Evidence implicating Epstein–Barr virus (EBV) in the pathogenesis of multiple sclerosis has been increasing over the last 30 years, and this issue of Brain contains an article and correspondence addressing the issue of EBV infection in the brains of people with multiple sclerosis. To appreciate how EBV might contribute to the development of multiple sclerosis, it is necessary to understand some key aspects of EBV infection.
EBV is a human herpesvirus with the unique ability to infect, activate and persist in latent form in B lymphocytes for the duration of the infected individual’s lifetime. The virus infects naïve B cells, driving them out of the resting state into proliferating lymphoblasts and exploiting the normal pathways of B cell differentiation so that the B blasts become memory cells (Thorley-Lawson and Gross, 2004). In normal B cell differentiation, naïve B cells are activated by antigen through the B cell receptor, and by CD4+ T cell help through the CD40 receptor, so that they proliferate and progress through a germinal centre reaction in lymphoid follicles to become antigen-specific memory B cells. Remarkably, EBV expresses two proteins, latent membrane protein 2A (LMP2A) and LMP1, which mimic the antigen-activated B cell receptor and the activated CD40 receptor, respectively (Mancao and Hammerschmidt, 2007; Rastelli et al., 2008). These drive naïve B cells through a germinal centre reaction, independent of antigen and T cell help, to become latently infected memory B cells (Thorley-Lawson and Gross, 2004). Terminal differentiation of the latter into plasma cells initiates the lytic replicative cycle of EBV with the production of infectious virus (Laichalk and Thorley-Lawson, 2005). EBV is normally kept under tight control by EBV-specific immune responses, especially cytotoxic CD8+ T lymphocytes that eliminate proliferating and lytically infected B cells (Hislop et al., 2007). In the developing world, most children become infected within the first 3 years of life, and EBV seropositivity reaches 100% within the first decade (Rickinson and Kieff, 2001). These early primary infections are almost always asymptomatic. In contrast, in the developed world, up to half of all children remain EBV seronegative at the end of their first decade and subsequently become infected in adolescence or young adulthood (Rickinson and Kieff, 2001). As many as 50% of these delayed primary infections manifest as infectious mononucleosis.
The first evidence for a role of EBV in the development of multiple sclerosis came in 1979 when Fraser et al. (1979) reported that peripheral blood lymphocytes from patients with clinically active multiple sclerosis have an increased tendency to spontaneous in vitro EBV-induced B lymphocyte transformation. Sumaya et al. (1980) reported a higher frequency of EBV seropositivity and higher anti-EBV antibody titres in patients with multiple sclerosis compared with controls. Subsequent studies have shown that patients with multiple sclerosis are almost universally seropositive for EBV (Ascherio and Munger, 2007), suggesting that EBV infection is a prerequisite for the development of the disease. Furthermore, infectious mononucleosis increases the risk of multiple sclerosis (Thacker et al., 2006). Prospective studies have demonstrated that a high titre of immunoglobulin G (IgG) antibodies to EBV nuclear antigen 1 (EBNA1) increases the risk of developing multiple sclerosis (Ascherio et al., 2001; Sundström et al., 2004; Levin et al., 2005).
Since 1981, it has been suggested that the epidemiological features of multiple sclerosis can be explained by late primary EBV infection in adolescence or young adulthood (Warner and Carp, 1981). Indeed, delayed primary infection with EBV can account for the association of multiple sclerosis with higher socio-economic status, the latitudinal variation in prevalence, the effects of migration and the occurrence of clusters and epidemics (Haahr and Höllsberg, 2006).
Initially, the favoured hypothesis for the role of EBV in the pathogenesis of multiple sclerosis was molecular mimicry between EBV and central nervous system (CNS) antigens leading to immunological cross-reactivity and subsequent autoimmune attack on the CNS (Lang et al., 2002). This hypothesis does not require EBV-infected B cells to be present in the CNS because activation of immunological cross-reactivity can occur in peripheral lymphoid organs. In 2003, the novel hypothesis was proposed that multiple sclerosis is driven by the accumulation of EBV-infected autoreactive B lymphocytes in the CNS—these cells make the oligoclonal IgG bands present in the cerebrospinal fluid, produce pathogenic autoantibodies and provide co-stimulatory survival signals to autoreactive T cells that would otherwise die in the CNS by apoptosis (Pender, 2003).
Recently, Serafini et al. (2007) reported that a substantial proportion of the B cells and plasma cells in 21 out of 22 multiple sclerosis brains are infected with EBV. They identified EBV-infected cells using in situ hybridization for EBV-encoded small RNA (EBER-ISH), the gold standard to detect EBV-infected B cells in histological material (Gulley and Tang, 2008), and by immunohistochemistry with antibodies specific for EBV proteins. EBV-infected B cells were not detected in the brains of patients with other inflammatory CNS diseases. The major sites of EBV persistence in the multiple sclerosis brain were meningeal structures resembling B cell lymphoid follicles with germinal centres, which the authors had previously demonstrated to be present in patients with secondary progressive disease (Serafini et al., 2004; Magliozzi et al., 2007). B cells expressing EBV latent proteins were regularly found in the brains of subjects with multiple sclerosis, whereas B cells expressing the EBV early lytic cycle protein, BFRF1, were restricted to acute lesions and meningeal B cell follicles. Their finding that EBV-infected B cells in meningeal follicles express LMP1, but not EBNA2, indicates that, here, EBV deploys the default latency transcription programme, which allows infected B cells to progress through a germinal centre reaction to become latently infected memory B cells with somatically hypermutated B cell receptor and Ig class switching from IgM to IgG production (Thorley-Lawson and Gross, 2004). It also accords with the observation that, in the tonsils of healthy EBV carriers, EBV-infected B cells expressing this viral transcription programme not only are functionally indistinguishable from classical germinal centre B cells but also are physically located in germinal centres (Roughan and Thorley-Lawson, 2009).
However, the demonstration of EBV infection of B lymphocytes and plasma cells in the CNS of people with multiple sclerosis is not easy to replicate. In stark contrast to the results of Serafini et al. (2007), a paper in this issue of Brain by Willis et al. (2009) finds EBV in only 2 of 24 multiple sclerosis brains. They do not detect EBV-infected B cells by EBER-ISH or immunohistochemistry using antibodies against LMP1 or EBNA2 in any of the specimens, but do detect low levels of EBV DNA in two samples by real-time polymerase chain reaction. The authors conclude that EBV infection is rare in multiple sclerosis brain tissue and is unlikely to contribute directly to the pathogenesis. A similar conclusion is reached by Peferoen et al. (2009) who, in correspondence published on-line in association with this issue of Brain, detect nuclear EBER-ISH in only one specimen out of 16 multiple sclerosis brains exhibiting prominent B cell infiltrates. So which position is correct?
Possible explanations for the discrepancy between these findings now reported and those of Serafini et al. are differences in experimental design and interpretation, tissue selection and processing, the degree of CNS B cell infiltration, and the sensitivities and specificities of the techniques used. The inability of Willis et al. (2009) to detect any EBV-infected B cells by EBER-ISH in their multiple sclerosis brain specimens might be explained on the basis that EBER RNA has been degraded by long tissue fixation times in formalin. Most of their specimens come from formalin-fixed autopsy tissue, whereas the positive controls of brain tissue are from biopsies likely to have been in formalin for a shorter time. Another important difference is that meningeal B cell follicles, which were found by Serafini et al. to be major sites of EBV persistence, are not found by either Willis et al. or Peferoen et al. The presence of EBV in the brain in multiple sclerosis, as reported by Serafini et al., is indirectly supported by the study of Cepok et al. (2005) who screened cerebrospinal fluid IgG for reactivity against 37 000 proteins and found the two most frequent reactivities to be against the EBV proteins, EBNA1 and BRRF2.
A major unanswered question is how EBV infection, if indeed it is present in the brain, might lead to the development of multiple sclerosis. Serafini et al. (2007) found a strong correlation between the frequency of CD8+ T cells and EBV-infected B cells in the CNS and proposed that the immune attack in multiple sclerosis is primarily directed against EBV, with bystander damage to the CNS. This scenario does not explain the evidence for a primary role of autoimmunity in the development of multiple sclerosis (Sospedra and Martin, 2005). It also questions why an EBV-directed immune response sufficient to cause CNS damage does not eliminate infected B cells from the CNS, given that EBV-infected B cells of multiple sclerosis patients can be killed by their own EBV-specific CD8+ T cells (Pender et al., 2009). An alternative explanation for the CNS damage is that some EBV-infected B cells if present in the brain produce pathogenic autoantibodies and/or provide co-stimulatory survival signals to autoreactive T cells, which then orchestrate an immune attack on the CNS (Pender, 2003). The autoimmune process itself could foster the EBV-infected autoreactive B cells by releasing CNS antigens and giving CD4+ T cell help, which would complement the B cell receptor and CD40 receptor signalling already provided by LMP2A and LMP1, respectively. This could lead to a vicious cycle wherein EBV-infected autoreactive B cells promote autoimmunity, which in turn promotes EBV infection in the CNS. Whether EBV-infected B cells in the CNS are autoreactive could be addressed by determining if they bind biotinylated CNS antigens in the same way that intrathyroidal germinal centre B cells specifically bind thyroid antigens in autoimmune thyroid disease (Armengol et al., 2001).
Another crucial question is why EBV-infected B cells might accumulate in the CNS in patients with multiple sclerosis but not in healthy subjects. One possibility is decreased CD8+ T cell immunity to EBV (Pender, 2003), which is supported by the finding that patients with multiple sclerosis have a decreased frequency of circulating CD8+ T cells specific for EBV-infected B cells (Pender et al., 2009). The deficiency of EBV-specific CD8+ T cells is most severe in patients with secondary progressive disease, who were found by Serafini et al. (2007) to have the highest frequency of EBV-infected B cells in the brain. Jilek et al. (2008) found a strong CD8+ T cell response to EBV peptides in patients with clinically isolated syndromes, although the CD8+ T cell response to EBV-infected B cells in these patients might have been suboptimal for dealing with the EBV load. Multiple sclerosis patients have an increased CD4+ T cell response to EBNA1 (Lünemann et al., 2006), possibly in compensation for a decreased CD8+ T cell response to EBV.
The possibility that EBV infection in the brain drives the development of multiple sclerosis has profound implications for prevention and treatment (Pender, 2009) and must be addressed by further studies. However, many will conclude that the jury is still out on whether the evidence is sufficient to proceed with detailed investigation of EBV infectivity as the basis for tissue injury in multiple sclerosis.