Epstein-Barr virus (EBV) is a common herpesvirus asymptomatically carried by over 90% of the human population. In the vast majority of infected individuals, the virus persists in B-cells without causing any symptomatic disease 27,28. However, in certain circumstances, such as in immunocompromised states, this virus can lead to life threatening diseases including malignancies 29–31. Unfortunately, several fundamental aspects of the biology of EBV and its association with human diseases remain poorly understood. The lack of a suitable animal model for EBV infection has contributed to this limitation. Recent evidence suggests that rabbits can be susceptible to EBV infection, and can be used to study long term EBV infection in vivo 21,22,32. In these animals, the virus persist asymptomatically for months. These features are similar to latent EBV infection seen in healthy human carriers 6,20. When infected rabbits are immunosuppressed, peripheral EBV load increases from undetectable to several thousand copies 21. Additionally, EBV infected rabbits have been shown to mount a strong humoral immune response against the virus 33. Using this rabbit model, we report, for the first time, the early events during primary EBV infection in vivo.
Intravenous introduction of EBV resulted in all animals contracting the virus. In contrast to the immunocompetent animals, the spleen from infected immunosuppressed group, showed marked splenomegaly and visible inflammatory nodules 21,32. On histology, the enlarged spleens showed evidence of extensive infection and destruction of the splenic architecture. It is likely that the weakened immune system allowed infected cells to undergo rapid proliferation. As a result, large number of EBV infected cells contributed to marked increase in the size of the spleen. Furthermore, of the three immunocompromised groups which were infected, the animals that were immunosuppressed for one week followed by EBV infection (CsA > EBV) had the most extensive infection, reminiscent of what has been observed in patients with congenital immunodeficiencies, such X-linked lymphoproliferative disease (XLP) 34. These patients typically develop EBV-associated lymphoproliferative diseases on primary infection and die early in childhood 35,36. In addition to the extensive infection, EBV infected cells in immunocompromised animals, were transcriptionally active and expressed a range of viral proteins. Despite the concomitant presence of infected cells expressing latent and lytic viral markers in the same follicle, co-expression of lytic and latent proteins was not observed at the individual cell level 7. This suggests that during early infection in vivo, at any given point, the cells are either in latent or lytic cycle, but not both. Another interesting observation was that, latently infected cells were not synchronous, and cells in all the latency programs (0–3) were observed. This is in contrast to long-term persistence infection in healthy seropositive individuals where the virus persists in memory B-cells in latency 0 37. It is noteworthy that some infected cells were distinctly large, and frequently multinuclear. These Reed-Sternberg (RS) like cells were typically LMP1 positive. Previous in vitro studies have reported that LMP1 could mediate the development of these multi-nucleated cells 38–40. In heavily infected cases, the loss of well-defined B-cell areas in the spleen, indicates that the increased burden of EBV infection can result in the breakdown of the well-defined follicular architecture. As a result, scattered, rather than clustered, B cells characterized the nearly destroyed follicle. Compared to their non-infected counterparts, infected B cells showed weaker CD79a, but higher IgM expression. This suggests that, EBV infection may alter the expression of some important B cell markers such as CD79a and IgM 41–44. The infected cells also seemed to reside primarily in the follicles, since splenic follicles are crucial site for B cell development. The majority of the infected cells inside the follicles were B cells. The infected cells expressed various B cell markers, including IgM, CD79a, CD20 and CD21. However, the intensity of staining for theses marker varied. We hypothesize that the infected cells were initially restricted to follicles, but as the infection spreads in situ, the normal structure of the follicles is destroyed. Although infected cells were abundantly present inside the follicles, they did not express the GC marker BCL6. This data is consistent with a previous study which reported that EBV infected cells can expand in the GCs, but do not necessarily participate in the GC reaction 45. It is possible that EBV infected cells increase in number without undergoing somatic hypermutation 23.
Another key observation was that most of the small EBER+ cells were found to be proliferating (Ki-67+), whereas the large LMP1+ cells were non-proliferating and p53+. This was somewhat surprising, as LMP1 is known to be an essential viral protein involved in cell proliferation and transformation. In epithelial cells, it has been reported that expression of LMP1 can result in p53 accumulation, but p53-mediated apoptosis is inhibited by LMP1 induced expression of anti-apoptotic factors such as BCL-2 and A20 46–48.
Taken together, we propose that in primary EBV infection in vivo, the virus targets naïve B-cells in secondary lymphoid organs, such as the spleen, become blast cells and undergo proliferation. These infected cells reach the follicles and occupy the marginal zone. Once in the marginal zone, some infected cells undergo productive cycle, shedding new virus particles, whilst others go through latent cycle expressing various latent genes. Among the cells expressing latent proteins, a portion of these cells undergo proliferation. While the infected cells are in the marginal zone, the GC remains intact and shows increased expression of BCL6 and Ki67. As the infection begins to spread through the mantle zone and further towards the GC, it begins to disrupt the structure of the GC and the size of the GC diminishes. These infected cells show increased expression of p53, which is not expressed in non-infected cells. As the number of EBV infected cells in the follicle increases, the number of BCL6+ GC cells gradually decline. Eventually, the entire follicle becomes filled with EBV infected cells with no intact GC. The EBV infected follicles expand, losing their well demarcated cellular zones. Multiple EBV infected follicles eventually fuse, resulting in the destruction of the typical follicular architecture. Based on our observations, we believe that follicles play an important role in EBV infection and proliferation, but GCs do not appear to be involved. However, further research is needed to verify this hypothesis. The rabbit model of EBV infection described here, will certainly help in further delineating the steps involved in primary EBV infection and the development of EBV induced malignancies. Additionally, this model may prove to be essential for vaccine and therapeutic developments against EBV.