Attenuated Human Parainfluenza Virus Type 1 (HPIV1) Expressing the Fusion Glycoprotein of Human Respiratory Syncytial Virus (RSV) as a Bivalent HPIV1/RSV Vaccine

ABSTRACT

Live attenuated recombinant human parainfluenza virus type 1 (rHPIV1) was investigated as a vector to express the respiratory syncytial virus (RSV) fusion (F) glycoprotein, to provide a bivalent vaccine against RSV and HPIV1. The RSV F gene was engineered to include HPIV1 transcription signals and inserted individually into three gene locations in each of the two attenuated rHPIV1 backbones. Each backbone contained a single previously described attenuating mutation that was stabilized against deattenuation, specifically, a non-temperature-sensitive deletion mutation involving 6 nucleotides in the overlapping P/C open reading frames (ORFs) (C^Δ170) or a temperature-sensitive missense mutation in the L ORF (L^Y942A). The insertion sites in the genome were pre-N (F1), N-P (F2), or P-M (F3) and were identical for both backbones. In vitro, the presence of the F insert reduced the rate of virus replication, but the final titers were the same as the final titer of wild-type (wt) HPIV1. High levels of RSV F expression in cultured cells were observed with rHPIV1-C^Δ170-F1, -F2, and -F3 and rHPIV1-L^Y942A-F1. In hamsters, the rHPIV1-C^Δ170-F1, -F2, and -F3 vectors were moderately restricted in the nasal turbinates, highly restricted in lungs, and genetically stable in vivo. Among the C^Δ170 vectors, the F1 virus was the most immunogenic and protective against wt RSV challenge. The rHPIV1-L^Y942A vectors were highly restricted in vivo and were not detectably immunogenic or protective, indicative of overattenuation. The C^Δ170-F1 construct appears to be suitably attenuated and immunogenic for further development as a bivalent intranasal pediatric vaccine.

IMPORTANCE There are no vaccines for the pediatric respiratory pathogens RSV and HPIV. We are developing live attenuated RSV and HPIV vaccines for use in virus-naive infants. Live attenuated RSV strains in particular are difficult to develop due to their poor growth and physical instability, but these obstacles could be avoided by the use of a vaccine vector. We describe the development and preclinical evaluation of live attenuated rHPIV1 vectors expressing the RSV F protein. Two different attenuated rHPIV1 backbones were each engineered to express RSV F from three different gene positions. The rHPIV1-C^Δ170-F1 vector, bearing an attenuating deletion mutation (C^Δ170) in the P/C gene and expressing RSV F from the pre-N position, was attenuated, stable, and immunogenic against the RSV F protein and HPIV1 in the hamster model and provided substantial protection against RSV challenge. This study provides a candidate rHPIV1-RSV-F vaccine virus suitable for continued development as a bivalent vaccine against two major childhood pathogens.

INTRODUCTION

Human respiratory syncytial virus (RSV) is the leading viral cause of severe acute respiratory infection (ARI) in infants and young children worldwide. RSV is an enveloped, nonsegmented, negative-strand RNA virus of the family Paramyxoviridae. RSV infects early in life and is responsible globally for an estimated 34 million annual pediatric cases of acute bronchiolitis and pneumonia, 4 million hospitalizations, and up to 199,000 pediatric deaths, 99% of which occur in the developing world (1). The human parainfluenza viruses (HPIVs) are also enveloped nonsegmented negative-strand RNA viruses of the family Paramyxoviridae. HPIV serotype 1 (HPIV1), HPIV2, and HPIV3 in particular are important agents of pediatric ARI and in aggregate are second in importance only to RSV (2). No vaccines are currently available for RSV or any of the HPIVs. Here we describe the development and evaluation of attenuated strains of HPIV1 as vectors to express the fusion (F) protein of RSV as a bivalent HPIV1/RSV vaccine.

A formalin-inactivated RSV vaccine evaluated in infants and children in the 1960s was poorly protective and, paradoxically, primed for greatly enhanced RSV disease (3, 4). Markers for RSV disease enhancement have also been observed with purified RSV subunit vaccines in experimental animals (5, 6). Therefore, inactivated and subunit vaccines are considered contraindicated in young RSV-naive recipients. In contrast, live attenuated RSV and live vectored RSV vaccines have been shown to be free of the effects of enhancing disease in experimental animals and in infants and children (7, 8). Similar to RSV, inactivated HPIV3 has also been reported to prime for enhanced HPIV3 disease in experimental animals following HPIV3 challenge (9). These observations indicate that live attenuated RSV and HPIV vaccines are preferred for infants and young children. These vaccines would be given intranasally (i.n.), which stimulates innate, cellular, and humoral immunity both systemically and in the respiratory tract. This immunization route in infants also decreases the virus-neutralizing and immunosuppressive effects of maternally derived serum antibodies (10). Local respiratory tract immunity is particularly effective in restricting the replication and transmission of these respiratory pathogens.

Our laboratory has been developing live attenuated RSV and HPIV vaccine strains by introducing a series of attenuating mutations into the respective viruses using reverse genetics (11, 12). Evaluation of these strains in clinical studies is ongoing. However, we are also pursuing the alternative approach described here of using attenuated HPIV strains to express RSV antigens because this strategy has a number of potential advantages. As noted above, this strategy provides a bivalent vaccine against RSV and the HPIV vector. In addition, the HPIVs replicate in cell culture to titers that are 10- to 100-fold greater than those of RSV, facilitating manufacture. RSV is also notorious for being susceptible to a loss of infectivity during handling, which complicates vaccine development, manufacture, and delivery. The HPIVs are substantially more stable, which may be critical for extending RSV vaccines to developing countries, where their need is the greatest. RSV grown in vitro typically forms long filaments that complicate manufacture, whereas the HPIVs form smaller spherical particles. RSV may also be inherently more pathogenic and possibly more immunosuppressive than the HPIVs, which would be another advantage of an HPIV-vectored RSV vaccine. We have also found in rodents that use of an HPIV-vectored vaccine as a boost subsequent to administration of a live attenuated RSV strain is more immunogenic than a second dose of the same attenuated RSV strain (unpublished data). This is likely because the RSV-specific immunity resulting from the primary immunization restricts a second dose of an attenuated RSV strain more efficiently than it does an HPIV-vectored virus.

The HPIV1 genome consists of 6 genes encoding the nucleoprotein (N), phosphoprotein (P/C), internal matrix protein (M), fusion glycoprotein (F), hemagglutinin-neuraminidase glycoprotein (HN), and large polymerase protein subunit (L) (2). Each gene encodes a major viral protein: N, P, M, F, HN, and L. The P gene carries an additional overlapping open reading frame (ORF) expressing a set of carboxy coterminal C accessory proteins that inhibit host interferon and apoptosis responses (13). Like other nonsegmented negative-strand RNA viruses, HPIV1 transcription initiates at the 3′ end of the genome and proceeds in a sequential start-stop process regulated by short gene start (GS), gene end (GE), and intergenic (IG) signals that flank each gene to generate a series of monocistronic mRNAs. There is a 3′-to-5′ gradient of decreasing transcription, with the promoter-proximal genes being expressed at higher levels (2, 14). Like other paramyxoviruses, complete infectious, replication-competent recombinant HPIV1 (rHPIV1) can be recovered in cell culture from transfected cDNAs by reverse genetics. HPIVs can accommodate and express several added foreign genes (15). However, we usually insert only a single foreign gene, because multiple genes can be overly attenuating and can accumulate point mutations. There are two RSV neutralization antigens that are also the major protective antigens: the F glycoprotein and the heavily glycosylated glycoprotein (G). The F protein is the RSV antigen of choice to be expressed from a vector because it is a more effective neutralization and protective antigen than G (16) and is also one of the most highly conserved proteins among RSV strains, whereas G is highly divergent.

Previous studies have described the development of an attenuated chimera of recombinant bovine and human parainfluenza virus type 3 (PIV3), namely, rB/HPIV3, expressing the RSV F protein as an experimental bivalent vaccine for RSV and HPIV3 (17–19). Clinical evaluation of an rB/HPIV3-RSV-F construct in seronegative children showed that it was infectious, well tolerated, and attenuated but was less immunogenic against RSV F than hoped (7). This appeared to be due at least in part to genetic instability that silenced expression of the RSV F insert in a substantial proportion of vector particles (20). However, further studies are under way to stabilize the RSV F insert and enhance immunogenicity (21). HPIV1 is another attractive vector for expressing RSV F protein. In particular, HPIV1 infects somewhat later in childhood (22, 23) than RSV or HPIV3 (23), and so an rHPIV1-vectored RSV vaccine might be used subsequent to a live attenuated RSV or rB/HPIV3-vectored vaccine to boost immune responses to RSV.

In the present study, we investigated the development of an rHPIV1-vectored vaccine expressing the RSV F protein from a codon-optimized ORF (Fig. 1; Table 1). We constructed two attenuated rHPIV1 backbones that each contained a different, previously described attenuating mutation (namely, C^Δ170 and L^Y942A; the viruses are designated rHPIV1-C^Δ170 and rHPIV1-L^Y942A, respectively) that had been designed for stability against deattenuation (Table 1) (24–26). The C^Δ170 mutation consists of a 6-nucleotide deletion in the overlapping P and C ORFs (Table 1). In the C ORF, this results in the deletion of 2 amino acids and the substitution of a third amino acid (specifically, the triplet 168-RDF-170 was changed to the single amino acid S), whereas in the overlapping P ORF, it results in the deletion of 2 amino acids (172-GF-173). The C^Δ170 mutation is non-temperature sensitive. It reduces the ability of C proteins to inhibit the host type I interferon response and apoptosis (13, 24, 26), resulting in viral attenuation. The other mutation is a missense mutation (942-Y to A, called L^Y942A) in the L ORF that was designed to involve 3 nucleotide substitutions so that it is highly resistant to deattenuation (Table 1), as has been confirmed previously (27). The L^Y942A mutation is temperature sensitive. Each of these mutations has been shown to be moderately attenuating in vivo (24–26). We compared the effects of inserting the RSV F gene at the first, second, or third gene position of the rHPIV1-C^Δ170 and rHPIV1-L^Y942A backbones. These six viruses were constructed and recovered by reverse genetics. They were analyzed, along with their respective attenuated parent and wild-type (wt) HPIV1, for replication, protein expression, and stability in vitro. The hamster model was used to assess in vivo replication (upper and lower respiratory tract), vaccine virus stability, immunogenicity, and protection against wt RSV challenge. The results identified an optimal construct suitable for further development.

FIG 1.

Construction of HPIV1 antigenomic cDNAs containing the RSV F gene inserted at the first (F1), second (F2), or third (F3) gene position. The rHPIV1 backbones contained either of two attenuating mutations: namely, the C^Δ170 or the L^Y942A mutation in the P/C or L ORF, respectively (indicated by * and ●, respectively). For the rHPIV1-F1 constructs, the RSV F gene was inserted at the MluI site located in the upstream nontranslated region of the N gene. For rHPIV1-F2, the RSV F gene was inserted between the rHPIV1 N and P genes at the AscI site located in the upstream nontranslated region of the P gene. For rHPIV1-F3, the RSV F gene was inserted between the rHPIV1 P and M genes at the NotI site situated in the downstream nontranslated region of the P gene. For all constructs, the RSV F ORF was codon optimized for human expression and contained the two HEK amino acid assignments (see the text). Each RSV F insert contained the N gene end (GE), intergenic (IG), and P gene start (GS) sequences, as shown, so that the sequence coding for RSV F would be expressed as an independent mRNA.

TABLE 1.

Attenuating mutations introduced in the HPIV1 backbone in the P/C or L ORF

Gene(s)	Mutation	ORF	Nucleotide changea → mutation	Type of mutation	Codon position	Amino acid change	No. (type) of nucleotide changes needed for reversion to wt
P/C	Δ170	C	AGG GAT TTC → AGC	Deletion	168–170	RDF → S	6 (insertions)
P/C	Δ170	P	AG GGA TTT C → AGC	Deletion	172–173	GF deletion	6 (insertions)
L	Y942A	L	TAT → GCG	Substitution	942	Y → A	3 (substitutions)

^aNucleotide changes (deletion or substitution) in the wt sequence are underlined in boldface.

MATERIALS AND METHODS

Cells and viruses.

LLC-MK2 (ATCC CCL-7) rhesus monkey kidney and Vero (ATCC CCL-81) African green monkey kidney cell lines were maintained in Opti-MEM I medium with GlutaMAX (Life Technologies, Grand Island, NY) supplemented with 5% fetal bovine serum (FBS; HyClone/Logan, UT) and 1 mM l-glutamine (Life Technologies). BHK BSR T7/5 cells are baby hamster kidney 21 (BHK-21) cells that constitutively express T7 RNA polymerase (28). These cells were maintained in Glasgow minimal essential medium (GMEM; Life Technologies) supplemented with 10% FBS, 2 mM l-glutamine, and 2% minimal essential medium amino acids (Life Technologies).

The strains used in this work both for viruses and for all cDNAs were HPIV1/Washington/20993/1964 (GenBank accession number AF457102) and RSV A2 (GenBank accession number M74568). HPIV1 and derivatives were propagated in LLC-MK2 cells in serum-free Opti-MEM I medium containing 1.2% trypsin (TrypLE Select; Life Technologies), 100 U/ml penicillin, 100 μg/ml streptomycin (Life Technologies), and 1 mM l-glutamine. HPIV1 titers were determined by 10-fold serial dilution in 96-well plates of LLC-MK2 cells in the same medium and incubation at 32°C for 7 days. Infected cells were detected by an hemadsorption (HAD) assay using guinea pig erythrocytes, and titers were calculated as the log₁₀ 50% tissue culture infective doses (TCID₅₀) per milliliter (24). The temperature-sensitive (ts) phenotype of each virus was studied by evaluating the efficiency of replication at 32, 35, 36, 37, 38, 39, and 40°C; this was done by serial dilution of viruses in 96-well replicate plates of LLC-MK2 cells incubated in sealed containers in temperature-controlled water baths at various temperatures for 7 days, followed by the HAD assay (29).

Design of rHPIV1-C^Δ170 and rHPIV1-L^Y942A expressing the RSV F protein.

The rHPIV1s were constructed using a previously described reverse genetics system for wt HPIV1 (30). The recombinant full-length antigenomic cDNA of HPIV1 was modified by site-directed mutagenesis to introduce 3 additional unique restriction sites: MluI (ACGCGT, pre-N position, nucleotide numbers 113 to 118), AscI (GGCGCGCC, N-P position, nucleotide numbers 1776 to 1783), and NotI (GCGGCCGC, P-M position, nucleotide numbers 3609 to 3616) (Fig. 1). Two attenuated backbones were generated by introducing either the C^Δ170 (25) or the L^Y942A (25, 27) mutation into the P/C or L ORF (Table 1), respectively, using a QuikChange Lightning mutagenesis kit (Agilent, Santa Clara, CA). For the rHPIV1 C^Δ170 mutation, the forward mutagenic primer was AAGAAGACCAAGTTGAGXCCAGAAGAGGTACGAAG and the reverse primer was CTTCGTACCTCTTCTGGXCTCAACTTGGTCTTCTT. These primers introduced a 6-nucleotide deletion (GGATTT), represented by the letter X in boldface, in the P/C ORF (Table 1), in keeping with the rule of six (31, 32). For the L^Y942A mutation, the forward primer was CCAGCTAACATAGGAGGGTTCAACGCGATGTCTACAGCTAGATGTTTTGTC and the reverse primer was GACAAAACATCTAGCTGTAGACATCGCGTTGAACCCTCCTATGTTAGCTGG; the site of the TAT (Y)-to-GCG (A) mutation at amino acid 942 in the L ORF is underlined. Clones with the desired mutation were identified and sequenced in their entirety by automated sequencing.

The RSV F ORF with HEK amino acid assignments Glu-66 and Pro-101 (33), which make its sequence identical at the amino acid level to that of an early passage of strain A2, was codon optimized for human expression as previously described (34) (GeneArt, Life Technologies, Grand Island, NY). The RSV F-gene insert was designed (Fig. 1) to include a set of HPIV1 transcription signals (GE-IG-GS) so that in the final construct the RSV F gene would be expressed as a separate mRNA. RSV F-gene inserts were generated by PCR with a flanking MluI, AscI, or NotI restriction site for insertion in the first pre-N (F1), second N-P (F2), or third P-M (F3) position, respectively (Fig. 1). The F1, F2, and F3 inserts were designed to adopt the hexamer spacing of the N, P, and P genes, respectively (32). The following PCR primers were used to generate the RSV F inserts. For the F1 gene position (MluI site), the forward primer was ACGCGTCCCGGGAACAATGGAACTGCTGATCCTGAAGGCCAACGCC and the reverse primer was ACGCGTCGTACGCATTCACCCTAAGTTTTTCTTACTTTCTATCAGTTGGAGAAGGCGATATTGTTGATGCCGG; for the F2 gene position (AscI site), the forward primer was GGCGCGCCCCCGGGAACAATGGAACTGCTGATCCTGAAGGCCAACGCC and the reverse primer was GGCGCGCCCGTACGCCATTCACCCTAAGTTTTTCTTACTTGATTCTATCAGTTGGAGAAGGCGATATTGTTGATGCCGG; and for the F3 gene position (NotI site), the forward primer was GCGGCCGCCCGGGAAGTAAGAAAAACTTAGGGTGAATGAACAATGGAACTGCTGATCCTGAAGGCCAACGCC and the reverse primer was GCGGCCGCCGTACGCTATCAGTTGGAGAAGGCGATATTGTTGATGCCGG. The restriction sites used for cloning are underlined, the GE-IG-GS signals are in boldface, and the translational start site (forward primer) and two stop codons (reverse primer) are italicized. PCR products were generated using an Advantage HF 2 PCR kit (Clontech, Mountain View, CA) and cloned into a TOPO TA cloning vector (Life Technologies), and the sequences were confirmed by automated sequencing. The RSV F MluI, AscI, and NotI fragments were cloned into the corresponding sites of the rHPIV1-C^Δ170 or rHPIV1-L^Y942A backbone. All viruses were designed to have the genome nucleotide length conform to the rule of six (32).

Recovery of rHPIV1-C^Δ170 and rHPIV1-L^Y942A expressing the RSV F protein.

The viruses were recovered in BHK BSR T7/5 cells as previously described (30, 35). Transfected cells were incubated overnight at 37°C and washed twice with Opti-MEM I medium (Life Technologies), and fresh Opti-MEM I medium containing 1 mM l-glutamine and 1.2% trypsin was added to the cells, followed by incubation at 32°C. At 48 h posttransfection, the cells were harvested by scraping them into the medium, and the cell suspension was added to 50% confluent monolayers of MK2 cells in Opti-MEM I medium containing 1 mM l-glutamine and 1.2% trypsin and incubated at 32°C. Virus was harvested after 7 days and was further amplified by one (rHPIV1-C^Δ170) or two (rHPIV1-L^Y942A) passages in LLC-MK2 cells at 32°C. All recombinant viruses were completely sequenced to confirm the lack of adventitious mutations except in the 82 and 164 nucleotides at the 3′ and 5′ terminal ends, respectively. For this, viral RNA was extracted (QIAamp viral RNA minikit; Qiagen, Valencia, CA) from virus stocks and treated with RNase-free DNase I (Qiagen) to remove the plasmid DNA used for virus rescue. RNA was reverse transcribed (SuperScript first-strand synthesis system for reverse transcription-PCR [RT-PCR]; Life Technologies), and overlapping genome regions were amplified by PCR (Advantage-HF 2 PCR kit). RT-PCR controls lacking reverse transcriptase were included for all viruses. These controls showed that the amplified products were derived from viral RNA and not from the antigenome cDNA used for virus recovery. The genome sequence of each virus construct was determined by automated sequencing of the uncloned overlapping amplified RT-PCR products and assembled using the Sequencher program, version 5.1 (Gene Codes Corporation, Ann Arbor, MI).

Analysis of RSV F and HPIV1 vector protein expression by Western blotting.

Vero cells (1 × 10⁶) were infected with the viruses at a multiplicity of infection (MOI) of 5 TCID₅₀ per cell, incubated at 32°C, and harvested at 48 h postinfection (p.i.) by lysis with 400 μl of 1× LDS sample buffer (Life Technologies). The lysates were reduced and denatured at 37°C for 30 min and subjected to electrophoresis on 4 to 12% bis-Tris NuPAGE gels (Novex-Life Technologies), and the separated proteins were transferred onto polyvinylidene difluoride (PVDF) membranes using an iBlot protein transfer system (Life Technologies). Membranes were blocked for 1 h in LI-COR blocking buffer (LI-COR Inc., Lincoln, NE) and probed with a murine monoclonal RSV F-specific antibody (catalog number ab43812; Abcam, Cambridge, MA) and the previously described (36) rabbit polyclonal antipeptide HPIV1 N-specific antiserum HPIV1-N-485, each of which was used at a 1:1,000 dilution in blocking buffer. Replicate blots performed with the same set of lysates were probed with rabbit polyclonal antipeptide antisera for HPIV1 P (SKIA-1), F (SKIA-15), or HN (SKIA-13) (produced by Mario Skiadopoulos), which were raised by immunization of rabbits with a keyhole limpet hemocyanin-conjugated peptide derived from each protein sequence (RDPEAEGEAPRKQESC, CYTLESRMRNPYMGNNSN, and KTNSSYWSTTRNDNSTVC, respectively) and which were used at a 1:200 dilution. A replicate blot probed with anti-GAPDH (anti-glyceraldehyde-3-phosphate dehydrogenase) antibody (catalog number ab9484; Abcam) was used as a loading control. After overnight incubation with the antibodies described above, the membranes were washed 4 times for 5 min each, followed by incubation for 1 h with infrared dye-labeled secondary antibodies diluted in the LI-COR blocking buffer: specifically, goat anti-mouse immunoglobulin IRDye 680LT, goat anti-rabbit immunoglobulin IRDye 800CW, and goat anti-mouse immunoglobulin IRDye 800CW (LI-COR). The membranes were scanned, and the blot images were acquired using an Odyssey infrared imaging system (LI-COR). The fluorescence intensities of the protein bands, derived from three independent experiments, were quantified by using the LI-COR image analysis suite (Image Studio) and are reported as the level of expression of RSV F or rHPIV1 vector proteins (N, P, F, and HN) relative to that of the respective F3 virus.

Double-staining fluorescence plaque assay to quantify viruses coexpressing HPIV1 and RSV antigens.

A double-staining fluorescence plaque assay was performed as previously described (34) with few modifications. Briefly, infected Vero cell monolayers were incubated for 6 days at 32°C under a 0.8% methylcellulose overlay containing 2 mM l-glutamine and 4% trypsin. For animal tissue-derived virus samples, ticarcillin-clavulanate (Timentin; 200 mg/ml), ampicillin (100 mg/ml), clindamycin (Cleocin; 150 mg/ml), and amphotericin B (250 μg/ml) were included in the overlay. Immunostaining was performed with a mixture of three RSV F-specific monoclonal antibodies (6) each at a 1:2,000 dilution plus an HPIV1-specific goat polyclonal antibody (Abcam) at a 1:1,600 dilution. The secondary antibodies were infrared dye-conjugated goat anti-mouse immunoglobulin 680LT and donkey anti-goat immunoglobulin 800CW (LI-COR), each of which was used at a 1:800 dilution. Images were acquired using an Odyssey infrared imaging system. The secondary antibody signals were pseudocolored to appear red and green for the detection of RSV F and HPIV1 antigens, respectively. The percentage of rHPIV1 plaques expressing RSV F protein was determined by merging the colors: plaques that appear yellow upon merging express both HPIV1 antigens and the RSV F protein, and those that appear green express HPIV1 antigens but not the RSV F protein.

Hamster studies.

All animal studies were approved by the National Institutes of Health (NIH) Institutional Animal Care and Use Committee (IACUC). Six-week-old golden Syrian hamsters were confirmed to be seronegative for HPIV1 and RSV by hemagglutination inhibition (HAI) assay and a plaque reduction assay (see below), respectively (37–39). To assess vector replication in vivo, hamsters in groups of 12 per virus were anesthetized, and each animal was inoculated i.n. with 0.1 ml of L15 medium (Life Technologies) containing 10⁵ TCID₅₀ of virus. Six hamsters per virus were euthanized per day on days 3 and 5 p.i., and nasal turbinates and lungs were collected and tissue homogenates were prepared in L15 medium containing ticarcillin-clavulanate (200 mg/ml), ampicillin (100 mg/ml), clindamycin (150 mg/ml), and amphotericin B (250 μg/ml), clarified, and titrated by serial dilution on LLC-MK2 cells by HAD assay. Virus titers are reported as TCID₅₀ per gram of hamster tissue.

To assess vector immunogenicity, each animal in groups of six hamsters per virus was inoculated i.n. with 10⁵ TCID₅₀ of the six rHPIV1-RSV-F vectors, and wt RSV, rHPIV1-C^Δ170, rHPIV1-L^Y942A, and rB/HPIV3-F2 were included as controls. Sera were collected at day 28 after immunization. The titers of RSV-specific neutralizing antibodies (NAbs) were determined by 60% plaque reduction neutralization tests (PRNT₆₀) on Vero cells in the presence of guinea pig complement (37) using enhanced green fluorescent protein (eGFP)-expressing RSV (40). The titers of HPIV1-specific NAbs on Vero cells were determined by a PRNT₆₀ using green fluorescent protein (GFP)-expressing rHPIV1 (rHPIV1-GFP), essentially as described above for RSV, with three modifications: (i) guinea pig complement was not used, as it was found to neutralize HPIV1; (ii) the inoculated Vero cells were washed twice with 1× phosphate-buffered saline after virus adsorption to remove serum; and (iii) the methylcellulose overlay lacked FBS and contained 4% trypsin.

Protection against RSV infection was tested by challenge infection of hamsters from the immunogenicity study described above at 30 days p.i. by i.n. inoculation with 0.1 ml L15 medium containing 10⁶ PFU of wt RSV strain A2. Hamsters were euthanized, and nasal turbinates and lungs were collected at 3 days postchallenge. The viral loads of the challenge RSV in these tissues were determined by plaque assay on Vero cells (41, 42).

RESULTS

Creation of two attenuated HPIV1 backbones (rHPIV1-C^Δ170 and rHPIV1-L^Y942A) expressing the RSV F protein from three different genome locations.

The C^Δ170 and L^Y942A mutations were individually introduced into rHPIV1 to create two attenuated versions of HPIV1, rHPIV1-C^Δ170 and rHPIV1-L^Y942A (Table 1). The RSV F ORF of strain A2 was optimized for human codon usage and engineered to be under the control of a set of HPIV1 transcription signals. Each was inserted into the rHPIV1-C^Δ170 and rHPIV1-L^Y942A backbones at three different genome locations, namely, at the first gene position (pre-N, yielding rHPIV1-C^Δ170-F1 and rHPIV1-L^Y942A-F1), at the second gene position (N-P, yielding rHPIV1-C^Δ170-F2 and rHPIV1-L^Y942A-F2), and at the third gene position (P-M, yielding rHPIV1-C^Δ170-F3 and rHPIV1-L^Y942A-F3) (Fig. 1). Each vector gene maintained its original hexamer phasing, while the F1, F2, and F3 inserts had the hexamer phasing of the N, P, and P genes, respectively. The RSV F protein also carried the HEK amino acid assignments Glu and Pro at residues 66 and 101, respectively (33), and therefore, its sequence is identical at the amino acid level to the sequence of RSV F from an early passage of strain A2 (see Discussion).

The rHPIV1-RSV-F viruses were recovered by reverse genetics. All viruses were readily rescued, except for the rHPIV1-L^Y942A-F2 construct, which required multiple passages to make a working pool. Complete genome sequencing showed that all of the rHPIV1-RSV-F working pools except for rHPIV1-L^Y942A-F2 were free of adventitious mutations. For rHPIV1-L^Y942A-F2, we had to prepare nine independently rescued clones in order to find one that lacked adventitious mutations. The other eight clones of this virus contained various adventitious mutations (four of these genomes were sequenced fully; the other four were sequenced for the RSV F insert and flanking gene regions). In five of these viruses, there were various mutations in the GE, IG, and GS transcriptional signals of the gene junction immediately upstream of the RSV F ORF and sometimes in the RSV F-gene sequence (not shown). These likely reduced or ablated the expression of RSV F, although that was not examined. The other three clones had one or two missense mutations in the N and/or L gene (not shown) that were not further investigated.

Replication of the rHPIV1-RSV-F vectors in Vero and LLC-MK2 cells.

Replication of the rHPIV1-RSV-F vectors in vitro was evaluated by determining their multistep growth kinetics in Vero (Fig. 2A and C) and LLC-MK2 (Fig. 2B and D) cells. On day 7, the final titers of all of the viruses in both cell lines were greater than 7.2 log₁₀ TCID₅₀/ml in Vero cells and 7.4 log₁₀ TCID₅₀/ml in LLC-MK2 cells, with slight differences among the viruses that were statistically insignificant (Fig. 2). These titers were also statistically indistinguishable from those of wt HPIV1.

FIG 2.

Multistep virus replication in Vero (A and C) and LLC-MK2 (B and D) cells. Triplicate wells of cell monolayers in 6-well plates were infected at an MOI of 0.01 TCID₅₀ with rHPIV1-C^Δ170 or -L^Y942A expressing the RSV F gene (F1, F2, or F3), with the rHPIV1-L^Y942A or rHPIV1-C^Δ170 empty vector, or with wt HPIV1. Cultures were incubated at 32°C. Aliquots of cell culture medium were collected at 24-h intervals, and virus titers (log₁₀ TCID₅₀ per milliliter) were determined by serial dilution on LLC-MK2 cells and HAD assay. Mean titers with SEMs are shown. One-way analysis of variance with Tukey's multiple-comparison posttest was used to determine the statistical significance of the difference between the titer of each virus and that of wt HPIV1 at day 2 p.i. (a time of exponential replication) and at day 7 p.i. (when the final titer had been reached), which is indicated by asterisks, as follows: *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P < 0.0001.

However, some differences were observed during the period of exponential replication, e.g., at day 2 p.i. In the case of the rHPIV1-C^Δ170 vectors (Fig. 2A and B), the rHPIV1-C^Δ170 empty vector and rHPIV1-C^Δ170-F3 construct replicated similarly to wt HPIV1 in both cell lines on day 2 p.i. However, the replication of rHPIV1-C^Δ170-F1 was significantly reduced in Vero (P < 0.001) and LLC-MK2 (P < 0.05) cells compared to that of wt HPIV1, and the replication of rHPIV1-C^Δ170-F2 in Vero cells was significantly reduced (P < 0.01) compared to that of wt HPIV1. For rHPIV1-L^Y942A (Fig. 2C and D), replication of the rHPIV1-L^Y942A empty vector was significantly lower than that of wt HPIV1 in Vero cells on day 2 p.i., but both grew to similar titers in LLC-MK2 cells. Highly significant reductions (P < 0.0001) in replication compared to that of wt HPIV1 were observed for rHPIV1-L^Y942A-F1, -F2, and -F3 in Vero cells on day 2 p.i. Compared to the growth of the rHPIV1-L^Y942A empty vector, rHPIV1-L^Y942A-F1 grew at the same rate in Vero cells, while the growth of rHPIV1-L^Y942A-F2 and -F3 (Fig. 2C) was reduced, but the differences were statistically insignificant. In LLC-MK2 cells, rHPIV1-L^Y942A-F1, -F2, and -F3 showed significantly reduced (P < 0.01, P < 0.0001, and P < 0.01, respectively) replication compared to that of wt HPIV1; only the replication of rHPIV1-L^Y942A-F2 was significantly reduced (P < 0.01) relative to that of the empty vector on day 2 p.i.

Expression of RSV F and HPIV1 proteins by the rHPIV1-RSV-F vectors.

Expression of the RSV F protein and the HPIV1 vector N, P, F and HN proteins was evaluated by Western blotting analyses. Representative blots from one of three independent experiments for the rHPIV1-C^Δ170 and rHPIV1-L^Y942A constructs are shown in Fig. 3A, and results from all three experiments are quantified in Fig. 3B and C for the rHPIV1-C^Δ170 and rHPIV1-L^Y942A constructs, respectively, with the values normalized relative to those of the F3 construct in each series as 1.0.

FIG 3.

Western blot analysis of the in vitro expression of RSV F and HPIV1 proteins by the rHPIV1-RSV-F vectors. (A) Vero cells were infected with the indicated viruses at an MOI of 5 TCID₅₀. At 48 h p.i., cells were lysed with SDS sample buffer. All samples were denatured, reduced, and subjected to SDS-PAGE. Proteins were transferred onto PVDF membranes and probed with either an RSV F-specific mouse monoclonal antibody or rabbit polyclonal antipeptide antibodies monospecific to the HPIV1 N, P, HN, or F protein. GAPDH was used as a loading control. Corresponding infrared dye-conjugated anti-mouse and anti-rabbit antibodies were used as secondary antibodies. Images were acquired using an Odyssey infrared imaging system. The RSV F and the rHPIV1 vector proteins appear red and green, respectively. The images shown are representative of three independent experiments. (B and C) The intensities of protein bands from the three independent Western blot experiments, one of which is shown in panel A, were quantified, and the expression is shown relative to that of the corresponding F3 virus, which was given a value of 1.0. Plots show the data as means ± SEMs. The statistical significance of the difference in expression of the indicated HPIV1 proteins between the indicated viruses was analyzed by one-way analysis of variance with Dunnett's multiple-comparison test using the 95% confidence interval and is indicated as follows: *, P < 0.05; **, P < 0.01; ****, P < 0.0001.

The RSV F protein is synthesized as a precursor (F₀) that is cleaved twice by a cellular protease into disulfide-linked F₁ and F₂ subunits and a short peptide, p27. Immunostaining with an F-specific monoclonal antibody detected the F₀ (70-kDa) and F₁ (48-kDa) forms. The rHPIV1 C^Δ170-F1, -F2, and -F3 constructs expressed substantial amounts of RSV F (Fig. 3A and B) that were only slightly higher for F1 and F2 than F3 (Fig. 3B), with the differences being statistically insignificant. Thus, contrary to expectations, a 3′-5′ polar gradient of expression of the RSV F protein from F1 to F2 was not observed for this vector backbone, and there was only a modest reduction from F2 to F3. In contrast, however, a strong polar gradient of RSV F expression was observed in the case of the rHPIV1-L^Y942A constructs, with the expression of RSV F by the F1 virus being significantly higher than that by the F2 (P < 0.05) and F3 (P < 0.05) viruses (Fig. 3A and C).

Evaluation of the vector N, P, F, and HN protein expression showed that it was generally not greatly affected by the C^Δ170 mutation: specifically, the rHPIV1-C^Δ170 empty vector had a vector protein expression profile similar to that of wt HPIV1 (Fig. 3B). For the three versions of the rHPIV1-C^Δ170 vector expressing RSV F, the F3 virus expressed N, P, F, and HN at levels similar to those for the empty rHPIV1-C^Δ170 vector. The F2 virus expressed N protein similarly to the empty rHPIV1-C^Δ170 vector but showed reduced expression of P, F, and HN proteins, of which only the HN reduction was statistically significant compared to the empty vector. The F1 virus demonstrated a significant reduction in the expression of all vector proteins, including N (P < 0.05), P (P < 0.01), F (P < 0.05), and HN (P < 0.05), compared to that of the empty vector. Thus, insertion of the RSV F gene into the rHPIV1-C^Δ170 vector reduced the expression of downstream vector genes in all constructs except F3.

The L^Y942A mutation had a greater effect on the expression of vector proteins: specifically, compared with the levels of protein expression by wt HPIV1, the rHPIV1-L^Y942A empty vector had significantly reduced levels of expression of the P (P < 0.05), F (P < 0.05), and HN (P < 0.01) proteins, with no significant difference being found for the N protein (Fig. 3A and C). In the case of the three versions of rHPIV1-L^Y942A expressing RSV F, the expression of the N, P, F, and HN proteins by the F1 virus was very similar to that by the rHPIV1-L^Y942A empty vector, whereas the expression by the F3 virus was somewhat lower than that by the empty vector, but not significantly so. In contrast, the F2 virus exhibited a highly pronounced and significant reduction in the expression of the N, P, F, and HN proteins compared to the expression by the rHPIV1-L^Y942A empty vector, although the reduction in the expression of N was less than that for the other proteins.

Temperature sensitivity of the rHPIV1-RSV-F vectors.

As noted above, the C^Δ170 mutation did not confer the temperature-sensitive (ts) phenotype in previous studies, whereas the L^Y942A mutation did (26, 27). Insertion of RSV F into an HPIV vector has also been shown to confer the ts phenotype (21). We therefore evaluated the ability of the rHPIV1-RSV-F constructs to grow in LLC-MK2 cells at 32, 35, 36, 37, 38, 39, and 40°C (Table 2). While the C^Δ170 mutation did not confer the ts phenotype in previous studies, in the present study the rHPIV1-C^Δ170 empty vector had a shutoff temperature of 40°C (Table 2) and was thus ts, whereas wt HPIV1 was not, but the effect was small. The rHPIV1-C^Δ170-F2 and -F3 constructs also had shutoff temperatures of 40°C and thus did not differ significantly from the empty vector. However, rHPIV1-C^Δ170-F1 had a lower shutoff temperature of 39°C (Table 2). In the present study, the rHPIV1-L^Y942A empty vector had a shutoff temperature of 36°C, similar to the values of 35 to 37°C associated with this mutation in previous studies (25, 27). The rHPIV1-L^Y942A-F1 and -F2 constructs were more ts than the empty vector, having shutoff temperatures of 35°C, while the rHPIV1-L^Y942A-F3 vector had the same 36°C shutoff temperature as the empty vector (Table 2). Thus, insertion of the RSV F gene into the F1 position of either attenuated backbone increased the ts phenotype, insertion into the F3 position of either backbone did not increase the ts phenotype, and insertion into the F2 position increased the ts phenotype only for the L^Y942A backbone.

TABLE 2.

Temperature sensitivity of recombinant viruses on LLC-MK2 cell monolayers^a

Virus	Virus titer (log10 TCID50/ml) at:
	32°C	35°C	36°C	37°C	38°C	39°C	40°C
wt HPIV1	7.7	8.5	7.5	7.5	8.0	7.2	6.2
rHPIV1-CΔ170	7.0	7.0	6.5	6.2	6.7	6.5	3.2
rHPIV1-LY942A	8.0	7.2	5.2	4.5	2.2	≤1.2	≤1.2
rHPIV1-CΔ170-F1	7.0	6.7	7.0	6.2	6.2	3.2	1.5
rHPIV1-CΔ170-F2	8.2	6.5	7.7	7.0	7.7	6.5	2.2
rHPIV1-CΔ170-F3	8.0	8.0	7.5	8.0	7.7	6.7	2.0
rHPIV1-LY942A-F1	7.0	4.7	4.7	1.5	1.5	≤1.2	≤1.2
rHPIV1-LY942A-F2	5.5	3.2	≤1.2	≤1.2	≤1.2	≤1.2	≤1.2
rHPIV1-LY942A-F3	8.2	7.2	5.5	4.2	≤1.2	≤1.2	≤1.2

^aSerial dilutions of each of the indicated viruses were inoculated on LLC-MK2 cells and incubated at the indicated temperature for 7 days. Virus was detected by HAD assay. The detection limit was 1.2 TCID₅₀/ml. Underlined values in boldface indicate the virus shutoff temperature, defined as the lowest restrictive temperature at which the mean log₁₀ reduction in virus titer at a given temperature versus that at 32°C was 2.0 log₁₀ TCID₅₀/ml or greater than that of wt HPIV1 at the same two temperatures. The shutoff temperature of wt HPIV1 is >40°C (not shown); viruses with shutoff temperatures of ≤40°C are considered temperature sensitive.

Stability of RSV F-protein expression by the rHPIV1-RSV-F vectors after in vitro replication.

The working pools of the rHPIV1-RSV-F constructs were evaluated for the frequency of RSV F expression in individual viral plaques using a double-staining fluorescence plaque assay with antibodies specific to RSV F and the HPIV1 antigens. A total number of 140, 77, and 59 plaques were counted for rHPIV1-C^Δ170-F1, -F2, and -F3, respectively, and 100% of the rHPIV1 plaques expressed the RSV F protein. F1 made smaller plaques than F2 and F3, while the sizes of the F2 and F3 plaques were similar to each other (data not shown). A total of 214, 70, and 192 plaques were counted for rHPIV1-L^Y942A-F1, -F2, and -F3, respectively, and 100%, 100%, and 97% of these rHPIV1 plaques expressed the RSV F antigen, respectively. Overall, rHPIV1-L^Y942A formed plaques that were much smaller than those formed by the rHPIV1-C^Δ170 constructs. Since the assay was performed at the permissive temperature of 32°C, the reductions in plaque size, which suggest a relatively more attenuated phenotype and slower spread, may not be a ts effect. The reduced plaque size was consistent with their relatively slower replication profiles (Fig. 2).

Replication of the rHPIV1-RSV-F vectors in the respiratory tract of hamsters.

Viruses were evaluated for their ability to replicate in the upper respiratory tract (URT) and lower respiratory tract (LRT) of hamsters. The virus titers in the nasal turbinates (Fig. 4A) and lungs (Fig. 4B) on days 3 and 5 are shown. The rHPIV1-C^Δ170 empty vector was significantly more attenuated than wt HPIV1 in the nasal turbinates on day 5 and in the lungs on both days, with no virus being detected in the lungs of 5/6 animals on day 5. The addition of the RSV F insert provided further attenuation to the rHPIV1-C^Δ170 vector: for the F1 derivative, this difference in attenuation from that of the empty vector was significant only on day 3 in the lungs, for the F2 derivative this was significant on both days in the nasal turbinates and on day 3 in the lungs, and for the F3 derivative this was significant on day 5 in the nasal turbinates and day 3 in the lungs.

FIG 4.

Replication of the rHPIV1-RSV-F vectors in the nasal turbinates (A) and lungs (B) of hamsters. Hamsters were inoculated i.n. with 10⁵ TCID₅₀ of the rHPIV1-C^Δ170 or -L^Y942A vectors expressing the RSV F gene (F1, F2, or F3), the rHPIV1-L^Y942A or rHPIV1-C^Δ170 empty vector, wt HPIV1, rHPIV1-C^R84GC^Δ170HN^553AL^Y942A (HPIV1 vaccine candidate), or rB/HPIV3-F2 (a chimeric bovine/human PIV3 expressing RSV F from the 2nd position). Animals were sacrificed on days 3 and 5 p.i. (6 animals per virus per day), and the nasal turbinates and lungs were collected and prepared as tissue homogenates. Virus titers (log₁₀ TCID₅₀ per gram of tissue) were determined by serial dilution on LLC-MK2 cells and HAD assay. Titers for individual animals are shown for day 3 (Δ) and day 5 (●). The mean for each group is shown by horizontal dashed and solid lines for days 3 and 5, respectively, and the values obtained on days 3 and 5 are shown in boldface and italics, respectively. The limit of detection (LOD) was 1.5 log₁₀ TCID₅₀/ml and is indicated with a dotted line. The statistical significance of the difference between each virus and wt HPIV1 or the respective empty vector was determined by one-way analysis of variance with Tukey's multiple-comparison posttest using the 95% confidence interval and is indicated at the lower and upper bars, respectively, as follows: *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001; ns, not significant (P > 0.05).

The rHPIV1-L^Y942A empty vector was significantly more attenuated than wt HPIV1 on both days in both the nasal turbinates and lungs, with no virus being detected in the nasal turbinates on day 5 or in the lungs on either day. Thus, this mutation was indeed strongly attenuating in vivo. Because the empty vector was so strongly attenuated (i.e., virus was detected only in the nasal turbinates and only on 1 day), the effect of further addition of the RSV F insert was difficult to determine. For the RSV F derivatives, no virus was detected on either day in the lungs (except for a single animal in the F3 group) and nasal turbinates (except for the F2 group on day 3). Thus, a general lack of detectable virus replication in the nasal turbinates and lungs precluded comparison of the constructs in these anatomical compartments.

We also included for comparison the rHPIV1-C^R84G/Δ170HN^553AL^Y942A virus (designated the rHPIV1 vaccine candidate in Fig. 4), which was previously demonstrated to be strongly attenuated in African green monkeys (25) and was recently found to be overattenuated in seronegative children (43). In the present study, this virus did not replicate detectably in either the nasal turbinates or the lungs of the hamsters. As another control, we also included the chimeric bovine/human PIV3 expressing RSV F from the second gene position (rB/HPIV3-F2). In a previous clinical study by others (7), a similar version of rB/HPIV3-F2 replicated to a moderate titer in 6- to 24-month-old children seronegative for RSV and HPIV3 and was well tolerated, suggesting that it has an appropriate level of attenuation (7). In the present study, this control replicated somewhat more efficiently than the rHPIV1-C^Δ170-based vectors in the nasal turbinates and replicated to low titers in the lungs, where the rHPIV1-RSV-F viruses were almost completely restricted.

Stability of RSV F-protein expression by the rHPIV1-RSV-F vectors after in vivo replication.

To evaluate the rHPIV1 vectors for in vivo stability, Vero cells were infected with serially diluted homogenates of the nasal turbinates and lungs of the infected hamsters. The double-staining immunofluorescence plaque assay was performed to determine the percentage of viral plaques expressing RSV F. Consistent with the general lack of replication in the lungs (Fig. 4B), no plaques could be detected in the lung homogenates for any vector. Similarly, no plaques were detectable for the rHPIV1-L^Y942A-RSV-F vectors in the nasal turbinates. Stability could be assessed for the rHPIV1-C^Δ170-RSV-F viruses in nasal turbinate specimens, as shown in Table 3. Of the 30 samples analyzed on days 3 and 5, 29 had 100% of plaques expressing RSV F and 1 had 98% of plaques expressing RSV F. These data suggest that the rHPIV1-C^Δ170 vectors expressing RSV F are relatively stable in the hamster model, with little evidence of the emergence of variants with silenced RSV F expression after in vivo replication for 3 to 5 days being present.

TABLE 3.

Percentage of virus population expressing RSV F after in vivo replication^a

Virus	Day 3		Day 5
	% PFU expressing RSV F	Total no. of plaques	% PFU expressing RSV F	Total no. of plaques
rHPIV1-CΔ170-F1	100 (6/6)b	283	100 (5/6)	132
			98 (1/6)	51
rHPIV1-CΔ170-F2	100 (6/6)	87	NDc	ND
rHPIV1-CΔ170-F3	100 (6/6)	503	100 (6/6)	131

^aThe percentage of the virus population expressing RSV F after in vivo replication was determined by a double-staining immunofluorescence plaque assay. Vero cells were infected with serially diluted tissue homogenates of the nasal turbinates collected on days 3 and 5 p.i. from hamsters infected with rHPIV1-C^Δ170-F1, -F2, and -F3 (n = 6 animals per virus per day; the results of the experiment are shown in Fig. 4) and incubated for 6 days under a methylcellulose overlay. Viral plaques were stained with mouse monoclonal anti-RSV F and goat polyclonal anti-HPIV1 antibodies, followed by detection with corresponding infrared dye-conjugated secondary antibodies. The percentages of plaques expressing both RSV F and HPIV1 antigens are shown. rHPIV1-C^Δ170-F1, -F2, and -F3 in lung tissue samples and rHPIV1-L^Y942A-F1, -F2, and -F3 in nasal turbinate and lung tissue samples could not be tested due to their lack of replication in these tissues (Fig. 4).

^bData in parentheses represent the number of hamsters in that group of 6 for which the indicated percentage applies.

^cND, no plaques were detected.

Induction of serum NAbs against RSV and HPIV1.

Sera from immunized hamsters were collected at 28 days p.i., and the neutralizing antibody (NAb) titers against RSV and HPIV1 were determined by PRNT₆₀ (Table 4). The rHPIV1-C^Δ170-F1, -F2, and -F3 constructs induced substantial titers of RSV-specific NAbs that were not statistically different from each other, although the titer of the F1 construct was the highest. However, these were significantly lower than that of rB/HPIV3-F2, a difference that likely stems from the significantly reduced in vivo replication of these viruses compared to that of rB/HPIV3-F2 (Fig. 4A and B; see Discussion). The rHPIV1-L^Y942A-F1, -F2, and -F3 viruses failed to induce a detectable NAb response to RSV (Table 4). This is consistent with their high degree of restriction in the hamster (Fig. 4).

TABLE 4.

RSV- and HPIV1-neutralizing serum antibody responses to the rHPIV1-RSV-F vectors^a

Immunizing virus	Neutralizing serum antibody responseb (mean reciprocal log2 PRNT60 ± SE) to:
	RSV		HPIV1
	Preimmunization	Day 28	Preimmunization	Day 28
rHPIV1-CΔ170	≤3.3	≤3.3 (A)	≤1	3.9 ± 0.3 (A)
rHPIV1-LY942A	≤3.3	≤3.3 (A)	≤1	≤1 (B)
rHPIV1-CΔ170-F1	≤3.3	7.3 ± 0.3 (B, C)	≤1	2.8 ± 0.5 (A)
rHPIV1-CΔ170-F2	≤3.3	4.7 ± 0.7 (C)	≤1	≤1 (B)
rHPIV1-CΔ170-F3	≤3.3	6.7 ± 0.8 (C, B)	≤1	≤1 (B)
rHPIV1-LY942A-F1	≤3.3	≤3.3 (A)	≤1	≤1 (B)
rHPIV1-LY942A-F2	≤3.3	≤3.3 (A)	≤1	≤1 (B)
rHPIV1-LY942A-F3	≤3.3	≤3.3 (A)	≤1	≤1 (B)
rB/HPIV3-F2	≤3.3	9.7 ± 0.4 (D)	≤1	≤1 (B)
wt RSV	≤3.3	11.3 ± 0.4 (D)	≤1	≤1 (B)

^aGroups of 6-week-old hamsters (n = 6) were immunized i.n. with 10⁵ TCID₅₀ of each indicated virus in a 0.1-ml inoculum.

^bSerum samples were collected on day 0 prior to immunization and at day 28 p.i. Antibody titers against RSV and HPIV1 were determined by using 60% plaque reduction neutralization tests (PRNT₆₀) and GFP-expressing viruses. The limits of detection were 3.3 and 1.0 reciprocal log₂ PRNT₆₀ for RSV and HPIV1, respectively. The statistical significance of the differences among the groups for RSV antibody titers was determined by one-way analysis of variance with Tukey's multiple-comparison posttest (P < 0.05), and that for HPIV1 antibody titers was determined by an unpaired t test. Mean neutralizing antibody titers were categorized into groups A, B, C, and D, indicated in parentheses. Titers with different letters are statistically different from each other; titers with two letters are not statistically different from those with either letter.

The HPIV1-specific serum NAb responses in hamsters were also evaluated (Table 4). Overall, the HPIV1 NAb titers were lower than the RSV NAb titers. As indicated in Materials and Methods, guinea pig complement was included for RSV but was excluded from the HPIV1 neutralization assay, as it directly neutralized HPIV1. The lack of complement likely accounts for the generally lower titers of HPIV1-specific NAbs. For example, the rHPIV1-C^Δ170-F2 and -F3 viruses did not induce detectable levels of HPIV1 NAbs, even though they induced significant titers of RSV-specific NAbs, as noted above. Only the rHPIV1-C^Δ170 empty vector and the rHPIV1-C^Δ170-F1 construct induced detectable HPIV1-specific NAb titers, which were not statistically different from each other. None of the rHPIV1-L^Y942A viruses, including the empty vector, induced detectable levels of HPIV1-specific NAbs, consistent with their highly restricted replication in hamsters.

Protection against wt RSV challenge.

The hamsters that were immunized as described above (Table 4) were challenged i.n. on day 30 postimmunization with 10⁶ PFU of wt RSV (strain A2). Hamsters were euthanized on day 3 postchallenge, and the RSV titers in the nasal turbinates and lungs were determined by plaque assay on Vero cells to assess protection against RSV replication (Fig. 5A and B). The protection provided by vaccine candidates generally correlated with their ability to induce RSV-specific serum NAbs. In the nasal turbinates, the rHPIV1-C^Δ170-F1 and -F3 viruses provided a significantly greater restriction of RSV replication (P < 0.0001 and P < 0.05, respectively) compared to that provided by the empty vector. In contrast, the F2 virus did not induce significant protection in the nasal turbinates. In the lungs, the rHPIV1-C^Δ170-F1, -F2, and -F3 viruses each reduced the mean RSV titers compared to that with the empty vector, but only the reduction by the F1 virus was statistically significant (P < 0.05). The rB/HPIV3-F2 control provided significant protection against RSV challenge in the nasal turbinates (P < 0.0001) and lungs (P < 0.05), consistent with our previous report (21). The levels of protection afforded by rHPIV1-C^Δ170-F1 and rB/HPIV3-F2 were very similar. The rHPIV1-L^Y942A-F1, -F2, and -F3 viruses did not provide protection against RSV challenge, with the titers of the challenge RSV loads being very similar to that of the rHPIV1-L^Y942A empty vector. This was consistent with their lack of in vivo replication and immunogenicity against RSV.

FIG 5.

Protection against challenge wt RSV replication in the nasal turbinates (A) and lungs (B) of hamsters previously immunized with the rHPIV1-RSV-F vectors. Hamsters (n = 6 per group) were inoculated i.n. with the viruses indicated along the x axis (the same animals were used to evaluate immunogenicity; Table 4), and at 30 days p.i., each animal was challenged i.n. with 10⁶ PFU of wt RSV A2. Three days later, the animals were sacrificed, the nasal turbinates and lungs were collected and prepared as tissue homogenates, and the titer of challenge RSV (log₁₀ PFU per gram of tissue) was determined by plaque assay on Vero cells. The mean value for each group is shown as a boldface number and by a horizontal bar. The statistical significance of the differences between viruses was determined by one-way analysis of variance with Tukey's multiple-comparison posttest using the 95% confidence interval and is indicated as follows: *, P < 0.05; **, P < 0.01; ****, P < 0.0001; ns, not significant (P > 0.05).

DISCUSSION

The goal of this study was to develop rHPIV1 as a vector expressing the RSV F protein to provide a live attenuated bivalent vaccine against RSV and HPIV1. This would enable simultaneous immunization against two important pediatric viruses with a single bivalent vaccine. Each of these pathogens lacks an approved vaccine, and the characteristics of epidemiology and the disease of RSV and HPIV1 overlap. Thus, an HPIV/RSV vaccine is a logical combination that would broaden the coverage against pediatric respiratory tract disease. Attenuation for RSV and the HPIVs must be done carefully to avoid overattenuation and loss of immunogenicity. Therefore, we evaluated several variables in an effort to identify suitable rHPIV1-RSV-F constructs.

First, we evaluated two rHPIV1 backbones that each contained a different HPIV1-attenuating mutation developed in previous studies (24–26, 35), namely, the C^Δ170 and the L^Y942A mutations. Each of these was designed for stability against deattenuation, and both are present (together with additional mutations) in the vaccine candidate rHPIV1-C^R84G/Δ170HN^553AL^Y942A that was recently found to be overattenuated in HPIV1-seronegative infants and children (43). Since the insertion of an added heterologous gene typically confers attenuation, in the present study we included only a single mutation, C^Δ170 or L^Y942A, into each backbone, in addition to the RSV F insert, in order to avoid overattenuation.

A second variable was that the C^Δ170 and L^Y942A mutations have different mechanisms of attenuation. Specifically, the C^Δ170 mutation reduces viral inhibition of host interferon and apoptosis responses and thus has the potential for increased immunogenicity, since these responses can have adjuvant effects (44). In addition, the C^Δ170 mutation does not confer temperature sensitivity and may thus allow some replication in the lungs that might increase immunogenicity. In contrast, the L^Y942A mutation is a strong temperature sensitivity-conferring mutation, which is thought to disproportionately restrict replication in the lower (warmer) respiratory tract and may thus reduce the possibility of lower respiratory tract reactogenicity.

As a third variable, we evaluated three different insertion sites, namely, the first, second, and third gene positions (F1, F2, and F3, respectively). Placement of a gene closer to the promoter in a nonsegmented negative-strand RNA virus typically increases its rate of transcription, and this can further influence attenuation in at least two ways: (i) increased expression of the RSV F protein could interfere with vector replication due to syncytium formation and other effects, and (ii) the presence of an added gene can reduce the transcription of downstream vector genes, with promoter-proximal inserts affecting a greater number of vector genes. In addition, we have found that, for reasons that are not clear or predictable, a particular construct can occasionally have an unexpectedly high degree of attenuation, as was observed with the rB/HPIV3-F3 construct in a previous study (21) and the rHPIV1-L^Y942A-F2 construct in the present study.

The vectors were constructed using an F ORF sequence that had previously been codon optimized for expression in human cells (34). The sequence of the expressed F protein was also designed previously (34) to be identical at the amino acid level to the sequence of the earliest available passage of strain A2, called HEK-7, because of the human embryonic kidney cell line used at that time. As previously discussed (34), the encoded HEK F protein has a hypofusogenic phenotype that may represent the fusion phenotype of the native A2 clinical isolate (34).

All six rHPIV1-RSV-F constructs with the exception of rHPIV1-L^Y942A-F2 were readily recovered. Nine independent transfections were required to recover a clone of the rHPIV1-L^Y942A-F2 construct that was free of adventitious mutations. In five of the eight mutated clones, the adventitious mutations were ones that likely reduce or ablate expression of the RSV F insert and were reminiscent of ones described previously in the rB/HPIV3-F2 virus from the nasal washes of vaccinated children (20). The other three mutant clones had mutations in the N and/or L protein whose effects are unknown. Since the RSV F-gene insert and the insertion site of the rHPIV1-L^Y942A-F2 virus were identical to those of the rHPIV1-C^Δ170-F2 virus this suggests that insertion into the second position was not sufficient alone to achieve these effects of inefficient recovery and the frequent occurrence of adventitious mutations. In addition, the observation that these effects were observed with the rHPIV1-L^Y942A-F2 virus but not the F1 and F3 derivatives suggests that the L^Y942A mutation alone was also not sufficient to achieve these effects; thus, they appeared to be due to the combination of the F2 gene position and the L^Y942A mutation.

All of the rHPIV1-RSV-F vectors, including rHPIV1-L^Y942A-F2, replicated in Vero and LLC-MK2 cells to final titers that were statistically indistinguishable from those of wt HPIV1 (Fig. 2). Some reductions in replication kinetics were observed during exponential-phase replication (e.g., day 2). For example, the L^Y942A mutation generally reduced the growth rate more than the C^Δ170 mutation, especially in Vero cells. Also, the attenuating effect of the RSV F insert was generally greater in the L^Y942A backbone, suggesting that the RSV F insert has a greater effect on a more attenuated backbone. There was some evidence of a greater reduction in the growth rate in association with promoter-proximal inserts: for example, among the rHPIV1-C^Δ170-RSV-F viruses, a greater reduction was observed for the F1 and F2 viruses in Vero cells and for the F1 virus in LLC-MK2 cells, whereas the growth rate of the F3 virus was similar to that of the empty vector and wt HPIV1. However, since these various differences were not reflected in the final titers, all of these constructs would be amenable to vaccine manufacture.

Analysis of intracellular protein expression in Vero cells using Western blotting showed that the rHPIV1-C^Δ170-F1, -F2, and -F3 constructs expressed high levels of RSV F protein with little evidence of a polar gradient. With the rHPIV1-L^Y942A vectors expressing RSV F, the expression of the RSV F protein by the F1 construct was high and comparable to that by the rHPIV1-C^Δ170-RSV-F constructs, whereas the expression by the F2 and F3 constructs was greatly reduced. With regard to the effect of the RSV F insert on the expression of vector proteins in the rHPIV1-C^Δ170 and rHPIV1-L^Y942A backbones, the results were variable. In two cases (the rHPIV1-C^Δ170-F1 and -F2 viruses), the expression of those genes that were downstream of the RSV F insert were reduced compared to those in the empty vector, consistent with the expected effect on the transcriptional gradient. In one case (the rHPIV1-L^Y942A-F2 virus), the expression of both upstream and downstream genes were reduced, an effect that might be related to the debilitated nature of this construct exemplified by the reduced efficiency of recovery and the frequent occurrence of adventitious mutations. In the other three cases (the rHPIV1-C^Δ170-F3 and rHPIV1-L^Y942A-F1 and -F3 viruses), there was no significant reduction in the expression of the vector genes compared to that in the respective empty vectors.

The insertion of RSV F into either vector backbone increased the level of temperature sensitivity in a number of cases. Virus with the C^Δ170 backbone was marginally temperature sensitive, and this was increased for the F1 derivative but not for F2 or F3. The L^Y942A construct was substantially more temperature sensitive, and this was further increased for the F1 and F2 derivatives but not for F3. The basis for the increases in temperature sensitivity is unclear: it may be related to reductions in vector protein synthesis. As noted above, the temperature sensitivity of a vaccine virus would be expected to reduce reactogenicity in the lower respiratory tract.

The rHPIV1 constructs were evaluated for replication and immunogenicity in hamsters. The rHPIV1-C^Δ170 empty vector was more attenuated than wt HPIV1 in the nasal turbinates and, to a greater degree, in the lungs. Compared to the rHPIV1-C^Δ170 empty vector, the F2 derivative was more attenuated in the nasal turbinates, and all three derivatives (F1, F2, and F3) were more attenuated in the lungs, with no replication being detectable in the lungs of most of the animals. The rHPIV1-L^Y942A empty vector was much more attenuated than wt HPIV1 in both the nasal turbinates and lungs on day 3, with no virus being detectable in the nasal turbinates of 3/6 of the animals or in the lungs of any of the animals; no replication in either compartment was detected on day 5. The rHPIV1-L^Y942A-F1 and -F3 derivatives were further attenuated in the nasal turbinates, whereas the general lack of replication in the lungs made it impossible to assess further attenuation in that compartment. Curiously, the F2 derivative replicated in the nasal turbinates somewhat more efficiently than the empty vector, which was surprising, given the somewhat debilitated nature of this virus in other assays. The rHPIV1-C^R84G/Δ170HN^553AL^Y942A vaccine candidate, which was included as a control, did not replicate detectably in the nasal turbinates or lungs of any animal. Since this virus is overattenuated in seronegative infants and children (43), an appropriate rHPIV1-RSV-F candidate should be less attenuated, which was the case with the rHPIV1-C^Δ170-F1, -F2, and -F3 constructs.

In the immunized hamsters, RSV-specific serum NAb responses were detected for the rHPIV1-C^Δ170-F1, -F2, and -F3 constructs but not for any of the rHPIV1-L^Y942A-based constructs. The titer of NAbs induced by the rHPIV1-C^Δ170-F1 construct was the highest, followed in order by the F3 construct and the F2 construct. The titers of NAbs correlated with the magnitude of in vivo replication rather than the amount of intracellular synthesis of the RSV F protein measured in vitro. These titers were significantly lower than those induced by rB/HPIV3-F2 or wt RSV. However, these controls proved to not be ideal. In the case of the rB/HPIV3-F2 control, this virus replicated much more efficiently in the nasal turbinates and lungs than any of the rHPIV1-RSV-F constructs, and therefore, it is not surprising that it was substantially more immunogenic. However, since wt HPIV1 did not replicate much more efficiently than the rB/HPIV3-F2 construct, even though the latter is attenuated and the former is not, it appears that hamsters are more permissive for the rB/HPIV3 backbone than the HPIV1 backbone. This would not be the case in humans, since the rB/HPIV3 backbone has been shown to be attenuated and well tolerated, whereas wt HPIV1 is not attenuated, and therefore, we anticipate that in humans the HPIV1-based constructs will have immunogenicity greater than that suggested by the hamster model. wt RSV was also used as a control for immunogenicity, but it was also not ideal because it replicates even more efficiently than rB/HPIV3-F2 (34) and expresses two RSV neutralization antigens, whereas only F is expressed by the HPIV vectors. Therefore, the immunogenicity of the rHPIV1-C^Δ170-RSV-F constructs is probably more promising than might be suggested by the present comparison to rB/HPIV3-F2 and wt RSV.

Protective efficacy was evaluated by challenge with wt RSV. Protection was statistically significant for rHPIV1-C^Δ170-F1 in both the nasal turbinates and lungs and for rHPIV1-C^Δ170-F3 in the nasal turbinates only. Protection was not significant for the rHPIV1-C^Δ170-F2 construct, despite its ability to induce a moderate titer of RSV-neutralizing serum antibodies. Importantly, the protection conferred by the rHPIV1-C^Δ170-F1 construct was statistically indistinguishable from that of the rB/HPIV3-F2 control. This was somewhat unexpected, because rHPIV1-C^Δ170-F1 induced significantly lower RSV NAb titers than rB/HPIV3-F2. Consistent with their lack of immunogenicity, the rHPIV1-L^Y942A-F1, -F2, and -F3 constructs did not provide detectable protection.

Previous clinical evaluation of the rB/HPIV3-F2 construct showed that the stability of RSV F expression is an important consideration (20). In that study, ∼2.5% of the virus in the clinical trial material used for human administration did not express RSV F, and approximately half of the nasal wash specimens from vaccine recipients contained virus with mutations expected to reduce or ablate expression of the RSV F insert (20). This suggests that in vitro and in vivo there was selection of variants in which expression of the RSV F insert had been silenced. In the present study, we evaluated the stability of RSV F expression by coimmunostaining for RSV F and HPIV1 antigens in a double-staining immunofluorescence plaque assay. This was done for our virus stocks and also for virus recovered from tissue homogenates of infected hamsters. This showed that expression of the RSV F protein by the rHPIV1 vectors was quite stable, with close to 100% of the plaques expressing the RSV F antigen. This suggests that there was not a strong selective pressure in vitro or in vivo for variants in which expression of the RSV F protein had been silenced. However, this will warrant careful monitoring in future studies.

In summary, this study evaluated a number of variables and identified the rHPIV1-C^Δ170 backbone as a promising vector for expressing the RSV F protein to develop an attenuated bivalent RSV/HPIV1 vaccine candidate. The F1, F2, and F3 sites appeared to be suitable in this backbone, with the F1 (pre-N) site appearing to be the most immunogenic and protective. This construct has several desirable features: (i) the C^Δ170 mutation has been stabilized against deattenuation; (ii) this mutation has been characterized, has already been used in a candidate evaluated in humans, and has the potential to enhance immunogenicity; (iii) this candidate replicated in Vero cells to final titers similar to that of wt HPIV1, an essential feature for vaccine manufacture; (iv) it was more attenuated than the rB/HPIV3-F2 construct, and its level of protective efficacy was statistically indistinguishable from that of the rB/HPIV3-F2 construct; (v) the construct was stable for RSV F-protein expression after in vitro and in vivo replication; and (vi) it was the most immunogenic vector inducing the highest titers of RSV and HPIV1 NAbs and was also the most protective against a wt RSV challenge. The rHPIV1-C^Δ170-F1 candidate is suitable for further evaluation in nonhuman primates and could also be modified to express a stabilized prefusion form of the RSV F protein (45) that might induce an immune response with an increased quantity and quality of RSV NAbs. An rHPIV1-vectored pediatric RSV vaccine could be used either as a primary RSV vaccine or to boost immunity induced by a previous live attenuated RSV vaccine.