As a newly identified type of coronavirus, SARS-CoV is a highly virulent disease with a 10% mortality rate that demands immediate international attention. Rapid development of a safe and effective prophylactic vaccine is critical in preventing the devastating effects of a potential episode of SARS resurgence. Prior to the outbreak of SARS-CoV in 2002, two human coronaviruses (HCoV-229E and HCoV-OC43)--responsible for causing 15-30% of common colds--had been discovered. In addition to the two human coronaviruses, there exist several animal coronaviruses for which multiple vaccines have been developed and tested. Thus, in order to develop an effective vaccine for SARS-CoV in these early stages when knowledge about human coronaviruses is scarce, we must study the clinical data from the SARS-CoV outbreak in order to gain insight into the immunology of SARS-CoV infection while also considering the existing vaccine approaches for animal coronaviruses.

Clinical data shows that re-infection with SARS-CoV is rare (unlike other HCoVs) and that many individuals (70-80% of young adults) infected with SARS-CoV recovered completely from acute illness, signifying that humans can develop protection from SARS pathology. (96) These data suggest that development of a SARS-CoV vaccine is feasible.

Animal Models

Researchers have worked with mice models for SARS and were able to elicit SARS-CoV neutralization and protective humoral immunity in mice with a DNA vaccine. Intranasal administration of SARS-CoV in mice leads to virus replication in the lungs and nasal turbinates within 1-2 da ys. (114) Immune protection against viral replication is used as a measure of vaccine efficacy in this model. The murine SARS pathology, however, does not present with sustained infection or high lethality. (135)

Cats and ferrets have also been studied as possible animal models for SARS-CoV infection, although it is unknown if the cat/ferret model faithfully replicates the complexity of the SARS infection in humans. (14)

The rhesus macaque model is a highly translational to humans. Adenoviral vector delivered vaccines encoding SARS-CoV spike protein S1 fragment, membrane protein, and nucleocapsid protein were tested in macaques, producing SARS-CoV-specific T-cell and virus neutralising antibody responses. Further study to determine if these responses are protective remains to be done. (43)

Immune State

There exists an immune state in SARS, as evidenced by the high recovery rate in the absence of effective medical therapy, especially in younger individuals. (The signficantly higher mortality rate in older populations infected with SARS-CoV has been hypothesized to result from a limited ability to generate new B and T cell responses in the elderly.) Furthermore, reinfection by the disease is rare, implying that protection from SARS-CoV is achievable. In addition, there are certain individual who have seroconverted but never reported any illness, suggesting that immunity to infection could be rapid enough in some individuals to prevent the onset of illness. This especially lends hope toward the development of an effective vaccine that would stimulate an immune response and induce memory. Recent studies have reported the successful use of DNA vaccine approaches to prevent SARS-CoV infection in animal models. However, because antibodies can actually exacerbate symptoms, further research must be done to prove safety and efficacy before use in humans. Currently, the main clinical intervention against SARS is the use of steroids to suppress local immune responses.

Correlates of Protection

To develop a vaccine to SARS we must first consider the correlates of protection. The initial SARS-CoV infection occurs primarily in the epithelial cells in the respiratory tract causing interstitial pneumonia. (56) This suggests that a vaccine against SARS-CoV should elicit secretion of immunoglobulin A (IgA), an antibody that prevents penetration of pathogens into the mucosal epithelium, as a first-line of immune defense against the invading virus. Eliciting an IgA response is often difficult but could potentially be achieved with a viral vector vaccine delivery system. However, there is also evidence that cell-mediated immunity is important in the rapid immune response to mucosal infections. (40) Specifically, cytotoxic T cells (CTL) play an important role in clearing infected cells, thereby decreasing the rate of pathogen replication in the mucosal epithelium.

Correlates of Protection: B Cell Immunity

Evidence from animal coronavirus models implicates both humoral and cellular immune responses as vital to the defense against acute SARS infection. In humans, a T cell response to infection would be expected to occur 2-4 days after infection, whereas antibody seroconversion to SARS-CoV occurs around day 10. (56) Antibody responses against several animal coronavirus structural proteins, including Spike glycoprotein (S-protein), nucleocapsid protein (N), membrane protein (M) and small envelope protein (E) have been documented, however, currently only the Spike protein has been confirmed as a major antigen and associated with neutralization of the virus. (58) Furthermore, roughly 40% of market traders in Chinese marketplaces handling animals with coronaviruses have been found to have anti-SARS-CoV antibodies without ever having presented with disease pathology, indicating that immunity to infection can occur rapidly after exposure and that disease is preventable. (46) Accordingly, there have been several attempts to make vaccines targeting the Spike protein in animal coronaviruses. An S1 subunit of IBV Spike protein expressed in fowl adenovirus ( FAV ) vector has been able to induce anti-S1 IgA antibodies and protect against IBV challenge. (61)

Correlates of Protection: T Cell Immunity

T cell immunity has also been demonstrated to be critical to an effective immune response and protection against coronaviruses. Several papers show that CD4+ and particularly CD8+ T cells play a critical role in the elimination of IBV virus during the acute phase and subsequent control of infection. (111) Further evidence supporting the role of T cell immunity (both CD4+ and CD8+) against MHV, BoCV, PEDV, and TCoV (80) , suggest that a vaccine against SARS-CoV must induce a T cell response. Finally, it seems that T cells are responsible for eradicating an existing MHV infection, whereas antibody is mainly involved in reducing viral load during acute infection. (99) It should be noted that some coronaviruses (found in the lungs of SARS patients (90 ) are able to form syncitia. If cell-to-cell viral transmission occurs via syncitia, a T cell response might be necessary to clear the infection. (80)

Vaccine Precautions

Before work begins on identifying SARS-CoV immunogens and subsequent vaccine development we must carefully consider the frequent rate of virus mutation and the apparent pathogenic interaction between the virus and host immune system in the late stages of natural SARS infection.

Vaccine Precautions: Immunopathogenesis

There is mounting evidence that an immune response developed against a natural coronavirus infection can lead to exacerbation of disease pathology. There is clinical data on SARS that suggests that adverse hyperimmunity can contribute to lung damage and disease progression. (96) There is also significant evidence on animal coronaviruses documenting a link between an increase in antibody and an increase in disease pathology. In the case of bovine viral diarrhea virus, there is evidence suggesting that development of a humoral response contributes to "Shipping Fever" (a condition attributed to BCoV-R ). (91) Furthermore, in feline coronoavirus ( FIPV ) infections, when anti-Spike antibodies were delivered via passive immunization or induced by experimental vaccines, enhancement of disease pathology was observed. (129) However, it is likely that mechanisms of SARS immunopathology are different from that of feline coronoavirus. Clinical data suggests that a proinflammatory cytokine release by SARS-CoV infected macrophages causes hyperreaction. Furthermore, there is encouraging evidence that passive immunization with convalescent SARS serum improves, not worsens, the condition of critically ill SARS patients. (132) Therefore, although we must take caution in eliciting an anti-SARS-CoV antibody response, the approach is promising and must be thoroughly explored.

In addition to the concerns of exacerbating disease with antibody, there is increasing evidence that an anti-coronavirus T cell response can lead to neural demyelination. Specifically, a protective CD4 and CD8 natural T cell response against murine hepatitis virus ( MHV ) could be the cause of observed demyelination of the brain and spinal cord following infection. (16)

Therefore, in developing a vaccine, we must test any B cell or T cell epitopes of SARS-CoV for eliciting an adverse immune reaction. However, this should not significantly slow the development of a vaccine since a candidate vaccine including any epitopes implicated in an adverse reaction could be quickly redesigned excluding these unwanted subsets of immunogens.

Vaccine Precautions: Virus Variability

Coronaviruses are extremely variable and capable of jumping species and changing organ tropism and pathology. There are two main theories about the emergence of SARS. First, it is possible that SARS-CoV jumped into humans from one of many animal coronavirus reservoirs. (125) There are several coronaviruses of striking similarity to SARS-CoV. Most notably are those that infect the Himalayan palm civet, hog-badger, and racoon-dog, animals frequently found in Chinese markets. Virus isolates from these animals are 99.8% identical to human SARS-CoV and have a characteristic omission of only 29bp in their genomes. (86) Although species jumping is a rare event, because such animal reservoirs are in constant proximity with humans, this theory is extremely plausible.

A second theory for the emergence of SARS-CoV suggests that the virus was already established in the human population causing mild asymptomatic disease and subsequently converted into a highly virulent form of the virus. Coronaviruses have been shown to change from enteric infection to respiratory, neurotropic or hepatotropic. Furthermore, the feline coronavirus FIPV has been shown to regain fatal virulence after circulating in a host population as relatively harmless strain. (32)

These two theories for the emergence of SARS emphasize the genetic variability of coronaviruses. Coronavirus RNA polymerase is highly errorprone (1 change per every 10,000 bases, on average), resulting in frequent mutations. Moreover, coronavirus polymerase is known to switch RNA templates within host cells, causing significant genomic deletions and even recombination between two coronavirus species in a co-infected host (a viable method of species jumping). (85) For example, the feline coronavirus FCoV type II is apparently the result of a homologous recombination between the feline FCoV type I coronavirus and the canine CCV coronavirus. (53)

The genetic variability of SARS is of great concern in the development of a vaccine. Several mutations in the SARS-CoV genome have already been described. (96) Comparison of 14 SARS genomes from a common source shows a total of 94 amino acid sequence mutations over a period of 2 months in the RNA polymerase, the Spike protein, the membrane nucleocapsid, and other uncharacterized proteins of SARS-CoV. (107) Genetic variation in the Spike protein, responsible for virus binding to cells, causes dramatic changes in virus tropism. Deletions of the Spike protein in enteric porcine TGEV resulted in a new porcine coronavirus with a novel respiratory pathology. (97) Mutations in coronavirus Spike protein have both been shown to produce a new neurotropic virus ( JHM ) from murine and to increase JHM virulence. (92)

Genetic variation of SARS-CoV is of great concern. Because of the ability for coronaviruses to jump species, adjust disease virulence, and rapidly evolve to immune pressure, it is critical that we develop a vaccine for SARS before the virus evolves into various strains and clades. Such a vaccine would need to be broadly protective to cope with antigen variability. Therefore, we must select genetically conserved B cell and T cell immunogens as vaccine targets.

B Cell Immunogens

The major consideration for B cell epitopes are accessibility on the viral surface and the preservation of protein conformation in developing recombinant immunogens for accurate antibody recognition of SARS-CoV. As aforementioned in Correlates of Immunity , antibodies are developed against many coronavirus proteins (S-protein, N-protein, M-protein, and E-protein). However, only the Spike S-protein has been shown to be a significant antigen so far. It has been suggested that anti-Spike antibodies are possibly capable of virus neutralization (58) . Experiments show that antibodies induced in mice against Spike and its soluble N-terminal fragments in a mammalian expression vector neutralize virus proliferation in vitro . (6) Therefore, conserved immunogenic regions of the variable Spike protein could be a promising approach towards development of an anti-SARS-CoV antibody response.

T Cell Immunogens

The significant importance of a T cell response to SARS-CoV has already been established. Human cytotoxic T lymphocytes (CTL) are specific for peptides presented in HLA molecules (human MHC--histocompatibility complex molecules). Pathogen peptides are processed in the cytosol of Antigen Presenting Cells (APCs) via limited proteolytic fragmentation of available proteins, transported to the endoplasmic reticulum where they are bound to HLA molecules. The HLA-peptide complex is the exported to the cell's surface and presented to CTLs. An important factor in this process is the specificity of the HLA molecules for the different peptides. HLA molecules are extremely polymorphic and vary from person to person and race to race. Accordingly, T cell vaccine development is often restricted by HLA types. Therefore, selection of T cell epitopes is primarily governed by epitope conservation, proteosome processing, and HLA selectivity. However, almost all HLA types can be categorized by nine "HLA supertypes"--each supertype selective for sequentially similar peptides (112) . Fortunately, a recent boom in genomics, bioinformatics and antigen immunogenicity and presentation prediction algorithms allows us to predict several immunogenic and conserved epitopes that can be presented in the nine HLA supertypes. Evidence suggests that up to 50% of these immunogens are capable of eliciting a CTL response (84) . Peptide-HLA complexes can then be produced in E. coli or by other expression mechanisms and subsequently tested for immune recognition against SARS survivors. Successful peptide-HLA complexes for the nine HLA supertypes could then be delivered via several vaccine methods. One interesting vaccine approach would be integration of all nine peptide-HLA-supertype complexes into a polytope vaccine design. (122, 123) . Polytope, or polyepitope vaccines have been shown to successfully elicit immune responses to large numbers of epitopes expressed in a single viral or DNA vector. Therefore, work must be done identifying, expressing, and testing T cell epitopes in order to create a safe and effective vaccine that stimulates cell mediated immunity.

Vaccines

There are several vaccine approaches to be considered for SARS. As we learn more about the correlates of immunity (humoral and cell mediated), we should be able to design a safe and effective vaccine. The following are the vaccine approaches to be considered in the attempt to develop a SARS vaccine.

Vaccines: Live

Several vaccines against animal coronaviruses have been developed and tested. One of the most common vaccine strategies for veterinary applications is live vaccines. Live vaccines are composed of live attenuated pathogens, live recombinant vaccines, and heterologous vaccines.

Live attenuated vaccines are viruses whose virulence has been reduced via in vitro culture manipulation (such as changed temperature or chemical modification). These live attenuated viruses replicate in the vaccine recipient without causing the standard disease pathology while still eliciting both cell mediated immunity and antibody response that subsequently recognizes the original virulent pathogen.

Live Attenuated Vaccines

Live recombinant vaccines are similar to live attenuated vaccines in that they originate from the virulent pathogen but are altered to decrease virulence by genomic alterations. Accordingly, live recombinant vaccines induce long-term humoral and cell mediated immune responses.

Heterologous vaccines are pathogens closely related to the virulent pathogen of interest that share common antigens and replicate within the host without causing disease. Like live attenuated and live recombinant vaccines, heterologous vaccines induce a long-term humoral and cell mediated immune response. The first vaccine ever developed was a heterologous vaccine: vaccine virus given to humans is relatively harmless yet causes immune recognition of the highly lethal and contagious variola virus, the pathogen that causes smallpox.

These vaccine strategies have had varied success in attacking animal coronaviruses. There are vaccines that have proven relatively effective: a combination live attenuated and inactivated IBV vaccine had some success in chickens (13) , and a heterologous vaccine ( TGEV ) showed a decrease disease pathology in pigs from PRCV (129) . However, it is important to note that attenuated vaccines against TGEV and IBV have only been shown to reduce disease pathology but fail to prevent infection (104) . Live recombinant vaccines against coronaviruses have been produced (35) , however, because of the genomic variability of SARS-CoV and the possibility of vaccine recombination, this approach cannot eliminate the possibility of virus reversal to its former virulence (64) .

Therefore, because live vaccines could potentially revert to their former virulence, a SARS-CoV vaccine must use a different approach to comply with higher safety standards.

Vaccines: Killed/Inactivated

A safer alternative to live vaccines are killed or inactivated vaccines. Because of the ease and low cost of production, these vaccines, like live vaccines, have been used against animal coronaviruses. Killed and inactivated vaccines are either whole killed vaccines or subunit vaccines.

Whole killed vaccines are made by culturing the pathogen in vitro and subsequently killing them (typically with beta -propiolactone or formaldehyde). After this treatment, the vaccine is unable to replicate and is therefore relatively safe.

Subunit vaccines are used when the known correlates of immunity suggest that immunity is raised against one or a few pathogen antigens. Subunit vaccines are made by culturing large amounts of the pathogen and then purifying for the proteins/antigens of interest.

Unfortunately, although very safe, inactivated vaccines induce a poor, antibody-only immune response. Furthermore, immune response is typically short lived.

These vaccine approaches have been used with some success against animal coronaviruses. A whole killed bovine BCoV vaccine was shown to be safe and induce an antibody response in cows (119) , and an inactivated CCoV dog vaccine was shown to protect pups from viral challenge (101) . However, because T cell immunity is considered so vital to protection from SARS-CoV, a killed or inactivated vaccine has little promise against SARS.

Vaccines: Recombinant Subunit

Recombinant subunit vaccines are immunogenic proteins of virulent organisms that are made by expressing the antigen's gene in an expression vector. Like inactivated vaccines, recombinant subunit vaccines only induce B cell antibody protection against the antigen. Because of the extreme genetic variability of SARS-CoV, only highly conserved antigens can be considered for a recombinant subunit vaccine. Although T cell immunity has been shown to be critical in coronavirus immunity, combining a recombinant subunit vaccine with a vaccine approach that induces T cell immunity should be considered.

Vaccines: Recombinant Vectored

A recombinant vectored vaccine is either the incorporation of antigens into a known virus that generates the desired immune response or the delivery of antigen DNA that is then taken up by host cells and expressed on the surface, inducing an immune response against the vaccine antigens.

Although there have only been few recent attempts to make recombinant vectored vaccines against coronaviruses, DNA or viral vectors are currently favored by coronavirus experts because of their ability to induce a specific humoral and cell mediated immune response. A recombinant flowpox virus containing cDNA of the S1 gene of IBV induced anti-S1 antibodies and neutralizing antibodies, protecting chickens from severe disease in a homologous virus challenge (128) . A DNA vaccine of the N protein of porcine TGEV expressed in a naked mammalian vector induced anti-N antibodies and likely Th1 response but not neutralizing antibodies (78) . Recently, a DNA vaccine of SARS-CoV Spike successfully induced IgG against SARS-CoV in 75% of mice after 3 immunizations (141) .

Recombinant viral vectors or DNA vaccines are supported by many coronavirus experts as the ideal approach since both humoral and cell mediated immunity are critical in protection from SARS. The many similarities between HIV and SARS suggest that a possibly rewarding approach would be a DNA prime vaccine followed by an adenovirus or MVA boost, as is being explored in a HIV vaccine. (34) .

Vaccines: Epitope-Based

An epitope-based vaccine is considered to be one of the best ways to address SARS. Immunogenic epitopes that have been screened against self-recognition can be easily delivered in a DNA, viral, or virus-like-particle vector. One epitope-based peptide vaccine for the S protein of murine coronavirus had some success protecting mice from severe disease (66) . Epitope-based vaccines offer the best method for inducing cross-strain hurmoral and cell-mediated protection via multiple epitope presentation. Furthermore, an epitope-based vaccine would be the best way to elicit a cell-mediated response with the most immunogenic epitopes.

A new type of epitope vaccine approach, termed "polytope" (or polyepitope) seems promising. The idea is simple: if one epitope is good, several epitopes must be even better. Presentation of multiple epitopes via DNA or viral vectors has been shown to be protective in mice. Although there is still much work on polytope vaccines to be done, the approach needs to be further investigated. (121, 122, 123)

Vaccine Delivery: Viral Vector

Live recombinant viral vector vaccines are constructed by inserting DNA for the desired immunogens into a live, infections but non-pathogenic virus that elicits a known immune response in humans. The most common viral vector is vaccinia , first detailed by Moss in 1987.   Based on vaccinia's success, there are now more than 20 different RNA and DNA viruses being tested for their applications as vaccine vectors. The main concern using a viral vector is the potential pathology the usually "harmless" virus can cause in immunocompromised individuals.

Viral Vector Vaccine

Viral vectors have been tested in many animal coronaviruses with promising results. This method of delivery should be considered for its ability to induce a strong immune response typical to the vector.

Vaccine Delivery: Microencapsulated Antigen

Antigen vaccines are a standard method of inducing immune recognition of the desired antigens and corresponding pathogen. The next generation of antigen vaccines are antigens enclosed in a biocompatible, biodegradable polymer. Microencapsulated antigens are phagocytized and hydrolyzed in APCs, which then present the antigen as usual. Microencapsulated antigens can be delivered by any route of administration, including parenteral, oral, and intranasal--a method of inducing the IgA response currently considered a first line of defense against SARS-CoV mucosal infection. Furthermore, microencapsulated antigen has been shown to induce a stronger immune response than naked antigen. (1,2)

Vaccine Delivery: Microencapsulated DNA

DNA vaccines deliv, er DNA of the desired epitopes into host cells where they are then presented on the cell surface for immune scrutiny. DNA vaccines have been shown to be effective against intracellular pathogens (such as a coronavirus) where a CTL immune response is necessary for pathogen clearing (105, 114, 63) . However, a DNA vaccine approach is usually less effective at inducing a B-cell antibody response than protein antigen vaccination. Until recently, DNA vaccines have been designed for intramuscular injection. However, the new technology of microencapsulation of DNA allows for oral or intranasal mucosal vaccination, critical in inducing an IgA response as desired for SARS (9, 36, 131, 62) .

Vaccine Delivery: Virus-Like Particles (VLPs)

Virus like particles typically induce a stronger immune response to selected protein epitopes through an array-like presentation of antigens in a structure similar to that of an authentic virus (102, 73) . Furthermore, VLP delivery vehicles improves uptake of antigen by APCs and enhances T cell immunity, thereby eliminating the need for chemical adjuvants (73) . Importantly, virus like particles are able to present several epitopes simultaneously with mosaic VLPs (67) .

Administration Route

Although most vaccines are given by injection, natural infections do not typically occur this way. The majority of infections enter or infect mucosal surfaces. Unfortunately, current methods of mucosal vaccination (e.g. intranasally) have been shown to induce local mucosal immunity but not distal immunity (108) . However, recent work on intranasal vaccination is promising. It has been shown that vaccine delivery via bare skin or nose can imporive vaccine efficacy due to the presence of a large amount of associated lymphoid tissues and APCs (26, 42) .

Thus, because of the apparent importance of an early IgA response against SARS, intranasal vaccination has the advantage over standard intramural vaccination of inducing the desired local mucosal antibody and cell mediated response. Nonetheless, standard intramural vaccination should be considered if intranasal vaccination proves ineffective at inducing a holistic immune response.