Newcastle disease virus, a host range-restricted virus, as a vaccine vector for intranasal immunization against emerging pathogens

Abstract

The international outbreak of the severe acute respiratory syndrome-associated coronavirus (SARS-CoV) in 2002–2003 highlighted the need to develop pretested human vaccine vectors that can be used in a rapid response against newly emerging pathogens. We evaluated Newcastle disease virus (NDV), an avian paramyxovirus that is highly attenuated in primates, as a topical respiratory vaccine vector with SARS-CoV as a test pathogen. Complete recombinant NDV was engineered to express the SARS-CoV spike S glycoprotein, the viral neutralization and major protective antigen, from an added transcriptional unit. African green monkeys immunized through the respiratory tract with two doses of the vaccine developed a titer of SARS-CoV-neutralizing antibodies comparable with the robust secondary response observed in animals that have been immunized with a different experimental SARS-CoV vaccine and challenged with SARS-CoV. When animals immunized with NDV expressing S were challenged with a high dose of SARS-CoV, direct viral assay of lung tissues taken by necropsy at the peak of viral replication demonstrated a 236- or 1,102-fold (depending on the NDV vector construct) mean reduction in pulmonary SARS-CoV titer compared with control animals. NDV has the potential for further development as a pretested, highly attenuated, intranasal vector to be available for expedited vaccine development for humans, who generally lack preexisting immunity against NDV.

The severe acute respiratory syndrome-associated coronavirus (SARS-CoV) emerged as a human pathogen in Southern China in 2002 and spread internationally, resulting in 8,096 infections and 774 deaths between November 1, 2002, and July 31, 2003, as well as considerable alarm and economic loss (1). There is a clear need for a safe and effective vaccine should an outbreak of a SARS-like virus reoccur in humans. In addition, the SARS experience highlighted the importance of having pretested, safe, and effective vectors on hand that can be used to accelerate initial steps of vaccine development should other new pathogenic agents emerge.

Inactivated whole-virus preparations of SARS-CoV have been shown to induce protective immunity in small-animal models (2–5) and primates (6, 7). However, large-scale growth of a highly pathogenic agent introduces the risk of laboratory infection, which indeed has occurred (8). Furthermore, incomplete inactivation can leave residual infectivity in such vaccines, posing further risk to vaccinees (9, 10). The SARS-CoV spike S glycoprotein is the viral neutralization and major protective antigen (11). Full-length or fragments of recombinant purified S have been shown to be immunogenic in small-animal models, but their safety and efficacy have not yet been evaluated in primates. Furthermore, inactivated whole-virus or protein subunit vaccines are not necessarily safe and effective: in the case of measles and respiratory syncytial virus, they induced altered immune responses, resulting in immune-mediated enhancement of disease upon subsequent exposure to the pathogen (12, 13).

Vectored vaccines offer a live-vaccine approach that does not involve the complete pathogen. A number of vectored vaccines against SARS-CoV have been described. A recombinant vesicular stomatitis virus (VSV) expressing the SARS-CoV S protein was protective in mice (14) but has not been tested in primates, and the safety of VSV in humans remains to be established. Recombinant modified vaccinia Ankara (MVA) virus expressing SARS-CoV S protein was immunogenic and protective in rodent and primate models (15–17); however, one group found that the MVA/SARS-CoV S-immunized ferrets developed hepatitis upon challenge with SARS-CoV (17). Vaccine constructs based on replication defective human adenovirus type 5 expressing a partial or full-length SARS-CoV S protein have been evaluated for immunogenicity in rats and monkeys (18, 19), but immunization depends on a high vaccine dose, and safety and protective efficacy remain to be demonstrated. We previously used an attenuated version of human parainfluenza virus type 3, a common pediatric respiratory pathogen, to express the SARS-CoV S protein and showed that a single intranasal (i.n.) and intratracheal (i.t.) inoculation was immunogenic and protective against SARS-CoV challenge in hamsters and African green monkeys (AGM) (20). This vector has a natural tropism for the respiratory tract and does not significantly spread beyond that site, properties desirable for use as a safe, restricted, topical respiratory tract vaccine. However, a concern that exists with any vector based on a common pathogen is that the adult population has significant immunity from prior exposure that will restrict infection and replication of the viral vector and reduce its immunogenicity. Indeed, comparisons of the immunogenicity of vaccinia virus-vectored and human adenovirus type 5-vectored vaccines in rodents, non-human primates, and humans demonstrated that preexisting immunity to the vector greatly reduced the immunogenicity of these vaccines (21–23).

An additional limitation of previous studies of vectored and inactivated SARS-CoV vaccines in non-human primates, the model that is most similar to humans, is that protective efficacy was measured by quantitation of challenge SARS-CoV shed in respiratory secretions (6, 7, 16, 20). It is now known that the virus replication in nasal and lung tissues occurs at a much higher level than previously appreciated and is poorly represented by shed virus (24). Also, in two of the above-mentioned studies (6, 7), protective efficacy of SARS-CoV vaccine candidates was evaluated by analysis of lung tissues collected by biopsy or necropsy on days 12 or 15 after the challenge; however, the virus is normally cleared by this time (24). Thus, evaluation of vaccine efficacy necessitates direct quantitative virological analysis of infected tissues at the peak of viral replication.

The present work involves Newcastle disease virus (NDV) as a potential vaccine vector against SARS-CoV. Like human parainfluenza virus type 3, NDV is a nonsegmented negative strand RNA virus of the Paramyxoviridae family. However, its natural host is birds, and it is antigenically distinct from common human pathogens. The severity of avian disease depends on the pathotype of the NDV strain: so-called lentogenic strains cause mild or asymptomatic infections that are restricted to the respiratory tract; mesogenic strains are of intermediate severity; and velogenic strains can cause systemic infections with high mortality (25). Lentogenic and, in some cases, mesogenic strains are used as live NDV vaccines for poultry (26). Pertinent to the present report, a major determinant of NDV virulence in birds is activation of the viral fusion F glycoprotein, which occurs by a cleavage event mediated by a cellular endoprotease. In velogenic and some mesogenic strains, the cleavage site of F is rich in basic amino acids and is readily cleaved by ubiquitous intracellular proteases such as furin, thus permitting replication in a variety of tissues. With lentogenic and some mesogenic strains, the site contains fewer basic residues and depends on a secreted protease found primarily in the lung for cleavage, thus restricting replication to the epithelial surface where the secretory protease is found (27).

Intranasal inoculation of mice with recombinant NDVs expressing influenza virus hemagglutinin or respiratory syncytial virus fusion protein resulted in an induction of antibody responses against the viruses and complete or partial protection against challenge (28–30). However, the level of host range restriction and immunogenicity of an NDV vector in mice is not necessarily predictive of its properties in primates. Intranasal and i.t. inoculation of AGM and rhesus monkeys with lentogenic and mesogenic strains of NDV resulted in low-level, scattered replication of the virus in the respiratory tract without any disease signs and with little or no virus shedding (31). The virus appeared to be restricted to the respiratory tract: virus was not detected in other organs or blood harvested at the peak of viral replication, and virus replication was not detected after parenteral inoculation (A.B., B.R.M., P.L.C., unpublished observations). In humans, infection by NDV appears to be limited and benign based on both anecdotal observations with bird handlers (32) and clinical studies using the virus as an oncolytic agent (33).

In the present study, we evaluated NDV as a topical respiratory vaccine vector with SARS-CoV as the target pathogen. Importantly, protective efficacy against SARS-CoV challenge was evaluated by direct analysis of nasal and lung tissues collected by necropsy at the peak of SARS-CoV replication (24).

Results

Design of the Vaccine Constructs.

Two NDV vectors were evaluated. One was a recombinant copy of the mesogenic Beaudette C strain (NDV-BC) that was previously shown to be highly attenuated in non-human primates (31). The second was a recombinant copy of the lentogenic LaSota strain [which also is highly attenuated in primates (see ref. 31)] that was modified so that the cleavage sequence of its F protein was replaced with that of NDV-BC, resulting in the virus NDV-VF. In chickens, the imported cleavage site rendered NDV-VF capable of replication beyond the respiratory tract, resulting in a virus that is intermediate in virulence in birds compared with its lentogenic and mesogenic parents (27). The NDV-BC and NDV-VF vectors were used to construct vaccine viruses expressing the full-length 1,255-aa SARS-CoV S protein, resulting in viruses NDV-BC/S and NDV-VF/S (Fig. 1). Because insertion of foreign genes into genomes of nonsegmented negative strand viruses such as NDV can confer attenuation that increases with insert size (34), we also made a separate NDV-BC-vectored vaccine construct expressing the N-terminal 762-aa S1 domain (NDV-BC/S1). The S1 domain has been reported to contain the receptor-binding site of S as well as major neutralizing epitopes (35). The various vaccine viruses were readily recovered and amplified, achieving peak titers of ≈10^7.5 to 10^8.0 pfu/ml in DF-1 cells and 10^8.5 to 10^9.0 in embryonated chicken eggs (data not shown). These values were similar to the peak titers obtained with the parental NDV-BC and NDV-LS(VF) viruses.

Fig. 1.

NDV vaccine constructs. The SARS-CoV S or S1 ORF was cloned by using XbaI sites (italicized) into NDV-BC or NDV-VF antigenomic cDNA under the control of a set of NDV gene-start (GS) and gene-end (GE) transcription signals that direct its expression as a separate mRNA. NDV genes are shown as gray boxes and the SARS-CoV S gene as black boxes. The nucleotide sequence spanning the inserted expression cassette was identical in both NDV vectors and is shown at the top of the figure in the DNA-positive sense. The intergenic nucleotide between each gene is indicated by an arrow.

In Vitro Characterization of the Vaccine Constructs.

To detect expression of the foreign gene inserts, DF-1 cell monolayers were infected with NDV-BC, NDV-BC/S, or NDV-BC/S1 and incubated for 24 h. Cell lysates were analyzed by Western blotting with NDV- and SARS-CoV S-specific antibodies. NDV-BC-based constructs produced high levels of S or S1 protein [see supporting information (SI) Fig. 4], with similarly high levels of S produced by the NDV-VF-based construct (data not shown). In some cases, expression of foreign viral transmembrane glycoproteins by recombinant nonsegmented negative strand viruses results in the incorporation of these proteins into virus particles (for review, see ref. 34). Therefore, we purified NDV particles from the medium of infected cells and assayed them for the presence of the expressed proteins. This assay showed that a distinct but faint SARS-CoV S-specific band could be seen in virions expressing the S full-length protein but not in virions expressing the S1 protein (SI Fig. 4), which suggests that the S protein was incorporated into the NDV virion and that incorporation required the presence of the transmembrane domain located at the C terminus. In some instances, incorporation of foreign surface glycoproteins by nonsegmented negative-strand viruses rendered the vector sensitive to neutralization by antibodies specific to the foreign protein and even allowed the particle to escape neutralization by vector-specific neutralizing antibodies (34). However, that was not observed here: the NDV vectors expressing the SARS-CoV S or S1 protein were fully sensitive to NDV-neutralizing antibodies but were not sensitive to SARS-CoV-neutralizing antibodies (data not shown). Comparison of the multi step growth kinetics of the viruses demonstrated that expression of SARS-CoV S protein did not alter viral replication in DF-1 chicken embryo fibroblasts but was associated with up to an ≈10-fold (NDV-BC backbone) or 100-fold (NDV-VF backbone) reduction in viral yield in Vero monkey kidney cells and in A549 human alveolar epithelial cells (data not shown). To evaluate whether incorporation of the S protein affects the pathogenicity of NDV for birds, we determined the mean embryo death time of NDV-BC and NDV-BC/S in 9-day-old embryonated eggs, as described previously (36). The mean embryo death time for the two viruses was found to be 60 and 61 h, respectively, suggesting that pathogenicity of NDV-BC/S is not increased compared with its NDV-BC parent.

Immunogenicity.

Groups of AGM were immunized as shown in Table 1. Sera were collected on days 0, 28, and 56 and analyzed by a SARS-CoV-neutralization assay. After one dose of either vaccine construct, the titers of S-specific antibodies were low; however, a strong increase in the viral titers was achieved after the second dose was administered. For NDV-BC/S and NDV-VF/S, the mean neutralization titers after the second dose were 1:97 and 1:630, respectively. Surprisingly, the latter titer was greater than the average titer of SARS-CoV-specific neutralizing antibodies in serum samples of four monkeys immunized against SARS-CoV and challenged with SARS-CoV in a previous study (1:363) (20), which were used as positive control sera in the present study (see the second footnote for Table 1). The development of a vector-specific serum antibody response, as evaluated by NDV hemagglutination inhibition (HAI), generally paralleled the SARS-CoV-specific response: a low-level response was detected after one dose followed by a substantial increase after two doses (Table 1).

Table 1.

Serum antibody responses against the NDV vector and SARS-CoV S insert in immunized AGM

Virus	Monkey	NDV HAI titer, reciprocal log2†			SARS-CoV-neutralization titer, reciprocal log2†‡
		Day 0	Day 28	Day 56	Day 0	Day 28	Day 56
NDV-BC, 2 doses	W389	≤1	4	7	2.5	3.0	2.0
	W410	≤1	5	8	2.7	2.3	2.0
	Mean	1	4.5	7.5	2.6	2.7	2.0
NDV-BC/S, 1 dose	W517	≤1	≤1	≤1	2.0	2.7	2.5
	W613	≤1	5	7	2.3	3.7	3.3
	W653	≤1	3	4	2.3	3.0	3.0
	W896	≤1	≤1	≤1	≤1	3.5	3.3
	Mean§	1	2.5	3.3	2.0	3.2**	3.0*
NDV-BC/S, 2 doses	X049	≤1	4	7	2.5	2.5	10.5
	X059	≤1	4	7	1.7	2.5	3.5
	X084	≤1	4	8	2.5	3.0	7.7
	X149	≤1	3	7	2.7	3.7	4.7
	Mean§	1	3.8**	7.3***	2.3	2.9	6.6*
NDV-VF/S, 2 doses	X181	≤1	6	7	2.5	3.5	8.0
	X184	≤1	5	8	3.0	4.7	10.5
	X185	≤1	4	7	2.5	3.5	8.0
	X192	≤1	≤1	7	3.0	4.7	10.5
	Mean§	1	4.0	7.3***	2.8	4.1*	9.3**

AGM were immunized by the combined i.n. and i.t. routes with 10⁷ pfu of the indicated virus per site on days 0 and 28 (the two-dose groups) or on day 0 only (the one-dose group). NDV-BC is an empty vector control. The NDV-BC/S1 construct was not included in this experiment.

^†Lower limits of antibody detection: 2.0 log₂ for the NDV HAI titer and 1.7 log₂ for SARS-CoV-neutralization titer. When the titers were below the detection limits, a value of 1 log₂ was assigned for calculation of the mean titers.

^‡As a positive control for SARS-CoV-neutralizing titer assays, serum samples from four AGM that had been successfully immunized against SARS-CoV using a parainfluenza virus type 3 vector expressing the SARS-CoV S protein and challenged 28 days later with SARS-CoV in a previous study were used; these samples had been collected on day 28 after the challenge (20). The SARS-CoV-neutralizing titers (reciprocal log₂) of these animals determined in the present assay are the following: 8.5, 9.7, 7.5, and 8.3; mean titer is 8.5.

^§Statistical significance of the antibody titers on days 28 and 56 compared with the corresponding values on day 0 (two-tailed Student's t test) are shown:

*, P < 0.05;

**, P < 0.01;

***, P < 0.001.

We also evaluated cellular responses against the S protein in the peripheral blood of immunized animals on days 10, 38 (10 days after the second dose), and 56 (28 days after the second dose). Only live, CD3-positive cells were included for analysis, and values were calculated as percentages of total CD8 or CD4 cells positive for IFN-γ or TNF-α. There was an increase in the mean numbers (horizontal bars) of IFN-γ-positive and TNF-α-positive CD8 T cells in animals after the second dose of the vaccine constructs; however, the increase was not statistically significant because of high variability between the individual animals (Fig. 2). We did not observe a similar increase in the number of CD4 T cells positive for TNF-α or IFN-γ (data not shown). There were no significant positive populations in the absence of stimulation, whereas nonspecific stimulation with phorbol 12-myristate 13-acetate and ionomycin resulted in percentages of total CD8 cells positive for IFN-γ and TNF-α of 35.5% and 39.6%, respectively, and CD4 cells positive for IFN-γ and TNF-α of 5.1% and 33.2%, respectively (data not shown), thus confirming the general responsiveness of the cells. The low numbers of SARS-CoV-specific T cells after immunization through the respiratory tract was expected because the numbers of T cells specific to common respiratory pathogens is much lower in human peripheral blood than in lung tissue (37). The design of the present study precluded sampling of lung tissues for isolation of pulmonary T cells.

Fig. 2.

Quantitation of SARS-CoV S-specific CD8+ T cells in the peripheral blood of AGM immunized with the NDV vaccine constructs. Peripheral blood mononuclear cells (PBMC) were stimulated with peptides specific to the S protein, stained, and analyzed by flow cytometry. Cells positive for either IFN-γ or TNF-α were plotted as a percentage of total CD8+ cells. The individual values and the mean values (horizontal bars) were plotted for each group of animals; each group contained four animals except that the empty NDV-BC vector control group contained two animals. The arrows indicate the timing of the first and second doses (days 0 and 28).

Protective Efficacy Against SARS-CoV Challenge.

AGM were immunized as described above with two doses of NDV-BC/S, NDV-VF/S, or NDV-BC/S1 and 28 days after the second dose, they were challenged by the i.n. and i.t. routes with SARS-CoV at a tissue culture 50% infectious dose (TCID₅₀) of 10⁶ per site. The animals were killed on day 2 after the challenge, which is the peak of viral replication in this species (24), and samples of respiratory tract tissues were collected by necropsy and analyzed for virus titer by titration on Vero cell monolayers (Fig. 3). In the control animals immunized with the empty NDV-BC vector, mean SARS-CoV titers reached 10^4.1 TCID₅₀ per cm³ in the nasal turbinates and 10^6.3 TCID₅₀ per cm³ in the trachea. In various parts of the lung, mean viral titers ranged from 10^4.9 TCID₅₀ per cm³ in the upper lobe of the left lung to as high as 10^6.9 TCID₅₀ per cm³ in the hilar region of the right lung. Of the vaccine constructs tested here, only NDV-BC/S1 was relatively ineffective against SARS-CoV challenge, although there were minimal reductions in mean SARS-CoV titer relative to the vector-immunized animals in all tissues except for the lower lobe of the right lung. Both NDV-VF/S and NDV-BC/S were effective at inducing protective immunity. Immunization with NDV-VF/S resulted in a 5-fold and 61-fold reduction in nasal turbinate and tracheal SARS-CoV titers, respectively, compared with the control animals. More importantly, this construct resulted in a 236-fold reduction of viral titer in the lung (average reduction from all seven lung tissue sections). The NDV-BC/S construct was even more effective, with average reductions in viral titer of 13-fold, 276-fold, and 1,102-fold in the nasal turbinate, trachea, and lung, respectively. To confirm the reduction in viral load seen in immunized animals, we also performed SARS-CoV N gene-specific quantitative RT-PCR on RNA extracted from three representative low-titer tissue samples and three representative high-titer tissue samples. Samples with a SARS-CoV titer between 10² and 10³ TCID₅₀ per cm³ tissue (i.e., protected animals) had a threshold cycle (C_T) that was 24–29 cycles later than the C_T for samples with a titer between 10⁶ and 10⁷ TCID₅₀ per cm³ tissue (i.e., unprotected animals) (data not shown). These results indicate that the observed differences in viral titer are caused by a true reduction of viral titer in vivo, and not because of in vitro neutralization after tissue homogenization.

Fig. 3.

SARS-CoV replication in the upper and lower respiratory tract of AGM after challenge. Animals were immunized by the i.n. and i.t. routes (10⁷ pfu per site) with the indicated NDV recombinant on days 0 and 28, and they were challenged on day 56 by the i.n. and i.t. sites with 10⁶ TCID₅₀ of SARS-CoV per site. Two days later, the animals were killed, and duplicate samples were taken from the nasal turbinates, the trachea, and the indicated lung regions. The replicate tissue specimens were each titrated in quadruplicate on separate days to measure infectious SARS-CoV. The reported value for each organ is the average of 8 log-transformed titers for each group (i.e., two samples per organ taken from each of four animals). The lower limit of detection was 10 TCID₅₀ per ml. The average ± SE of the log-transformed TCID₅₀ per ml values are shown. P values were calculated by using a two-tailed Student's t test. ∗, P < 0.05; ∗∗, P < 0.01; ∗∗∗, P < 0.001.

Discussion

Inoculation of the upper respiratory tract of AGM with recombinant NDV expressing the SARS-CoV S protein was immunogenic and protective against a high challenge dose of SARS-CoV. Immunization with two doses of the NDV-VF/S or NDV-BC/S vaccine candidate resulted in a titer of serum SARS-CoV-neutralizing antibodies that equaled or exceeded that achieved in animals that had received a protective immunization with a parainfluenza virus type 3 vector expressing the S protein followed by i.n. challenge with a high dose of SARS-CoV (i.e., animals whose serum neutralizing titers were reflective of secondary exposure to SARS-CoV). Based on this strict standard, the NDV vectors seemed to be very immunogenic despite their high level of attenuation, which resulted in a 236-fold or 1,102-fold (depending on the NDV vector) reduction in pulmonary SARS-CoV challenge virus, a reduction that likely would be sufficient to prevent mortality and reduce morbidity associated with SARS-CoV infection. Because respiratory viruses typically induce a strong mucosal response in the respiratory tract, the NDV vaccine constructs should be particularly effective in reducing virus shedding from the respiratory tract and thus reducing transmission.

A single dose of the NDV vaccine candidates was weakly immunogenic for the SARS-CoV S insert protein while being moderately immunogenic for the NDV vector protective antigens. We did not evaluate protective efficacy after one dose, and it is not known whether this weak primary immunization would confer a significant amount of protection against SARS-CoV. The second dose of the NDV vaccine constructs was immunogenic despite the presence of HAI antibodies against the NDV vector induced by the primary immunization, indicating that the second vaccine dose replicated sufficiently well to provide a boost. Administration of two vaccine doses would be readily accomplished for preepidemic immunization. In the setting of an outbreak, it would be preferable to have a single immunization. One possibility is that this topical vaccine could be combined with a parenteral immunization by using another SARS-CoV vaccine to augment protection from a single immunization.

Unexpectedly, the NDV-BC/S1 construct, expressing the N-terminal 60% of the S protein, failed to induce significant protection against SARS-CoV challenge. We had evaluated expression of this smaller fragment as an alternative strategy in case expression of the complete protein proved to overattenuate the vector. The lack of significant immunogenicity of NDV-expressed S1 might be explained by its lack of incorporation into the NDV particle, which was likely because of the absence of the transmembrane domain. Past studies have demonstrated that incorporation of a foreign protein into virion particles significantly increased its immunogenicity (38). Alternatively, the reduced immunogenicity of NDV-BC/S1 might mean that the S2 domain of the S protein contains or contributes to one or more important neutralizing epitopes (39).

We evaluated two different NDV vectors, namely the mesogenic NDV-BC strain and NDV-VF, a version of the lentogenic LaSota strain that had been modified to possess the F cleavage site of the mesogenic virus. The introduction of this more easily cleaved sequence was done primarily for a practical reason: the parental LaSota strain depends on added trypsin for activation of infectivity during in vitro growth, a requirement that was eliminated by the change in cleavage sequence. Thus, both vectors have an F protein that is readily cleaved without extracellular protease and thus have the potential for replication in a wide range of tissues. Even so, NDV-BC and NDV-VF resembled the previously evaluated LaSota strain (31) in being highly restricted and limited to replication in the respiratory tract, which suggests that the host range restriction of NDV in primates is severe and is based largely on factors other than activation of the F protein. Differences between avian and primate cells with regard to sialic acid receptors for viral attachment comprise one possible factor in the host range restriction. It also might be that NDV is less able to block the type I IFN response in primate cells, as suggested by previous studies (40). In the case of the host range restriction of bovine parainfluenza virus type 3 in primates, each gene appeared to make a contribution (41). Regardless of the mechanism, it is clear that NDV is very highly restricted for replication in primates, and the level of restriction appears to be similar for the mesogenic and lentogenic pathotypes based on this work and that performed previously (31).

The use of NDV as a vaccine vector offers a number of advantages over alternative strategies mentioned above. NDV is antigenically distinct from common human pathogens and vaccines and therefore should be infectious and immunogenic in the general human population. NDV has had considerable use as an oncolytic agent (www.cancer.gov/cancertopics/pdq/cam/NDV/healthprofessional), demonstrating that it is safe when administered parenterally to humans (33). An outbreak of NDV in poultry workers, as subsequently documented by their seroconversion, was characterized by a transient conjunctivitis that, in most cases, lasted 3–4 days with a lack of systemic symptoms (32). A high degree of attenuation also has been demonstrated in non-human primates, where i.n. and i.t. inoculation of 10⁷ pfu resulted in only scattered, low-level replication that was restricted to the lungs, with no apparent adverse clinical effects (31). The restriction of NDV replication to the epithelial cells of the respiratory tract allows for stimulation of local and systemic immunity while reducing safety concerns that would be associated with viremia and viral spread to secondary sites. In the present work, we used a combined i.n. and i.t. route of immunization. In clinical studies, the i.n. route alone would be appropriate. We are currently testing the efficiency of vaccine delivery by the i.n. route alone. Should that be insufficiently immunogenic, alternative methods of delivery will be developed. The needle-free inoculation also offers a practical benefit in that untrained personnel would be able to administer the vaccine, which would be a considerable advantage in developing countries. We found that all of the NDV constructs replicated to acceptably high titers in Vero cells qualified for the production of vaccines for human use. It also is noteworthy that infection of non-human primates with NDV resulted in only low-level virus shedding into respiratory secretions (31) and hence into the environment. Furthermore, genetic exchange involving nonsegmented negative-strand RNA viruses such as NDV is insignificant. This consideration essentially eliminates the concern of recombination with circulating viruses, which is particularly relevant for live vaccines against coronaviruses and influenza viruses.

A hypothetical concern in SARS-CoV vaccine development is the possibility of immune-mediated disease enhancement upon exposure to the authentic virus. It was found previously that antibodies to feline infectious peritonitis coronavirus, acquired by immunization or by passive immunization, led to enhanced disease in felines after exposure to the virus (42). However, differences in the biology of that virus compared with SARS-CoV make the relevance of this precedent unclear. The only potential evidence of a similar phenomenon for SARS-CoV was the finding that MVA expressing the S protein may have led to hepatitis in ferrets after challenge with SARS-CoV (17); however, it is unclear how this hepatitis occurred or whether it has any relevance to immune-enhanced disease. We did not see any evidence of enhanced clinical disease on day 2 after infection with SARS-CoV; however, the present study was not designed to address the possibility of enhanced disease at later time points. Another potential concern is that the vaccine construct might spread to birds and cause disease. However, as already noted, lentogenic and, occasionally, mesogenic strains have widespread use as live NDV vaccines for poultry (26). In the present work, although the foreign S protein was incorporated into the vector particle, the virus was not susceptible to neutralization by antibodies specific to the S protein and did not lose sensitivity to NDV-specific antibodies. Moreover, presence of the S insert did not increase pathogenicity of the virus in birds, as evaluated by the mean embryo death time. In recent studies, the insertion of foreign genes, including that encoding the hemagglutinin protein of highly pathogenic avian influenza virus, into the genomes of lentogenic or mesogenic strains of NDV resulted in reduced rather than increased pathogenicity in birds, even when the foreign glycoprotein was from a highly pathogenic avian influenza virus and was incorporated into the vector particle (38, 43). Taken together, these data suggest that the vaccine constructs present no environmental or agricultural risk, although this possibility will continue to be carefully monitored.

In summary, we have evaluated two promising vaccine candidates, NDV-BC/S and NDV-VF/S, and demonstrated a substantial level of immunogenicity and protective efficacy against SARS-CoV challenge in a primate model by direct analysis of respiratory tract tissues at the peak of the challenge virus replication. These topical respiratory vaccine candidates are very highly restricted in non-human primates, do not shed significantly, should be further evaluated in clinical studies, and may provide a new vector suitable for use against SARS-CoV or other emerging viruses.

Materials and Methods

Viruses.

The full-length (1,255-aa) SARS-CoV S ORF (GenBank accession number AAP13441.1) or the S1 subunit (the first 762 codons) was amplified by PCR using the full-length SARS-CoV S cDNA as a template (20). The primers were designed to add flanking XbaI sites and an NDV gene junction while maintaining the genome length as a multiple of six nucleotides (44) (Fig. 1). Each PCR product was subcloned into an XbaI site that had been inserted into the P–M junction of a full-length cloned cDNA of the NDV antigenome (27, 31). NDV recombinants were recovered as described previously (43), passaged twice on HEp-2 cells, twice on chicken embryo DF-1 cells, and once in embryonated chicken eggs. The integrity of the inserts was confirmed by RT-PCR of viral RNA and sequence analysis. NDV titers (pfu/ml) were determined by plaque titration on monolayers of DF-1 cells in 24-well plates.

SARS-CoV (Urbani strain) was provided by L. Anderson and T. Ksiazek of the Centers for Disease Control and Prevention (Atlanta, GA) and propagated in Vero cells (24). All experiments involving infectious SARS-CoV were done under biosafety level 3 conditions. SARS-CoV titers (TCID₅₀ per ml) were determined by limiting dilution on monolayers of Vero cells in 96-well plates.

Immunization and Challenge of AGM.

Juvenile AGM (Cercopithecus aethiops), seronegative for NDV by HAI with turkey erythrocytes, were immunized i.n. and i.t. as described previously (20) with 10⁷ pfu of NDV per site on days 0 and 28 or on day 0 only for animals receiving one dose (Table 1). Serum samples were taken at days 0, 28, and 56 to determine antibody titers. Blood samples were taken on days 0, 10, 38, and 56 for isolation of peripheral blood mononuclear cells (PBMCs). In addition, blood samples were taken at days 2 and 4 (after the first immunization) and days 30 and 32 (days 2 and 4 after the second immunization) and evaluated for the presence of NDV by titration on DF-1 cells. They were found to be negative (data not shown). Animals were observed for any signs of clinical symptoms throughout the course of the work.

To assess the protective efficacy, a separate group of animals was immunized as described above and challenged on day 56 by the combined i.n. and i.t. routes with 10⁶ TCID₅₀ of SARS-CoV per site. Two days after challenge, animals were killed, and duplicate 1-cm³ samples were taken from the nasal turbinates, the trachea, and from seven regions of the lung: the left hilar region, the left upper and lower lobes, the right hilar region, and the right upper, middle, and lower lobes. Tissue samples were homogenized, and SARS-CoV titers were determined by titration in Vero cells. As an independent assessment of viral load in the tissue samples, RNA was extracted, and viral genomes were quantified by using a quantitative RT-PCR assay as described previously (45). A complete set of serum samples was not available from the challenged monkeys, precluding evaluation of the antibody response in these animals. All primate experiments were performed at Bioqual, Inc. (Rockville, MD), a site approved by the Association for Assessment and Accreditation of Laboratory Care International with a protocol approved by the Animal Care and Use Committee of the National Institute of Allergy and Infectious Diseases, National Institutes of Health.

SARS-CoV S-Specific Cellular Immune Responses in AGM.

PBMCs from whole blood were banded by centrifugation over lymphocyte separation medium (Cellgro, Herndon, VA), washed, and resuspended in RPMI medium 1640 (Invitrogen, Carlsbad, CA) containing 10% FBS and 10% dimethyl sulfoxide. PBMC samples were incubated overnight in medium containing GolgiStop (BD Biosciences, Franklin Lakes, NJ) and either phorbol 12-myristate 13-acetate/ionomycin (nonspecific stimulation) or 2.5 μg/ml of a pool of 169 17- to 19-mer peptides encompassing the entire SARS S protein. The samples were then washed and stained with FITC-conjugated mouse anti-human CD8α, peridinin chlorophyll-conjugated mouse anti-human CD4, and R-phycoerythrin-conjugated mouse anti-human CD3. The cells were washed, fixed, and permeabilized with Cytofix/Cytoperm solution (BD Biosciences) and stained with allophycocyanin (APC)-conjugated mouse anti-human IFN-γ or APC-conjugated mouse anti-human TNF-α (all from BD Biosciences). Flow cytometric analysis (FACSCalibur; BD Biosciences) involved 30,000 events per sample.

Abbreviations

AGM

African green monkeys

BC

Beaudette C

HAI

hemagglutination inhibition

i.n.

intranasal

i.t.

intratracheal

MVA

modified vaccinia Ankara

NDV

Newcastle disease virus

PBMC

peripheral blood mononuclear cell

SARS-CoV

severe acute respiratory syndrome-associated coronavirus

TCID₅₀

tissue culture 50% infectious dose.