This study focuses on early cellular events following encounter of B cells with EBV. EBV differentially infects B cells such that sub-populations of cells with distinct phenotypic and functional characteristics emerge. While expression of CD23, CD58, and IL6 have been examined individually either after infection with EBV [6, 7, 20, 21] or after transfection of EBV latency gene LMP1 , to our knowledge, this is the first report of identification of sub-populations of B cells based on co-expression of these molecules, marking EBV-infected cells early for different outcomes. The expression pattern CD23hiCD58+IL6- predicts the identity of infected cells destined for proliferation as early as three days after exposure to EBV. Another sub-population of CD23+ B cells, also infected with EBV and expressing EBNA2 and LMP1, expresses IL6 but fails to proliferate.
Earlier investigations have shown that only cells co-expressing Epstein-Barr nuclear antigens and CD23 undergo immortalization . These investigations did not differentiate between the different EB nuclear antigens. Our experiments demonstrate that expression of EBV latency genes, LMP1 and EBNA2, and CD23, are not sufficient for proliferation. While low levels of LMP1 transcripts relative to EBNA2 transcripts in CD23loCD58+ cells suggested lack of transition to high levels of expression of LMP1 as a potential reason for their inability to proliferate, LMP1 protein expression levels in individual cells did not substantiate this possibility. Greater than 80% of CD23loCD58+ cells expressed LMP1 at levels comparable to those observed in CD23hiCD58+ cells. Furthermore, the LMP1 gene product appeared to be a latency protein since we did not observe expression of lytic gene BZLF1 in any of the sub-populations of cells (data not shown). Thus, B cell differentiation, abundance of expression of CD23, and other cellular determinants are among the likely causes of non-proliferation of CD23loCD58+ cells. Whether the cell surface molecular expression patterns that characterize the two sub-populations are causal to the distinct outcomes, have other functional significance, or simply mark the sub-populations is unclear.
We initially infected total peripheral blood B cells with high titers of EBV to include both naïve and non-naïve B cells that are thought to be targets of infection during primary infection with EBV and during the early stages of development of B cell-EBV lymphomas in immunocompromised hosts. To better understand whether the dichotomy in outcome was related to the differentiation stage of target B cells, we exposed purified naïve and memory B cells to EBV. Proliferating and non-proliferating EBV-infected cells did not exclusively derive from one or the other type of B cells. A related question, which has been the subject of earlier investigations, is whether memory B cells serve as direct targets for EBV-mediated proliferation and immortalization. Based on experiments that have relied on expression of viral latency genes for evidence of "immortalization", it has been extrapolated that EBV can immortalize both memory and naïve B cells [2, 3, 22]. However, using viral latency gene expression as evidence for immortalization may be fallacious as our data demonstrates that expression of viral latency genes and CD23 during the early stages of EBV infection does not necessarily correlate with proliferation and therefore potentially immortalization. We have found that while EBV infects both memory and naïve B cells (Figure 6D) ex vivo, using the marker pattern that we have identified, it appears that memory cells can serve as direct targets for EBV-driven proliferation (Figure 5). The inability of LMP1+-naïve B cells to proliferate in our study may be related to the absence of other types of cells or cytokines such as IL6 in an ex vivo setting.
EBNA2 RNA was the only latency gene product detected in CD23loCD58- cells. Since EBNA2 is not easily amenable to FACS staining, we were unable to determine if EBNA2 protein was expressed and if so, in what fraction of this sub-population, and at what levels. However, since EBNA2 is a major transactivator of LMP1 in the early stages of infection [23, 24], low levels of expression of LMP1 protein in a very small fraction of CD23loCD58- cells may serve as an indirect indicator of deficient EBNA2 protein. Low levels of expression of CD23 within cells in this sub-population may also contribute to lack of cell proliferation. Certainly lack of expression of CD23 in EBNA+ cells is known to prevent immortalization of B cells [6, 7].
IL6 is predominantly expressed by CD23loCD58+LMP1+ cells. Expression of IL6 by non-proliferating cells is consistent with the observation that only rare EBV-immortalized tonsillar blasts expressed IL6 during primary infection . IL6 is a growth factor for LCL in culture  and in SCID mice . A strong positive correlation exists between development of post-transplant EBV-lymphomas in humans and elevated levels of serum IL6 . It is tempting to speculate that IL6-producing cells aid the proliferating sub-population, perhaps during the early stages of EBV-infection via paracrine mechanisms. While the results of the experiment in Figure 4 argue in favor of this hypothesis, post-sort mixing experiments will be necessary to determine the effects of the different sub-populations and IL6 on the viability and proliferation of CD23hiCD58+ cells.
Transfection of LMP1 gene was found to increase expression of CD58 , an adhesion molecule, suggesting that LMP1 drives expression of CD58 following EBV infection. We observed a discernible increase in the fraction of CD23+CD58+ cells as early as 18 h after exposure to EBV (Table 1); yet, LMP1 expression has not been detected earlier than 48 h after exposure to EBV . This raises the question of an LMP1-independent mode of expression for CD58. In support of an LMP1-independent mode of expression of CD58, we found that greater than 50% of CD23- cells expressed CD58 around day 3 (Table 1) while fewer than10% of CD23- cells expressed LMP1 on day 3 (data not shown).
In the absence of EBV infection, CD58 was expressed almost exclusively on CD23- cells (Table 1, time 0), while after exposure to EBV, expression of CD58 was up-regulated greatly on CD23+ cells. Since only a fraction of CD23+CD58+ cells proliferated, it is unlikely that CD58 plays a direct role in proliferation. CD58 may exert its effects in a broader and more global capacity to indirectly affect the proliferating population. Indeed this rapid up-regulation of CD58 may be part of the immune response alerting the host to infection with EBV. Interaction of CD58 with its ligand CD2 on T cells may elicit EBV-directed T cell responses. Interactions between B and T cells were noted following gene transfer of LMP1 .
Although using B cells from healthy EBV-seronegative individuals may have yielded two potential advantages, we used cells from healthy EBV-seropositive subjects. First, depletion of T cells might not have been necessary and second, "contaminating" naturally infected B cells, although very few in EBV-seropositive subjects, would have been absent in cells derived from EBV-seronegative individuals. However, since most EBV-seronegative individuals are children and adolescents, it is difficult to obtain large amounts of blood with each draw. As large volumes of blood were necessary for our experiments, the majority of experiments were performed with B cells from healthy EBV-seropositive individuals. To ensure that naturally infected B cells did not give rise to proliferating cells, we tested B cells from all five EBV-seropositive subjects for outgrowth in culture in the absence of T cells and exogenous EBV. None of the cells showed outgrowth as shown for one subject in Figure 3C. Thus, although 1-50 out of 106 peripheral B cells in EBV-seropositive subjects are naturally infected with EBV , these cells are unable to grow out in culture even in the absence of T cells. Experiments with cells from two EBV-seronegative subjects demonstrated emergence of similar sub-populations of B cells after exposure to EBV (Figure 2B).
Consistent with the work of others , between 5% (Figure 1A) and 15% (data not shown) of peripheral B cells are CD23lo while the rest are CD23- prior to exposure to EBV. Mature B cells express low levels of CD23 . Whether the initial drop in the fraction of CD23lo cells following exposure to EBV (Figure 1A) is due to down-regulation of cell surface expression of CD23 or death of CD23-expressing cells is unclear. Since CD27+ memory B cells tend to be CD23-  and give rise to CD23hi cells (Figure 5B), it is likely that CD23hi cells derive from CD23- cells. This is consistent with the work of Azim et al. which suggests that CD23- cells could serve as targets for EBV-mediated immortalization .