Two different types of vaccines have been developed to prevent disease and limit primary infection with EBV. Soluble EBV gp350 reduced the rate of infectious mononucleosis by 78% in young adults 
. Alternatively, induction of cellular immunity to EBNA-3 has been proposed to limit the events occurring immediately after primary infection including virus replication in the throat and the expansion of virus-infected B cells 
. Prior studies have shown that EBNA-3 epitopes are primary targets for EBV-specific CTLs in healthy persons, and therefore an EBV vaccine containing EBNA-3 epitopes has been proposed 
. A peptide corresponding to EBNA-3A elicited peptide-specific T cell responses in EBV-seronegative human volunteers; 4 of 4 seronegative volunteers seroconverted to EBV asymptomatically, while 1 of 2 placebo recipients infected with EBV developed infectious mononucleosis 
. Since rhesus LCV is considered one of the best animal models for EBV infection, we compared rhesus LCV soluble gp350 with VRPs expressing gp350 or VRPs expressing a combination of gp350, EBNA-3A and EBNA-3B. Animals received three doses of the vaccines at 0, 4, and 12 weeks. We found that rhesus LCV soluble gp350 induced better protection against challenge virus than VRPs expressing a combination of gp350, EBNA-3A and EBNA-3B.
Animals vaccinated with soluble gp350 produced the highest levels of antibody to the glycoprotein and these levels were higher than those seen in monkeys naturally infected with rhesus LCV. Prior studies have shown that antibody to gp350 is likely the predominant component of neutralizing antibody to EBV 
. In addition gp350 induces antibody-dependent cellular cytotoxicity which may also be important in controlling EBV infection 
. Animals vaccinated with VRP expressing gp350, or VRPs expressing gp350, EBNA-3A, and EBNA-3B developed lower levels of antibody to gp350 and had less protection against acute infection than animals that received soluble gp350. Thus, the high levels of antibody to gp350 are likely important for protection against acute infection with rhesus LCV.
We compared soluble rhesus LCV gp350 with VRPs expressing gp350 with the expectation that expression of the viral glycoprotein in cells infected with VRPs might enhance the immunogenicity of gp350 beyond its ability to induce antibody. Animal studies have shown that neutralizing antibody to gp350 alone does not always correlate with protection from disease. When cottontop tamarins were vaccinated with replication-defective adenovirus expressing gp350, non-neutralizing antibody to gp350 was induced, but the animals were protected against lymphoma 
. In contrast, when cottontop tamarins were vaccinated with gp350 in liposomes, high titers of neutralizing antibodies were induced, but the animals were not always protected from lymphoma 
. These studies showed protection from development of lymphoma, rather than protection from infection. Immunization of common marmosets with gp350 in alum resulted in neutralizing antibodies in some animals, but protection from infection (defined by absence of seroconversion after challenge) did not correlate with the presence of neutralizing antibodies 
. Somewhat surprisingly we found that rhesus LCV soluble gp350 induced better protection against challenge virus than VRP expressing gp350. Animals vaccinated with VRP expressing gp350 had antibody to the glycoprotein at levels comparable to animals naturally infected with rhesus LCV; however, the levels were significantly lower than in animals vaccinated with soluble gp350.
Animals vaccinated with alphavirus VRPs expressing EBNA-3A and EBNA-3B developed CD4 and CD8 cell responses to these proteins, while those vaccinated with VRPs expressing gp350 did not have detectable cellular responses to the glycoprotein. It is possible that the different methods used to present these antigens (LCLs naturally expressing EBNA-3A and EBNA-3B versus cells infected with MVA expressing gp350) could be responsible for these differences. Alphavirus VRPs target dendritic cells, which are highly efficient antigen presenting cells, and are effective for inducing cellular immunity 
. Prior studies in humans show that EBV EBNA-3A, EBNA-3B, and EBNA-3C are the main targets of CD8 T cells in humans, while EBV EBNA-1 is the principal target of CD4 T cells (reviewed in 
). While EBV gp350-specific CD8 T cells have been detected in patients during infectious mononucleosis 
and gp350-specific CD4 T cells have been detected in healthy EBV carriers 
, the level of these T cells has not been quantified relative to those against EBNA-3. In general, the level of T responses to structural proteins is generally lower than that to latent proteins in healthy EBV carriers (reviewed in 
After challenge of animals with rhesus LCV, animals vaccinated with soluble rhesus LCV gp350 had the best level of protection based on levels of rhesus LCV DNA or RNA in the blood and lower rates of seroconversion. While animals that received VRP-gp350, VRP-EBNA-3A, and VRP-EBNA-3B had the next best level of protection from challenge and might have better protection from reactivation, than those receiving the other vaccine candidates, we could not test for protection against reactivation with the small number of animals in the current study. Although soluble gp350 induced the highest levels of antibody to gp350 and the best protection from acute infection, addition of potent EBV-specific T cell responses in combination with high levels of antibody might enhance the effectiveness of an EBV vaccine.
Although an ideal vaccine would protect from infection with EBV, a vaccine that reduces the EBV DNA load might also be useful. The EBV DNA load is a predictor for development of certain EBV-associated malignancies 
. EBV DNA is increased in the blood of transplant recipients prior to the development of EBV post-transplant lymphoproliferative disease 
, and rituximab which lowers the viral load in the blood likely reduces the rate of post-transplant lymphoproliferative disease 
. Patients with primary EBV infection after transplantation have high viral loads, and a 24-fold increased risk of post-transplant lymphoproliferative disease compared with seropositive transplant recipients 
. Similarly, patients with HIV who progressed to B cell lymphoma had elevated levels of EBV in PBMCs and the level increased several months before developing lymphoma 
. In order to determine if our vaccine reduced the level of rhesus LCV DNA in the blood, we developed a more sensitive assay for detection of viral DNA. With this assay we found that animals vaccinated with soluble gp350 that became infected with rhesus LCV after challenge had lower levels of rhesus LCV DNA in PBMCs at 23 and 34 months compared with animals that received vaccine control (PBS). Taken together these finding suggest that an EBV vaccine that reduces the viral load after infection might also reduce the risk for development of certain EBV-associated malignancies.
In summary, our findings indicate that a subunit vaccine that induces primarily humoral, rather than cellular immunity can result in a low virus load in animals that develop breakthrough infection after challenge with wild-type virus. At 23 months after challenge, animals vaccinated with soluble gp350 that became infected with rhesus LCV had ≥100-fold lower levels of rhesus LCV DNA in PBMCs than those vaccinated with VRP-gp350, or the combination of VRP-gp350, VRP-EBNA-3A, and VRP-EBNA-3B. Rhesus LCV DNA was still lower in PBMCs from animals vaccinated with soluble gp350 at 34 months after challenge compared with animals that received PBS. Thus, antibodies to a viral glycoprotein before challenge likely alter the primary infection in such a way as to result in a lower viral load years later. While the largest EBV subunit vaccine study performed to date showed that soluble gp350 protected against infectious mononucleosis, breakthrough infection still occurred; however, the authors did not report on the level of EBV DNA in the blood after breakthrough infection 
. Based on our data, as well as observations of EBV DNA in PBMCs in certain malignancies, future EBV vaccine studies should test the ability of the vaccine to reduce viral loads in persons that become infected.