Both human and murine monoclonal antibodies (MAbs) have been developed against three late-phase SARS-CoV strains, strains Urbani, Tor-2, and HKU-39849, and in vitro neutralizing activity has been described (
46-
48). The recent development of a method to isolate a large number of MAbs from SARS patients provides the reagents needed to characterize the homologous and heterologous neutralizing responses after natural SARS-CoV infection (
47). Although studies using pseudotyped lentiviruses and recombinant SARS-CoV RBD protein have shown some cross-neutralizing or cross-reactive activity (
12,
24,
43,
56,
57), the neutralizing activities of these MAbs have not been tested against actual heterologous SARS-CoV strains from the middle, early, or zoonotic phase of the epidemic or in lethal models of disease. This is potentially problematic, since the absence of human cases over the past 2 years suggests that future epidemics will likely result from zoonotic transmission. Consequently, antibodies that provide robust cross-neutralization activity are essential to interrupt zoonotic transmission and contain future epidemics (
3,
36).
We recently developed several lethal SARS-CoV challenge models with BALB/c mice that recapitulated the age-related clinical signs, weight loss exceeding 20% as well as severe lung pathology, by using recombinant SARS-CoV bearing the S glycoprotein of early human and zoonotic strains (
37). A second pathogenic model for young mice was also developed by serial passage of the Urbani isolate in BALB/c mice, resulting in MA15, which replicates to high titers in the lung and causes clinical disease, weight loss exceeding 20%, and severe alveolitis (
33). In the present study, we used a panel of isogenic SARS-CoVs bearing human and zoonotic S glycoproteins to categorize the human MAbs into six distinct neutralization profiles. Moreover, we identify four neutralizing antibodies that neutralize all zoonotic and human strains tested, and we demonstrate that three of these MAbs engage unique epitopes in the S glycoprotein, providing for the development of a broad-spectrum therapeutic that protects young and senescent mice from lethal homologous and heterologous challenge. A cocktail of these antibodies would likely provide robust protection from lethal SARS-CoV infection in humans.
MATERIALS AND METHODS
Viruses and cells.
The generation and characterization of the recombinant infectious clones (ic) of Urbani—icCUHK-W1, icGZ02, icHC/SZ/61/03, icA031G, and icMA15—have been described previously (
33,
37). Briefly, the Urbani spike gene in icUrbani was replaced by the various spike genes of CUHK-W1, GZ02, HC/SZ/61/03, and A031G. All recombinant icSARS-CoV strains were propagated on Vero E6 cells in Eagle's minimal essential medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum (HyClone, Logan, UT), kanamycin (0.25 μg/ml), and gentamicin (0.05 μg/ml) at 37°C in a humidified CO
2 incubator. All work was performed in a biological safety cabinet in a biosafety level 3 laboratory containing redundant exhaust fans. Personnel were equipped with powered air-purifying respirators with high-efficiency particulate air and organic vapor filters (3M, St. Paul, MN), wore Tyvek suits (DuPont, Research Triangle Park, NC), and were double gloved.
Human MAbs.
Human MAbs against SARS-CoV were generated as described previously (
47). The MAbs were initially screened for their capacity to bind to SARS-CoV S-expressing cells and were subsequently tested for their ability to neutralize the Frankfurt isolate of SARS-CoV (GenBank accession number AY310120). A panel of 23 SARS-CoV S-specific MAbs and a control MAb (D2.2) specific for diphtheria toxin were used for further study.
Neutralization assay.
Neutralizing titers were determined by either a microneutralization assay or a plaque reduction neutralization titer assay (
37). For the microneutralization assay, MAbs were serially diluted twofold and incubated with 100 PFU of the different icSARS-CoV strains for 1 h at 37°C. The virus and antibodies were then added to a 96-well plate with 5 × 10
3 Vero E6 cells/well and 5 wells per antibody dilution. Wells were checked for cytopathic effect (CPE) at 4 to 5 days postinfection, and the 50% neutralization titer was determined as the MAb concentration at which at least 50% of wells showed no CPE. For the plaque reduction neutralization titer assay, MAbs were serially diluted twofold and incubated with 100 PFU of the different icSARS-CoV strains for 1 h at 37°C. The virus and antibodies were then added to a 6-well plate with 5 × 10
5 Vero E6 cells/well in duplicate. After a 1-h incubation at 37°C, cells were overlaid with 3 ml of 0.8% agarose in medium. Plates were incubated for 2 days at 37°C and then stained with neutral red for 3 h, and plaques were counted. The percentage of neutralization was calculated as [1 − (number of plaques with antibody/number of plaques without antibody)] × 100. All assays were performed in duplicate. Importantly, a good correlation has been noted between the two assays (data not shown).
Inhibition of binding of SARS-CoV spike glycoprotein to ACE2.
Serial dilutions of MAbs in phosphate-buffered saline (PBS)-1% fetal calf serum were incubated for 20 min at 4°C with 5 μg/ml SARS-CoV S glycoprotein (S1 domain, aa 19 to 713 of the WH20 isolate [99.8% amino acid homology with Urbani]; accession number AY772062) fused to the Fc region of human immunoglobulin (Ig) (Aalto Bio Reagents, Dublin, Ireland). The mixture was added to a single-cell suspension of 4 × 104 ACE2-transfected DBT cells that had been sorted for stable and relatively uniform levels of ACE2 expression. After 20 min, the cells were washed and stained with phycoerythrin-conjugated F(ab′)2 fragments of a goat anti-human Fcγ specific antibody (Jackson ImmunoResearch Laboratories). The percentage of binding inhibition was calculated as [1 − (% positive events for the sample/Bmax)] × 100, where maximum binding (Bmax) is represented by the average of six wells. The concentration of the antibody needed to achieve 50% binding inhibition was calculated with GraphPad Prism software using nonlinear regression fitting with a variable slope.
Detection of human MAbs.
The reactivities of MAbs with native or denatured Urbani S recombinant protein were determined by enzyme-linked immunosorbent assays (ELISA). Briefly, 96-well plates were coated with 1 μg/ml of recombinant Urbani S glycoprotein (NR-686; NIH Biodefense and Emerging Infections Research Resources Repository, NIAID, NIH). Wells were washed and blocked with 5% nonfat milk for 1 h at 37°C and were then incubated with serially diluted MAbs for 1.5 h at 37°C. Bound MAbs were detected by incubating alkaline phosphatase-conjugated goat anti-human IgG (A-1543; Sigma) for 1 h at 37°C and were developed by 1 mg/ml p-nitrophenylphosphate substrate in 0.1 M glycine buffer (pH 10.4) for 30 min at room temperature. The optical density (OD) values were measured at a wavelength of 405 nm in an ELISA reader (Bio-Rad model 680).
Competition for binding to SARS-CoV S glycoprotein.
MAbs were purified on protein G columns (GE Healthcare) and biotinylated using the EZ-Link NHS-PEO solid-phase biotinylation kit (Pierce). An ELISA was used as described above to measure the competition between unlabeled and biotinylated MAbs for binding to immobilized SARS-CoV S glycoprotein. Unlabeled competitor MAbs were added at 5 μg/ml. After 1 h, biotinylated MAbs were added at a limiting concentration (0.1 μg/ml) that was chosen to give a net OD in the linear part of the titration curve, allowing the inhibitory effects of the unlabeled MAb to be quantitated. After incubation for 1 h, the plates were washed, and the amount of biotinylated MAb bound was detected using alkaline phosphatase-labeled streptavidin (Jackson ImmunoResearch). The percentage of inhibition was calculated with the means of triplicate tests as (1 − [(ODsample − ODnegative control)/(ODpositive control − ODnegative control)]) × 100.
Escape mutant analysis.
Thirty micrograms of a neutralizing antibody was incubated with 1 × 10
6 PFU of icGZ02 for 30 min at room temperature in a 0.25-ml volume and was then inoculated onto six-well dishes containing 1 × 10
6 cells. After a 1-h incubation, the inoculating virus was removed, and 1 ml of medium containing 30 μg of the appropriate antibody was added to the culture wells. The development of CPE was monitored over 72 h, and progeny viruses were harvested at ∼25 to 50% CPE. Antibody treatment was repeated two additional times, and more-rapid CPE was noted with each passage. Passage 3 viruses were plaque purified in the presence of a MAb, and neutralization-resistant viruses were isolated and designated GZ02-230 and GZ02-109-1 and -2. The S genes of individual plaques were sequenced as previously described (
37). The neutralization titers for wild-type and MAb-resistant viruses were determined as described above.
Structural analyses.
The crystal structure coordinates of the SARS-CoV RBD interacting with the human ACE2 receptor (PDB code 2AJF) (
21) were used as a template to generate each set of mutations using the Rosetta Design Web server (
http://rosettadesign.med.unc.edu/ ). In each case, the SARS-CoV RBD structure was analyzed by using the molecular modeling tool MacPyMol (DeLano Scientific) to determine which amino acid residues were proximal to the amino acid being targeted for replacement. Briefly, each amino acid to be altered was highlighted, and all other amino acid residues within an interaction distance of 5 Å were identified. Using the Rosetta Design website, the amino acid replacements were incorporated, and all amino acid residues within the 5-Å interaction distance were relaxed to allow the program to repack the side chains to an optimal energetic state. This process was repeated with each mutation and series of mutations. Ten models were generated for each set of mutations, and the best model was selected based on the lowest energy score and was further evaluated using Mac Pymol. In all cases, the lowest energy scores were identical for several of the predicted models, suggesting an optimal folding energy of the chosen model.
Passive immunization.
Female BALB/cAnNHsd mice (age, 10 weeks or 12 months; Harlan, Indianapolis, IN) were anesthetized with a ketamine (1.3 mg/mouse)-xylazine (0.38 mg/mouse) mixture administered intraperitoneally in a 50-μl volume. Each mouse was intranasally inoculated with 106 PFU (icUrbani, icGZ02, or icHC/SZ/61/03) or 105 PFU (icMA15) of icSARS-CoV in a 50-μl volume.
In experiments 1 and 2 (Table
1), 12-month-old mice were injected intraperitoneally with 25 or 250 μg of various human MAbs (D2.2, S109.8, S227.14, or S230.15) in a 400-μl volume at 1 day prior to intranasal inoculation with 10
6 PFU of the different icSARS-CoV strains (
n = 3 per MAb per virus per time point). In experiment 3 (Table
1), 12-month-old mice were injected with a cocktail of S109.8, S227.14, and S230.15 (containing 83 μg of each MAb) with a total concentration of 250 μg MAb in 400 μl at 1 day prior to inoculation with 10
6 PFU icHC/SZ/61/03 (
n = 3 per time point). In experiment 4 (Table
1), 10-week-old mice were injected with 250 μg of S230.15 at 1 day prior to inoculation with 10
6 PFU of icHC/SZ/61/03 (
n = 4). In experiment 5 (Table
1), 10-week-old mice were injected with 25 μg of D2.2, S109.8, S227.14, or S230.15 at 1 day prior to inoculation with 10
5 PFU of icMA15 (
n = 3 per MAb per time point). In experiment 6 (Table
1), 12-month-old mice were injected with 250 μg of S230.15 at −1, 0, 1, 2, or 3 days after inoculation with 10
6 PFU icGZ02 (
n = 5 per treatment per time point). All animals were weighed daily, and at 2, 4, or 5 days postinfection, serum and lung samples were removed and frozen at −70°C for later determination of viral titers by plaque assays. Lung tissue was also removed for histological examination on day 4 or 5 depending on whether animals had to be euthanized due to >20% weight loss.
All mice were housed under sterile conditions in individually ventilated Sealsafe cages using the SlimLine system (Tecniplast, Exton, PA). Experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee at the University of North Carolina, Chapel Hill.
Virus titers in lung samples.
Tissue samples were weighed and homogenized in 5 equivalent volumes of PBS to generate a 20% solution. The solution was centrifuged at 13,000 rpm under aerosol containment in a tabletop centrifuge for 5 min; the clarified supernatant was serially diluted in PBS; and 200-μl volumes of the dilutions were placed on monolayers of Vero cells in 6-well plates. Following a 1-h incubation at 37°C, cells were overlaid with a medium containing 0.8% agarose. Two days later, plates were stained with neutral red, and plaques were counted.
Histology.
All tissues were fixed in 4% paraformaldehyde in PBS (pH 7.4) prior to being submitted to the Histopathology Core Facility (University of Norrth Carolina, Chapel Hill) for paraffin embedding, sectioning at a 5-μm thickness, and hematoxylin and eosin staining. Lung pathology was evaluated in a blinded manner.
DISCUSSION
SARS-CoV is a newly emergent human respiratory pathogen that caused a major outbreak in community settings around the world in 2002 to 2003. Several laboratory-acquired cases have been reported with subsequent spread of the disease into communities, resulting in additional outbreaks in 2003 to 2004 (
25,
31). Since these epidemics, no human cases have been reported, and epidemic human strains are believed to be extinct. However, several SARS-CoV strains have been sequenced from possible zoonotic reservoirs, including palm civets, raccoon dogs, and bats, and new human strains will likely evolve from these reservoirs (
11,
23). In support of this hypothesis, we have shown that many of these animal strains encode an S glycoprotein that can utilize human ACE2 for docking and entry (
37). Additionally, the SARS-CoV infections involving researchers underscore the need for medical countermeasures for postexposure prophylaxis. Therefore, protection against zoonotic and early human-to-human transmission, especially in more vulnerable elderly populations, should receive high priority (
3,
5,
36).
The goal of this study was to identify and characterize cross-neutralizing human MAbs that efficiently neutralize an extensive panel of variant SARS-CoV isolates bearing S glycoproteins from both the human and the zoonotic phases of the epidemic (
37). For the first time, we evaluated the use of human MAbs in the prevention and treatment of lethal homologous and heterologous SARS-CoV infections in murine models. Uniquely, this study not only carefully mapped the cross-reactive neutralization effectiveness of a large panel of human MAbs against human and zoonotic isolates but also identified at least three MAbs that recognize unique epitopes and neutralize all SARS-CoV isolates tested in vitro and in vivo. A similar strategy has recently been used to identify cross-reactive protective human MAbs against influenza virus H5N1 (
41).
Several studies have shown that the generation of neutralizing antibodies against SARS-CoV is a major component of protective immunity (
47,
48). A few studies have focused on evaluating the cross-neutralizing potential of human and murine antibodies and have measured cross-neutralization indexes using a small number of pseudotyped viruses bearing S glycoproteins from SARS-CoV isolates throughout the epidemic (
12,
56). These studies have demonstrated mixed results ranging from no cross-neutralization to enhanced infection and even robust cross-neutralization, complicating the interpretation of data.
Given the variable responses reported with pseudotyped viruses (
30,
55), we used an isogenic panel of recombinant SARS-CoVs bearing variant human epidemic and zoonotic S glycoproteins to evaluate the role of S glycoprotein heterogeneity in neutralization by MAbs (
37). Using this panel of viruses, we identified 23 human MAbs that effectively neutralized one or multiple SARS-CoV S isolates, dividing the antibodies into six distinct neutralizing categories. More importantly, we identified four MAbs that efficiently neutralize both human and zoonotic SARS-CoV isolates. MAb S230.15 has recently been shown to have cross-neutralizing in vitro and in vivo activities against GD03-S and SZ16-K479N recombinant viruses (
57); however, we expanded on these studies to include several more MAbs and three more SARS-CoV spike isolates as well as mapping of the putative epitopes.
In agreement with previous studies, sequence analysis, competition studies, and inhibition of SARS-CoV S binding to ACE2 showed that the majority of our panel of MAbs recognized epitopes within the RBD, with the exception of two MAbs (S132 and S228.11, constituting group I) that likely recognize epitopes in the N terminus of the S1 domain. The majority of the human MAbs (groups III and IV) likely recognize a set of overlapping conformational epitopes, since reactivity was lost by denaturation of the antigen, as seen with the hepatitis B virus surface antigen (
38).
At least three of the cross-neutralizing MAbs (S109.8, S230.15, and S227.14) were mapped within the RBD. Since S230.15 and S227.14 compete with S215.17, all the cross-neutralizing antibodies likely map in the RBD. The epitope of MAb S230.15 likely overlaps with the epitopes that were recognized by 80R and m396, since at least one amino acid (aa 487) was identified as an important residue both for S230.15 and for 80R and m396 (
15,
32,
43). MAb S227.14 was shown to recognize an epitope that partially overlapped with the S230.15 epitope but is likely distinct from the 80R and m396 epitopes. Recognition of partially overlapping but distinct epitopes has also been reported for other viruses, including human immunodeficiency virus type 1 and hepatitis A virus (
18,
29). Although MAb S109.8 likely recognizes an epitope within the RBD, the site is located away from residues in direct contact with ACE2. The mechanism of neutralization by S109.8 is unknown, but we hypothesize that binding of S109.8 to the RBD may result in either structural changes that affect its binding to ACE2, blocking of structural changes in the RBD needed for efficient binding to ACE2, or steric hindrance during ACE2 binding, as has been described for poliovirus, avian sarcoma-leukosis virus, and human papillomavirus (
7,
53,
54).
The locations and mechanisms of neutralization for the other MAbs identified in this study remain unknown, underscoring the need for more-detailed analysis of escape mutants and the efficacy of protection in vivo.
We successfully used escape mutant analysis to identify key residues involved in the neutralizing activities of two cross-neutralizing human MAbs. Escape variant analysis has been used previously to identify the importance of P462 in the neutralization of SARS-CoV by MAb CR3014 (
46) as well as to identify epitopes for various other viruses, including influenza virus (
17). Although escape mutants can be helpful in identifying important neutralizing residues, they compromise the use of the MAb as prophylaxis or treatment. Previous studies with human MAbs neutralizing SARS-CoV and rabies virus have shown that the use of a MAb cocktail may circumvent this problem (
2,
46). Our data show that individual MAb escape mutants can be efficiently neutralized by the other cross-neutralizing MAbs and that these MAbs should therefore be used as a cocktail to prevent the generation of escape mutants. This application is especially useful given that we were unsuccessful in isolating neutralization escape mutants against one of the MAbs (S227.14).
The use of neutralizing MAbs in passive immunizations has been well established (
27,
39). This form of immunization has the advantage of providing immediate protection in the absence of a humoral immune response and may be especially advantageous for elderly populations, since they show increased morbidity and mortality caused by infectious diseases in general and SARS-CoV in particular (
1,
4,
5,
44). This is generally accepted to be due to a compromised senescent immune system (
1,
44), as demonstrated by the limited protection against a heterologous SARS-CoV in vaccinated aged mice (
8). Therefore, the use of MAbs shown to cross-neutralize human and zoonotic SARS-CoV strains may be an attractive alternative for use in the elderly as well as in laboratory personnel after accidental exposure.
Several studies have shown that MAbs and polyclonal sera can effectively protect young mice from homologous SARS-CoV replication in the lung (
42,
43). These models evaluated virus replication only in the absence of clinical disease, and the aged-mouse model recapitulating human SARS-CoV infection allows a more relevant evaluation of efficacy (
34). We recently developed lethal models of SARS-CoV bearing variant S glycoproteins that recapitulated the age-related highly pathogenic phenotype observed in acute SARS cases in humans (
37). Aged mice have been used to test whether immune serum protects from weight loss, reduces viral titers, and prevents histopathologic changes in the lungs (
51). However, to date no MAbs have been tested in aged-animal models of SARS-CoV, especially following heterologous challenge with lethal viruses. These models have the advantage of testing treatment regimens in the most vulnerable populations, the elderly, as well as generally evaluating the efficacy of immunoprophylactic therapy in this population.
In the present study, we tested the abilities of three broad-spectrum human MAbs to cross-protect against challenge with one homologous and three heterologous SARS-CoV isolates. We showed complete protection against clinical disease and virus replication in the lungs of mice treated with S227.14 or S230.15 and challenged with a late- or early-phase human isolate. Interestingly, animals were also protected against clinical disease after challenge with the palm civet isolate despite the presence of high viral titers in the lung. Although animals were not completely protected against viral replication on day 2 postinfection, no virus could be detected by day 5, thereby reducing the chance of transmission. A cocktail of multiple MAbs was as capable of protecting against lethal challenge as the most potent individual MAbs, providing a strategy for minimizing the emergence of MAb escape mutants (
2,
46).
Aged animals were less effective at clearing heterologous virus than were the younger animals. This may be due to an impaired innate immune system in the elderly (
1) or to differences in the efficiency of the human MAbs at reaching the lung mucosa and neutralizing the virus. Bidirectional IgG transport across epithelial barriers is mediated in part by the major histocompatibility complex class I-related Fc receptor, which has been shown to be downregulated by age (
9). To our knowledge, this is the first study to compare and document differences in passive immunization efficacy between young and aged animals, potentially documenting an important clinical consideration in the treatment of the elderly. The effect of age on the efficacy of passive immunizations should be studied in further detail, perhaps by evaluating antibody efficacy in progressively older animals challenged with lethal viruses.
Ideally, these antibodies could also be used in postinfection treatment of SARS-CoV. MAb S230.15 could effectively protect against or reduce the clinical symptoms of SARS-CoV infection only when mice were treated with the MAb 1 day prior to challenge or on the day of challenge. This was particularly surprising because similar treatments were very successful with H5N1 influenza virus infections in mice (
41). However, we used a much higher dose of SARS-CoV to challenge the mice, and the SARS-CoV infection models are much more acute than the model of influenza virus H5N1 infection in mice. In addition, the murine Fc receptor has a relatively low affinity for the human Fc portion, which may also explain the difference in protection between S230.15 and S227.14 in young mice challenged with the highly lethal MA15 virus. Therefore, future experiments to assess the role of the Fc portion in the effector neutralizing potential of the MAbs may be necessary. Interestingly, mice treated at 2 or 3 days postchallenge did reduce viral loads by clearing virus replication by day 4, but these animals were not protected against clinical disease. These data suggest that the clinical disease course is set within the first 24 h of infection, since mice were not protected with MAbs after this time and died even after virus had been cleared. This is in agreement with observations of human SARS cases, where a large subset of cases showed clinical disease after virus clearance (
14).
Our data show that cross-reacting MAbs exist that can efficiently neutralize all human and zoonotic SARS-CoV isolates reported to date. It is interesting that broadly neutralizing MAbs were isolated at a late time point after infection (2 years), suggesting a long-lasting maintenance of potent SARS-CoV-neutralizing MAbs in the B-cell memory repertoire, which would protect survivors against epidemic and zoonotic SARS-CoV emergence. While escape mutants could be generated against some of the MAbs, these could still be neutralized by the other MAbs, stressing the importance of combination therapy. In addition, we showed that escape mutant analysis coupled with a time-ordered panel of outbreak isolates provides a novel set of reagents with which to map neutralizing epitopes. These methods can potentially be used to identify and characterize cross-neutralizing epitopes of novel emerging viral pathogens. Cross-neutralizing epitopes will be important targets for the development of a vaccine that protects against the reemergence of SARS-CoV from its zoonotic reservoir. These MAbs are attractive candidates for prophylactic treatment of SARS-CoV infection, and further development should include testing in larger-animal models of SARS-CoV infection such as ferrets and nonhuman primates, including transmission models. We also showed the importance of testing multiple heterologous SARS-CoV strains and the use of robust animal models with multiple readouts, e.g., virus replication, pathology, clinical signs, and mortality. These models will be essential for successful vaccine development in aged populations. Finally, understanding the mechanisms of neutralization may provide insight into the SARS-CoV entry process.