SARS-CoV expresses several structural proteins, including nucleocapsid, membrane, envelope, and spike (S) proteins (1). All may serve as antigens to induce neutralizing antibodies and protective responses. In general, prior to identification of the protein that contains the major neutralizing epitopes, the inactivated virus may be used as the first-generation vaccine because it is easy to generate whole killed virus particles. However, once the neutralizing epitopes are identified, the inactivated virus vaccine should be replaced by vaccines based on fragments containing neutralizing epitopes since they are safer and more effective. Several reports have showed that SARS-CoV inactivated with formaldehyde, UV light, and β-propiolactone can induce virus-neutralizing antibodies in immunized animals (8–11), and the first inactivated SARS-CoV vaccine is being tested in the clinical trials in China. However, safety of the inactivated vaccine is a serious concern; production workers are at risk for infection during handling of concentrated live SARS-CoV, incomplete virus inactivation may cause SARS outbreaks among the vaccinated populations, and some viral proteins may induce harmful immune or inflammatory responses, even causing SARS-like diseases (12,13).

The S protein of SARS-CoV, a type I transmembrane glycoprotein, is responsible for virus binding, fusion, and entry and is a major inducer of neutralizing antibodies (1,14). S protein consists of a signal peptide (SP: amino acids [aa] 1–12) and 3 domains: an extracellular domain (aa 13–1193), a transmembrane domain (aa 11194–1215), and an intracellular domain (aa 1216–1255). Its extracellular domain consists of 2 subunits, S1 and S2 (14), although the cleavage site between these subunits has not been clearly defined. The S1 subunit is responsible for virus binding to the receptor, angiotensin-converting enzyme 2 (ACE2) (15,16). A fragment located in the middle region of the S1 subunit (aa 318–510) is the receptor-binding domain (RBD) for ACE2 (17–19). SARS-CoV may also bind to cells through the alternative receptors DC-SIGN or L-SIGN (20,21), but the binding sites for these alternative receptors have not been defined. The S2 subunit, which contains a putative fusion peptide and 2 heptad repeats (HR1 and HR2), is responsible for fusion between the viral and target cell membranes. Infection by SARS-CoV is initiated by binding of RBD in the viral S protein S1 subunit to ACE2 on target cells. This forms a fusogenic core between the HR1 and HR2 regions in the S2 domain that brings the viral and target cell membranes into close proximity, which results in virus fusion and entry (22–24). This scenario indicates that the S protein may be used as a vaccine to induce antibodies for blocking virus binding and fusion.

Several recombinant vector-based vaccines expressing SARS-CoV S protein have been assessed in preclinical studies. Yang et al. (25) reported that a candidate DNA vaccine encoding the full-length S protein induced neutralizing antibodies (neutralizing titers ranging from 1:50 to 1:150) and protected mice from SARS-CoV challenge. Using DNA vaccines encoding the full-length and segments of S protein to immunize rabbits, Wang et al. have produced higher titers of neutralizing antibodies and demonstrated that major and minor neutralizing epitopes are located in the S1 and S2 subunits, respectively (26). Other groups also found neutralizing epitopes in the S2 subunit (27,28). Bisht et al. (29) have shown that intranasal or intramuscular inoculations of mice with highly attenuated modified vaccinia virus Ankara (MVA) vaccines encoding full-length SARS-CoV S protein also produce neutralizing antibodies with mean neutralizing titers of 1:284. Bukreyev et al. (30) reported that mucosal immunization of African green monkeys with an attenuated parainfluenza virus expressing S protein resulted in production of neutralizing antibodies and protected animals from infection by challenge with SARS-CoV. These data suggest that the S protein can induce neutralizing antibodies and protective responses in immunized animals.

Using convalescent-phase sera from SARS patients and a set of peptides spanning the entire sequence of the SARS-CoV S protein, we have identified 5 linear immunodominant sites (IDS) in the S protein (Figure 2A). IDS I, II, III, and V reacted with >50% of the convalescent-phase sera from SARS patients, while IDS IV was reactive with >80% of SARS sera, suggesting that IDS IV is the major immunodominant epitope on the S protein (31). Synthetic peptides corresponding to IDS could induce high titers of S protein–specific antibodies, but none of these antibodies possesses neutralizing activity. These findings suggest that the IDS in S protein may not induce neutralizing antibodies. Whether these antibodies enhance infection by heterologous SARS-CoV strains or mediate harmful immune responses is unclear. The S protein of FIPV expressed by recombinant vaccinia can cause antibody-dependent enhancement of disease if vaccinated animals are subsequently infected with wild-type virus (32). Our previous studies on HIV-1 showed that antibodies against some immunodominant epitopes in the HIV-1 envelope glycoprotein could enhance infection by heterologous HIV-1 strains (33). Most recently, Yang et al. (6) demonstrated that the polyclonal and monoclonal antibodies against S protein of the late SARS-CoV (Urbani strain) could neutralize infection by the relevant late SARS-CoV strains. However, these antibodies enhanced infection by an early human SARS-CoV isolate (GD03T0013) and the civet SARS-CoV–like viruses. These investigators have shown that the ACE2-binding domain mediates the antibody-dependent enhancement of civet SARS-CoV–like virus entry (6). Theoretically, some antibodies to the ACE2-binding domain may enhance infection if these antibodies closely mimic the receptor ACE2 and induce similar conformational changes, as the receptor likely does. The S protein with truncation at aa 1153 failed to cause antibody-dependent enhancement of infection, although it still induced neutralizing antibodies. This finding suggests that removal of the aa 1153–1194 region may abrogate induction of virus infection–enhancing antibodies (6). Vaccination of ferrets with MVA-based SARS vaccine expressing full-length S protein caused liver damage after animals were challenged with SARS-CoV (34). These findings raised concerns about the efficacy and safety of the vaccines containing or expressing full-length S protein.

RBD, a fragment (≈193 aa residues) in the middle of S1 subunit of S protein (Figure 2B), is responsible for virus binding to the receptor on target cells. We have demonstrated that the antisera from SARS patients and from animals immunized with inactivated SARS-CoV reacted strongly with RBD (9,35). Absorption of antibodies by RBD from these antisera results in the removal of most of the neutralizing antibodies, and RBD-specific antibodies isolated from these antisera have potent neutralizing activity (35,36). We have also shown that rabbits and mice immunized with RBD produced high titers of neutralizing antibodies against SARS-CoV with 50% neutralizing titers at a >1:10,000 serum dilution (37). The immunized mice were protected from SARS-CoV challenge (unpub. data). The antibodies purified from the antisera against SARS-CoV significantly inhibited RBD binding to ACE2 (9,36–38). Using spleen cells from mice immunized with RBD, we have generated a panel of 25 monoclonal antibodies (MAbs) that recognize different conformational epitopes on RBD and possess potent neutralizing activity (38). Our result is in agreement with the report by van den Brink et al. (39), who identified 3 human neutralizing anti-S MAbs from antibody phage display libraries by using inactivated SARS-CoV as the target. These researchers also found that all of these MAbs specifically bound to RBD and blocked interaction between RBD and ACE2. These findings suggest that RBD contains the major neutralizing epitopes in the S protein and is an ideal SARS vaccine candidate because RBD contains the receptor-binding site, which is critical for virus attachment to the target cell for infection (15,17–19). Antibodies specific for RBD are expected to block binding of virus to the target cell. RBD induces higher titers of neutralizing antibodies than those vaccines expressing the full-length S protein (25,26,29,30,37,38). RBD sequences among the late SARS-CoV strains are highly conserved. When the early and late SARS-CoV strains are compared, only 3 to 5 aa residues are variable among the 193 residues in RBD and most of the isolates vary by only 1 residue (4). van den Brink et al. (39) showed that 1 human MAb (CR3014) specific for RBD of SARS-CoV strain FM1 can effectively bind to most RBDs of the early and late SARS-CoV strains. These data suggest that antibodies directed against RBD of a SARS-CoV isolate may neutralize infection by a broad spectrum of SARS-CoV strains. Therefore, recombinant proteins containing RBD or vectors encoding RBD may be used as vaccines for preventing infection by SARS-CoV with distinct genotypes.