetic mechanism underlying this phenotype remains uncharacterized. The mutations present in the 3' leader and in the N gene (rcp45 3'N) resulted in a shutoff temperature of 40°C and did not appreciably restrict replication of these mutants in hamsters and, therefore, do not independently appear to play a major role in attenuation. In contrast, mutations in the 5' untranslated region (UTR) of the attenuated poliovirus vaccine strains are major determinants of the att phenotype (13). The 5' UTR mutations appear to result in defective interaction between the secondary structure adopted by the UTR and cellular factors involved viral RNA replication and translation.

The finding that cp45 contains multiple ts and non-ts attenuating mutations explains the high level stability of this virus even after extensive replication in vitro and in vivo. The complex interaction between the mutations may also enhance the stability of this virus. For example, loss of the three mutations in the L protein would yield a virus that was more ts than the cp45 virus. The identification of the genetic basis of the ts and att phenotypes of cp45 will allow us to monitor the presence of the attenuating mutations during all phases of manufacture and use in humans. Furthermore, if the cp45 vaccine candidate is found to be insufficiently attenuated in expanded phase II clinical studies, then the cp45 cDNA would serve as an excellent genetic backbone for the introduction of additional attenuating mutations, just as the respiratory syncytial virus cpts248/404 serves as an attenuated backbone for the introduction of additional attenuating mutations (36).

The identification of critical ts and att mutations in cp45 also will allow us to design attenuated candidate vaccines for the PIV1 and PIV2 viruses (33), using a PIV3 cp45 backbone. The strategy for producing a vaccine against PIV1 or PIV2 involves the substitution of the PIV1 or PIV2 HN and F glycoproteins for the PIV3 glycoproteins in rcp45. One possible obstacle to success in this strategy is that the mutations in the F and HN genes of cp45 are major determinants of att and, therefore, could not be easily replaced. Thus the PIV1- or PIV2-cp45 chimeric recombinant viruses might be less attenuated than PIV3 rcp45. In this study, we have shown that the mutation in the HN gene of cp45 does not play a major role in attenuation. However, one or both of the mutations in the F gene of cp45 are only moderately attenuating. In addition, it is likely that the attenuating mutations in the C and L proteins are dominant over the those in F, and the loss of the mutations in F might not have a detectable effect on overall attenuation. To address this issue, we are currently comparing the level of attenuation of rcp45 viruses that lack the F mutations with that of rcp45 itself. Importantly, it should also be possible to identify additional sites in the PIV3 genome, other than in the F or HN gene, that are targets for attenuation and can be used for this purpose in the construction of PIV1/PIV3 and PIV2/PIV3 chimeras for inclusion in a live attenuated viral vaccine. The development of a polyvalent PIV1/PIV2/PIV3 vaccine based on the attenuated cp45 background, using reverse genetics, should now be achievable.