INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Bovine respiratory syncytial virus (BRSV) and Human respiratory syncytial virus (HRSV) are closely related members of the Pneumovirus genus within the subfamily Pneumovirinae, family Paramyxoviridae, order Mononegavirales (6, 36). BRSV is a major cause of respiratory tract disease in cattle (32, 44). HRSV is the most important causative agent of viral pediatric respiratory disease worldwide (6). A licensed vaccine against HRSV is not available, though several promising candidates for attenuated live vaccines recently have been developed (references 17, 46, 47, and 48, and references therein).

The genomes of HRSV and of BRSV are single-stranded, negative-sense RNAs of 15,222 nucleotides (nt) (HRSV A2) and 15,140 nt (BRSV ATue51908) (1, 4). Their genome organizations are identical, comprising 10 genes (encoding 10 mRNAs containing 11 translational open reading frames [ORFs]) in the order 3'-NS1-NS2-N-P-M-SH-G-F-M2-1/M2-2-L-5'. Like all members of Mononegavirales, the respiratory syncytial virus (RSV) genomic RNA is tightly encapsidated by the nucleoprotein (N) and is associated with the phosphoprotein (P) and the viral RNA-dependent RNA polymerase (L). This encapsidated RNA is the template for replication and transcription (13, 49). The viral polymerase enters the genome in the 3' leader region and synthesizes the 10 monocistronic mRNAs by linear sequential start-stop transcription. This is directed by the respective gene start signals, which are located on the upstream end of each gene and are highly conserved between BRSV and HRSV, and the gene end/polyadenylation signals, which are located at the downstream end of each gene and are semiconserved (21, 22, 50). RNA replication yields a full-length, positive-sense intermediate, the antigenome, which serves as template for the synthesis of full-length genomic RNA. The N, P, and L proteins are necessary and sufficient to direct RNA replication, whereas efficient synthesis of full-length mRNAs and efficient sequential transcription also require the M2-1 transcription antitermination factor (5, 9, 13, 14, 49).

In addition, RSV encodes four envelope-associated structural proteins. One of these is the nonglycosylated matrix (M) protein, which is essential for virion formation (42) and is thought to mediate interaction between the nucleocapsid and the envelope. The other three are transmembrane surface glycoproteins: the attachment glycoprotein G (25), the fusion glycoprotein F (45), and the small hydrophobic protein SH (2). For both BRSV and HRSV, the G and F proteins are the major neutralization and protective antigens (7, 31, 41). The G protein is highly variable between the HRSV subgroups (16) and between HRSV and BRSV (24), with a conserved central structure (16) that might be involved in receptor binding. Between the BRSV strain ATue51908 and the HRSV A2 used in this work, the overall amino acid identity of the G protein is 28%. The F protein mediates viral penetration and syncytium formation (45). F is synthesized as an F₀ precursor which is cleaved endoproteolytically into the F1 and F2 subunits that remain linked by disulfide bonds. BRSV ATue51908 and HRSV A2 F proteins are 82% identical.

Live vaccinia virus has the distinction of being both the first vaccine and the most successful, having led to the eradication of smallpox as a human disease. The original vaccinia virus introduced by Jenner in 1798 was of bovine origin. The use of an animal virus as a live vaccine against its human counterpart is called the "Jennerian" approach. This strategy requires that the human virus of interest has an animal counterpart which is attenuated in the nonnatural human host, yet grows sufficiently well and is sufficiently related antigenically that it can induce effective protection against the human virus. As a current example, bovine parainfluenzavirus (BPIV) type 3 is under clinical evaluation as vaccine against human parainfluenzavirus (HPIV) type 3 and appears to have appropriate levels of attenuation and immunogenicity (19). The recently licensed quadrivalent rotavirus vaccine exemplifies both the Jennerian approach and a "modified Jennerian" approach (33). This vaccine is based on a rhesus rotavirus that is attenuated in humans and has sufficient antigenic relatedness to the human rotavirus serotype 3 that it can confer resistance in humans to this human rotavirus serotype. Effective immunoprophylaxis against human rotavirus disease also requires resistance to the three other major serotypes of human rotavirus. Therefore, the rhesus rotavirus was used to construct, by the natural mechanism of gene reassortment during mixed infection in cell culture, three single-gene reassortant viruses. In each, the rhesus rotavirus gene encoding VP7, a major neutralization antigen, was replaced by its counterpart from human rotavirus serotype 1, 2, or 4. The resulting three chimeric reassortant viruses were combined with the rhesus rotavirus parent to formulate the quadrivalent vaccine. Chimeric animal/human viruses represent the modified Jennerian strategy.

BRSV and HRSV are related antigenically (11, 32), and BRSV has been considered as a possible Jennerian vaccine against HRSV (34, 35). In previous work (34, 35), BRSV was evaluated in several experimental animal models, most notably in chimpanzee, which is the animal that is the most permissive for HRSV replication and disease. When BRSV was administered intranasally to two animals, replication was detected in only one animal, and only at a very low level, and the animals were not significantly resistant to subsequent HRSV challenge (35) (R. M. Chanock, personal communication).

Recently, we reported the cDNA cloning and sequencing of the entire BRSV genome and the establishment of a reverse genetics system allowing recovery of infectious recombinant BRSV (rBRSV) (1). In the present work, we reevaluated the Jennerian approach by assaying the replication of rBRSV in chimpanzees and its protective efficacy against HRSV. Furthermore, we made use of this system to generate chimeric rBRSV in which the glycoprotein F gene alone, or F and G genes together, were replaced by their HRSV counterpart(s), thus combining the replication features of BRSV and the antigenic properties of HRSV. This represents the first step in developing a chimeric BRSV/HRSV virus as a vaccine against HRSV, and this modified Jennerian vaccine candidate also was evaluated in chimpanzees. The construction of bovine/human chimeric viruses also addresses the fundamental issue of the contribution of the G and F glycoproteins in determining the host range of RSV.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of cDNAs encoding chimeric BRSV/HRSV antigenomes. We previously constructed a cDNA encoding BRSV antigenomic RNA (1). This was modified by site-directed mutagenesis (20) in two ways: first, a unique synthetic NotI site in the NS1 noncoding region was removed to regenerate the ATue51908 wild-type sequence (GenBank accession no. AF 092942). Second, synthetic restriction sites were introduced into the SH/G (SalI, ATue51908 nt 4673), G/F (SphI, ATue51908 nt 5539), and F/M2 (XhoI, ATue51908 nt 7471 and ClaI, ATue51908 nt 7485) intergenic regions (Fig. 1B). The resulting plasmid was termed pBRSV18. This cDNA served as parent for the construction of BRSV/HRSV chimeras.

View larger version (55K): [in this window] [in a new window]   FIG. 1.   Construction of chimeric rBRSV in which the F gene alone was replaced by its counterpart from HRSV Long to create rBRSV/LongF or in which the F and G genes were both replaced by their counterparts from HRSV A2 to create rBRSV/A2. (A) Diagram of the rBRSV genome, illustrating the replacement of the F and G genes with those of HRSV. The location of the ORFs are shown as shaded (BRSV) or open (HRSV) rectangles. The G and F genes are shown as enlargements, in which gene end/polyadenylation signals are represented by bars, gene start signals are shown as filled triangles, and the locations of synthetic restriction sites are indicated by filled arrowheads with the position in the complete antigenome sequence shown underneath. Naturally occurring PstI and EcoRI restriction sites used for differentiation of PCR products obtained from recombinant viruses are indicated by open arrowheads. The lengths of the fragments framed by the marker restriction sites are indicated in parentheses. The genome length of each recombinant virus is indicated on the left in parentheses. The roman numerals (I, II, and III) refer to intergenic regions which are expanded in panel B. (B) Alignment of the SH/G (I), G/F (II), and F/M2 (III) intergenic regions of biologically derived BRSV strain ATue51908 (top line), its recombinant version rBRSV (second line), the chimeric virus rBRSV/LongF (third line), the recombinant chimeric virus rBRSV/A2 (fourth line), and rHRSV strain A2 (bottom line). The sequences are shown as DNA-positive strands. Gaps are indicated by dots, gene start and gene end signals are shaded, and translation start codons are in boldface. Restriction sites are indicated, with recognition sequences shaded. Numbering refers to the antigenomic positions (for BRSV, ATue51908, GenBank accession no. AF092942; for HRSV strain A2, GenBank accession no. M74568, further modified as described in reference 4).

Recovery of recombinant RSVs. Recombinant RSVs were recovered from cDNA essentially as described before (1). Dishes (diameter, 32 mm) of subconfluent BSR T7/5 cells stably expressing phage T7 RNA polymerase were transfected with 5 µg of the respective antigenomic plasmid and a set of four support plasmids (2.5 µg of pN, 2.5 µg of pP, 0.5 µg of pL, and 0.5 µg of pM2) from which the BRSV N, P, L, and M2-1 proteins are expressed. All cDNA constructs were under control of a T7 promoter (1). Transfections were carried out by using Superfect (Qiagen). Two hours after transfection, the supernatant was removed, and cells were washed and maintained in minimum essential medium supplemented with 3% fetal calf serum (FCS). Three days after transfection, the cells were split in a 1 to 3 ratio. Cells and supernatant were harvested 7 days after transfection.

Viruses and cell culture. Recombinant BRSV and BRSV/HRSV chimeric viruses were propagated in monolayer cultures of MDBK cells. Cells were infected at a multiplicity of infection (MOI) of 0.1 and were maintained at 37°C in minimum essential medium supplemented with 3% FCS. When the cytopathic effect (CPE) was pronounced, the medium was adjusted to 100 mM MgSO₄ and to 50 mM HEPES (pH 7.5) (10), and cells and medium were collected and subjected to three rounds of freezing and thawing, and the clarified medium supernatants were stored at 70°C and titered by limiting dilution method on BSR T7/5 cells (1). HRSV Long (kindly provided by G. Herrler, Hannover, Germany) was propagated on HEp-2 cells by the same general procedure. HRSV Long is very similar to strain A2 with regard to sequence analysis (16, 26), with 98% amino acid sequence identity for the F protein, and also is closely related based on reactivity with most monoclonal antibodies (MAbs) (12). The Long strain was used as the wild-type HRSV control for the in vitro studies.

RT-PCR. Total RNA of MDBK cells infected individually with each RSV was prepared (RNeasy, Qiagen) when extensive CPE was observed. To verify the identity of the recombinant viruses, regions containing the marker restriction sites shown in Fig. 1B were amplified by RT-PCR (1).

View larger version (37K): [in this window] [in a new window]   FIG. 2.   Demonstration of marker restriction sites in the genomes of biologically derived BRSV ATue51908, rBRSV, the chimeras rBRSV/A2 and rBRSV/LongF, and HRSV Long. RT-PCR was performed on infected-cell RNA, and the PCR products were subjected to restriction analysis and separated on a 2% (panels A and B), 3% (panel C), or 0.8% (panel D) agarose gel. M, 1-kb marker DNA ladder (Life Technologies) with the size of some marker fragments indicated. (A) Analysis of RT-PCR products encompassing the SH/G intergenic region and demonstration of the presence of a SalI site in rBRSV and rBRSV/A2 and its absence in biologically derived BRSV ATue51908. The RT-PCR products were consistent with the predicted size of approximately 1,274 bp, and the SalI digestion products were consistent with the predicted sizes of approximately 1,062 and 212 bp. (B) Analysis of RT-PCR products encompassing the G/F intergenic region and demonstration of the presence of an SphI site in rBRSV and a StuI site in rBRSV/A2. The PCR products were approximately 592 bp (ATue51908 and rBRSV) or 614 bp (rBRSV/A2). SphI digested only rBRSV, resulting in fragments of 425 and 167 bp. StuI digested only rBRSV/A2, resulting in fragments of 443 and 171 bp. (C) Analysis of RT-PCR products encompassing the F/M2 intergenic region, and demonstration of an XhoI site in rBRSV and rBRSV/A2 and its absence in BRSV ATue51908. The RT-PCR product was 637 (rBRSV/A2) or 648 bp (ATue51908 and rBRSV). XhoI digestion resulted in fragments of 381 (rBRSV and rBRSV/A2), 267 (rBRSV), or 256 bp (rBRSV/A2). (The last two bands each contain 14 additional nonviral nucleotides contributed by the oligonucleotide primers.) (D) Analysis of RT-PCR products confirming the identity of rBRSV/LongF. One series of reactions (RT-PCR1) was carried out using a BRSV-specific primer pair hybridizing upstream and downstream of the F gene. This yielded products of 2,187 bp (BRSV strain ATue51908 and rBRSV) or 2,161 bp (rBRSV/LongF), whereas the BRSV-specific primers did not produce a product from HRSV Long. SphI cleaved the products representing rBRSV and rBRSV/A2, yielding bands of 2,028 and 159 bp (rBRSV) or 2,002 and 159 bp (BRSV/LongF). BRSV strain ATue51908 was not cleaved. EcoRI cleaved BRSV ATue51908 and rBRSV at a naturally occurring site in the BRSV F gene to yield two fragments of 1,334 and 853 bp, whereas the fragment originating from rBRSV/LongF was not cleaved by EcoRI. PstI cleaved rBRSV/LongF at two naturally occurring sites in the Long F gene to yield fragments of 1,351, 586, and 224 bp, whereas rBRSV F gene in BRSV ATue51908 and rBRSV remained uncleaved. A second series of reactions (RT-PCR2) was performed using a primer pair that hybridized at either end of the HRSV Long F gene, yielding a PCR product of 1,904 bp for rBRSV/LongF and the HRSV Long parental virus. As expected, a product was not obtained from rBRSV.

Indirect immunofluorescence assay. HEp-2 cells were infected with an MOI of 0.1 with rBRSV, rBRSV/A2, rBRSV/LongF, or HRSV Long, were incubated for 36 h, were fixed with 80% acetone, and were incubated at 37°C for 30 min with MAb G66 directed to BRSV G (1:1,000 dilution), kindly provided by G. Taylor (11, 23), or MAb 021/1G directed against HRSV G (1:40 dilution), MAb 2F reacting with both HRSV and BRSV F (1:40 dilution), or MAb 44F, specific to HRSV F (1:40 dilution), kindly provided by J. A. Melero (12, 27). Cells were stained with a fluorescein-isothiocyanate-conjugated goat anti-mouse immunoglobulin G (Dianova, Hamburg, Germany) and were counterstained with 0.01% Evans Blue.

Growth analysis. Subconfluent MDBK cells or HEp-2 cells were infected with rBRSV, rBRSV/A2, rBRSV/LongF, or HRSV strain Long at an MOI of 0.1. After 90 min of adsorption, the inoculum was removed and the cells were washed twice with medium containing 3% FCS and were incubated at 37°C. At various times after infection, the medium was adjusted to 100 mM MgSO₄ and 50 mM HEPES (pH 7.5) (10), and the cells and medium were quickly frozen and thawed three times. The medium was clarified by centrifugation and was stored at 70°C and analyzed later in duplicate by plaque assay.

Electron microscopy. For immunolabeling experiments, MDBK cells were infected with rBRSV, rBRSV/A2, rBRSV/LongF, or HRSV Long at an MOI of 0.1. At 72 h postinfection, when CPE was pronounced, the material was scraped off the plate and was collected by low-speed centrifugation, and the cell pellet was resuspended in a small amount of phosphate-buffered saline (PBS). Glow-discharged formvar-coated 300-mesh copper grids, stabilized with carbon, were floated on drops of cell suspension. Nonspecific adsorption of antibodies was blocked by treating the grids with 1% cold water fish gelatin (Sigma, Deisenhofen, Germany) in PBS containing 1% bovine serum albumin, fraction V (PBS-BSA). Subsequently, the grids were floated for 45 min on drops of PBS-BSA which each contained one of the four MAbs described above. After several washings in PBS-BSA, bound antibodies were detected with 10-nm-colloidal-gold-labeled goat anti-mouse immunoglobulin Fab (GAF₁₀; BioCell Int., Cardiff, United Kingdom) and negatively stained with phosphotungstic acid, pH 6.0. The grids were examined using a transmission electron microscope (EM 400T; Philips).

Evaluation of replication and protective efficacy in chimpanzees. Young chimpanzees were confirmed to be seronegative for RSV by enzyme-linked immunosorbent assay against HRSV F glycoprotein. Animals were inoculated by both the intranasal and intratracheal routes with a dose of 10⁷ PFU per ml per site of rBRSV or rBRSV/A2 (see Table 2). Each virus was administered to two chimpanzees. Following inoculation, nasopharyngeal swab samples were taken daily on days 1 through 10 and 12, and tracheal lavage samples were taken on days 2, 5, 6, 8, and 12. Specimens were frozen, and RSV titers were measured later by plaque assay on HEp-2 cells. The amount of rhinorrhea, a measure of upper respiratory tract illness, was estimated daily and was assigned a score of 0 to 4 (0 = none, 1 = trace, 2 = mild, 3 = moderate, 4 = severe). The results were compared (see Table 2) to historic controls of animals which had been inoculated in the same way with 10⁴ PFU of HRSV A2 wild-type virus per site (47) or inoculated with 10⁵ PFU of the live attenuated cpts248/404 strain A2 vaccine candidate per site (46).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of cDNAs encoding BRSV/HRSV chimeric antigenomes. We previously described a system for generating rBRSV entirely from cDNA (1). The plasmid encoding the full-length BRSV antigenomic RNA (1) was modified to remove a synthetic NotI marker and thereby restore the wild-type sequence in the NS1 noncoding region. The antigenomic cDNA was further modified by insertion of synthetic SalI, SphI, XhoI, and ClaI restriction sites into the SH/G, G/F, and F/M2 intergenic regions (Fig. 1), resulting in plasmid pBRSV18. The length of the antigenome was unchanged. rBRSV recovered from this plasmid was indistinguishable from wild-type BRSV ATue51908 by in vitro growth characteristics (data not shown).

Recovery and identification of chimeric rBRSV bearing HRSV glycoprotein genes. Plasmids pBRSV/A2-10 and pBRSV/LongF-12, encoding chimeric BRSV/HRSV antigenomic RNAs, were transfected separately into BSR T7/5 cells, which stably express T7 RNA polymerase. A set of expression plasmids for the BRSV N, P, L, and M2-1 proteins was cotransfected. In parallel, transfections were done with pBRSV18 to generate parental rBRSV as a control. Five days after transfection, CPE consisting of rounding cells typical for RSV could be observed in all transfected dishes. The recovery rates of rBRSV, rBRSV/A2, and rBRSV/LongF were comparably high. Depending on the experiment, typical yields were approximately 30 to 300 foci per 32-mm-diameter dish as determined from duplicate monolayers which were overlaid with methylcellulose 2 h after transfection, were fixed 96 h posttransfection, and were analyzed by indirect immunofluorescence staining using a MAb directed to RSV F. Virus stocks were produced by two passages of the supernatants from transfections on MDBK cells.

Expression of the HRSV glycoproteins in cells infected with rBRSV/A2. HEp-2 cells were infected with rBRSV, rBRSV/A2, or HRSV Long at an MOI of 0.1. Thirty-six hours postinfection, cells were fixed and stained with MAbs specific for the BRSV and HRSV G glycoproteins (Table 1 and Fig. 3) (see Materials and Methods). As expected, the BRSV G-specific MAb G66 reacted with rBRSV but not with rBRSV/A2 or HRSV. In contrast, the HRSV G-specific MAb 021/1G reacted with rBRSV/A2 and HRSV, but not with rBRSV. This confirmed that rBRSV/A2 directed the abundant expression of HRSV-specific G protein in infected cells and did not express BRSV-specific G protein.

View larger version (123K): [in this window] [in a new window]   FIG. 3.   Indirect immunofluorescence of HEp-2 cells infected with rBRSV (top panels), rBRSV/A2 (middle panels), or HRSV Long (bottom panels). Cells were infected at an MOI of 0.1 and were processed 36 h later for indirect immunofluorescence. The panels illustrate cells that were reacted with (left) MAb G66, specific to BRSV G protein, or (right) MAb O21/1G, specific to the HRSV G protein. The CPE produced by rBRSV and rBRSV/A2 in HEp-2 cells is comparable and not very pronounced. In contrast, HRSV produces an extensive CPE with large syncytia.

Growth of rBRSV/A2 in cell culture. Multicycle growth of rBRSV, rBRSV/A2, and HRSV Long was compared in MDBK and HEp-2 cells which were infected at an MOI of 0.1. At various times postinfection, duplicate wells were harvested, and the virus in clarified medium supernatants was titrated (Fig. 4). In HEp-2 cells, HRSV produced approximately 10-fold more virus than did rBRSV (Fig. 4A), whereas the situation was reversed in MDBK cells (Fig. 4B). In MDBK cells, HRSV reached a maximum titer after 2 days, whereas rBRSV continued to grow and reached a maximum after 4 to 5 days. These observations identify a partial restriction of growth for BRSV and HRSV in human and bovine cells, respectively. The replication of the chimeric rBRSV/A2 virus in HEp-2 cells was intermediate between that of HRSV and BRSV, consistent with its chimeric nature. Thus, transfer of the human glycoproteins improved the growth of the chimera in human cells, but the effect was not complete, and the peak titer observed for the chimera more closely resembled that of its BRSV parent than that of HRSV. In bovine cells, the growth of the chimera more closely resembled that of BRSV although, interestingly, the chimeric virus grew somewhat better than its BRSV parent. The ability to replicate efficiently in this multicycle growth experiment indicated that the HRSV G and F proteins were functional in the rBRSV background and that the chimeric virus is fully competent for growth in two cell lines, one of human origin and one of bovine origin. The growth characteristics of rBRSV/LongF resembled those of rBRSV (data not shown), indicating that this chimera was fully competent for multicycle growth in both cell lines even though the G and F glycoproteins were of BRSV and HRSV, respectively.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Comparison of the multicycle growth of rBRSV, rBRSV/A2, and HRSV Long in human HEp-2 cells (A) and bovine MDBK cells (B). Duplicate cell monolayers in 24-well dishes were infected with the indicated virus at an MOI of 0.1. Monolayers were harvested at the indicated times, were stored at 70°C, and were titrated later in duplicate. Each value is the mean titer of material from two wells of infected cells.

Electron microscopy of the rBRSV/A2 and rBRSV/LongF chimeric viruses. Virions of rBRSV/A2, rBRSV/LongF, rBRSV, and HRSV Long grown in MDBK cells were examined by immunoelectron microscopy using antibodies against the F and G proteins (Table 1), followed by labeling with gold-conjugated goat anti-mouse immunoglobulin and negative staining. The detected virions were filamentous with a diameter of 100 to 200 nm with highly variable lengths up to several micrometers. Some virions seemed to be branched. No spherical virion particles were detected (not shown).

View larger version (168K): [in this window] [in a new window]   FIG. 5.   Immunogold labeling using MAbs reacting with both the HRSV F and BRSV F protein (MAb 2F; A, C, E) or exclusively with the HRSV F protein (MAb 44F; B, D, F). Labeling was performed on preparations of rBRSV (A, B), rBRSV/A2 (C, D), and HRSV Long (E, F). Bar, 150 nm.

View larger version (180K): [in this window] [in a new window]   FIG. 6.   Immunogold labeling using MAbs reacting with the BRSV G protein (MAb G66; A, C, E) or with the HRSV G protein (MAb 021/1G; B, D, F). Labeling was performed on preparations of rBRSV (A, B), rBRSV/A2 (C, D), and HRSV Long (E, F). Bar, 150 nm.

View larger version (104K): [in this window] [in a new window]   FIG. 7.   Immunogold labeling of the surface proteins of rBRSV/LongF. Labeling was performed with the following MAbs: MAb G66 specific to the BRSV G protein (A), MAb 021/1G specific to the HRSV G protein (B), MAb 2F specific to the F proteins of both HRSV and BRSV (C), MAb 44F specific to HRSV F protein (D). Bar, 150 nm.

Evaluation of the replication and protective efficacy of the rBRSV/A2 chimeric virus in chimpanzees. rBRSV and rBRSV/A2 were each administered to two chimpanzees by both the intranasal and intratracheal routes at the high dose of 10⁷ PFU per ml per site (Table 2). Virus replication in the upper and lower respiratory tracts was monitored by nasopharyngeal swab and tracheal lavage, respectively. The results were compared (Table 2) to historic controls of animals which had been inoculated in the same way with 10⁴ PFU of HRSV A2 wild-type virus per site (47) or 10⁵ PFU of the live attenuated cpts248/404 strain A2 vaccine candidate per site (46).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Previously, we described reverse genetics systems for generating rHRSV (4) or rBRSV (1) from cDNA. In the work described here, these systems were used to generate rBRSVs in which the F gene alone, or the F and G genes together, were replaced by their HRSV counterparts. The heterologous F glycoprotein alone, or F and G glycoproteins together, were efficiently incorporated into the virion envelope based on immunoelectron microscopy and appeared to be fully functional based on efficient multicycle growth of the chimeras in vitro. The chimeric virus rBRSV/A2, which was studied in greater detail because it bears the greater number of the HRSV surface proteins, more closely resembled BRSV than HRSV with respect to host range in cell culture and in chimpanzees. These data indicated that the G and F proteins contribute to the host range restriction of BRSV but, surprisingly, are not its major determinants. This study also evaluated Jennerian and modified Jennerian strategies for employing BRSV for the development of a live-attenuated vaccine against HRSV.

One potential obstacle to replacing the F and G proteins of BRSV with those of HRSV was that the proteins might be incompatible. By analogy to other mononegaviruses, it seemed likely that interaction between the internal domain of one or more glycoproteins and internal proteins such as M would be important in virion assembly and infectivity (3, 28, 29, 38). However, recent observations indicated that the G and SH proteins were unlikely to play critical roles, at least in the appropriate cell line. For example, RSV B1 multiply passaged at low temperature was recently shown to have sustained the spontaneous deletion of the SH and G genes, and this increased rather than decreased virus growth at low temperature in cell culture (18). Studies with rHRSV and rBRSV confirmed that the SH and G genes can be deleted without loss of growth fitness in cell culture (2) (M. Teng and P. L. Collins, unpublished data and U. J. Buchholz, unpublished data). In a minigenome system, the production of transmissible particles was strongly dependent on M and F, was less dependent on G, and was independent of SH (42). Taken together, these studies indicate that F and M are critical for virion formation and infectivity in vitro, whereas SH and G are dispensible under the appropriate in vitro conditions. The F proteins of these BRSV and HRSV strains share only 82% amino acid identity, with the cytoplasmic and transmembrane domains being 67 and 79% identical, respectively. This degree of divergence apparently had no effect on the ability of the chimeric virus to grow in vitro. This is not completely surprising, since we previously reported that the hemagglutinin-neuraminidase (HN) and F proteins of HPIV3 could be replaced by their HPIV1 counterparts without significant loss of growth efficiency even though the percentage of amino acid identity between serotypes was very low: HN transmembrane domain identity, 30%; F transmembrane identity, 22%; HN cytoplasmic tail identity, 9%; F cytoplasmic tail identity, 11% (40). It is also noteworthy that the HRSV F and BRSV G proteins could operate together in a single virus, rBRSV/LongF, without a loss in efficiency of incorporation into virions or of replication in cell culture. This was somewhat surprising, because the G proteins of HRSV and BRSV are very divergent, exhibiting 28% identity overall. The ability of nonmatched G and F glycoproteins to function in the same envelope might mean they do not interact closely or, alternatively, that this interaction is not highly sequence specific. This is in contrast to the situation with members of the Paramyxovirinae such as Sendai virus and other viruses of the genus Parainfluenzavirus, in which the processing and function of the HN and F proteins appear to be closely linked with respect to fusion (39, 43).

The chimpanzee is the experimental animal model that most closely resembles humans with regard to permissiveness for HRSV replication and disease. When a high dose of rBRSV was administered to two animals in this study, there was no evidence of virus shedding or disease. In a previous study, one of two chimpanzees inoculated with BRSV shed a low level of virus (35). In contrast, inoculation with a 1,000-fold-lower dose of HRSV resulted in high levels of virus shedding and the induction of disease. Thus, chimpanzees provide an unambiguous and highly authentic assay of the difference in host range between BRSV and HRSV. When a high dose of rBRSV/A2 was administered to two animals, virus shedding was detected in both animals but was very low. The observation that the transfer of the HRSV G and F proteins to BRSV improved its growth somewhat in chimpanzees indicates that these two glycoproteins contribute to the host range phenotype. However, the observation that the chimera remains highly restricted compared to HRSV indicates that one or more other viral proteins also make important contributions to the host range phenotype. It also is possible that cis-acting RNA sequences are involved, but this seems less likely because the transcription signals are highly conserved between HRSV and BRSV as noted above and because the leader region of HRSV can be substituted into BRSV with little effect (1).

These findings also have implications with regard to strategies for improving the permissiveness of convenient small experimental animals such as mice for HRSV infection and disease. One strategy would be to identify the human cellular receptor(s) for HRSV and express it in a transgenic mouse, such as has been done for measles virus (30, 37). However, since the permissiveness of BRSV for replication in chimpanzees was not greatly improved by the insertion of G and F glycoproteins of HRSV origin, it seems unlikely that expression in a mouse of the human molecules which interact with G and perhaps F would have a major effect on improving permissiveness for HRSV.

Reverse genetics is being used to engineer rHRSV to develop candidate live-attenuated vaccines (references 2, 15, 17, 46, 47, and 48, and references therein). In particular, the HRSV reverse genetics system was used to characterize and combine sets of attenuating mutations present in HRSV strains attenuated by classical passage and mutagenesis methods (references 17, 47, and 48, and references therein). Because of the strategy by which they were selected, most of these mutations confer the temperature-sensitive phenotype in addition to the attenuation phenotype. One limitation of the temperature-sensitive phenotype is that the virus can be overly restricted for replication in the lower respiratory tract while being underattenuated in the upper respiratory tract. This is because there is a temperature gradient within the respiratory tract, with temperature being higher (more restrictive) in the lower respiratory tract and lower (less restrictive) in the upper respiratory tract. The ability of an attenuated virus to replicate in the upper respiratory tract can result in complications, including congestion, rhinitis, fever, and otitis media. Thus, attenuation achieved primarily by temperature-sensitive mutations may not be ideal. One solution has been to supplement the temperature-sensitive mutations with deletion mutations, such as deletion of the entire SH or NS2 gene (2, 46). Such mutations are not temperature sensitive and, furthermore, would be genetically more stable than point mutations.

Here, we have pursued an alternative approach to develop a live-attenuated HRSV vaccine, namely to make use of the host range differences between HRSV and BRSV. It would be anticipated that a growth restriction in a nonnatural host would involve multiple genes and many sequence differences, and this is consistent with the present findings that the G and F glycoproteins contribute only part of this phenotype. Not surprisingly, experience with other viruses indicates that a host range restriction is a very stable phenotype (19, 33). Also, host range restrictions typically are not associated with temperature sensitivity. We reexamined the Jennerian approach, in which BRSV (in this case a recombinant version) was used directly as a live-attenuated vaccine against HRSV. This had been investigated previously in rodents, in monkeys, and in two chimpanzees with the B/097 strain (35). Whereas previous studies in rodents or monkeys suggested that BRSV might be a satisfactory Jennerian vaccine, the present results with strain ATue51908 confirm that BRSV itself has very low efficacy in chimpanzees as a vaccine against HRSV. This is likely due in large part to its very poor replication as described above.

We also sought to use reverse genetics to improve BRSV as a live-attenuated vaccine against HRSV, pursuing a modified Jennerian strategy. The first step, described here, involved substituting the G and F glycoproteins of BRSV with those of HRSV. Since G and F are the major protective antigens of RSV, their substitution would address the issue of antigenic divergence and also had the potential for improving the efficiency of replication of BRSV in primates. As noted above, the replication of BRSV in chimpanzees was only marginally improved. Since rBRSV/A2 remains overattenuated, the next step will be to improve its capacity for replication in primates by substituting additional HRSV genes. Specifically, the difference between rBRSV/A2, which is highly restricted in chimpanzees, and HRSV, which is highly permissive, lies in eight genes: NS1, NS2, N, P, M, SH, M2-1/M2-2, and L. The systematic substitution of these eight genes should identify a chimera with an appropriate level of attenuation and will provide new information on the determinants of host range in a paramyxovirus. A complementary strategy will be to start with HRSV and replace internal HRSV genes with their BRSV counterparts in an incremental, stepwise fashion. In conclusion, we have developed a chimeric rBRSV that bears the major antigenic determinants of HRSV and have shown that it is viable and safe in chimpanzees and thus represents a starting point for developing a live-attenuated chimeric vaccine against HRSV.