Search

Epstein-Barr virus (EBV), a human herpesvirus, is strongly linked with two relatively rare forms of B-cell lymphoma and with a much more prevalent epithelial malignancy, undifferentiated nasopharyngeal carcinoma (NPC). The availability of suitable culture systems has allowed detailed analysis of EBV-induced growth transformation in B lymphocytes, but little is known about the virus--epithelial cell interaction or about the possible effector role of viral proteins in the pathogenesis of NPC. Here we describe an experimental system to monitor the effects of introduced viral or cellular genes upon human epithelial cell growth and differentiation. We transfected a human epithelial cell line, which retains several features of normal keratinocyte behaviour in vitro, with the EBV gene encoding latent membrane protein (LMP), one of only two viral proteins known to be expressed in NPC cells in vivo. LMP expression was accompanied by changes in the epithelial cell surface phenotype, mimicking surface changes observed in NPC cells, and by severe impairment of the cellular response to differentiation signals. The ability of LMP to inhibit terminal differentiation indicates a mechanism whereby EBV infection of squamous epithelium could contribute to the multi-step pathogenesis of NPC.

One of the central unsolved questions of Epstein-Barr virus (EBV) biology is the mechanism of virus persistence in the immunologically competent host. In vitro this virus displays potent growth-transforming activity for one of its major target cells, the B lymphocyte; in vivo virus infection of the B-cell pool appears to be very largely asymptomatic. Much current work on the biology and immunology of EBV infection is aimed at resolving this apparent paradox. The present paper reviews recent progress in the field and develops a provisional model for asymptomatic EBV persistence. The principal gaps in our knowledge will be identified as a stimulus for future work.

Epstein-Barr virus (EBV) persists within the B cell system and, in the blood of healthy virus carriers, is sequestered in the immunoglobulin (Ig)Dneg, CD27+ (“class-switched”) memory B cell compartment. This selective colonisation is apparent even in the blood of infectious mononucleosis (IM) patients undergoing primary EBV infection. Exactly how this is achieved is not understood, one theory being that the virus infects naïve B cells preferentially in vivo and drives them into memory using the physiologic process of germinal centre (GC) transit within lymphoid tissues. Here we have studied the distribution of EBV-infected B cells in tonsil cell preparations from acute IM patients and from chronic carriers. The following subsets of tonsillar B cells were isolated to high purity by FACS sorting and their EBV DNA loads were then determined by quantitative PCR: Naïve (CD38neg, IgD+, CD27neg); Class-switched memory (CD38neg, IgDneg, CD27+); Non-switched memory (CD38neg, IgD+, CD27+); Germinal centre (CD38+); and Plasma cell (CD38hi). In 15 carrier tonsils, the highest EBV load was consistently detected in class-switched memory, followed by significant amounts in non-switched memory. However, viral load in the CD38+ GC subset was usually very low, comparable to that in the naïve B cell fraction. This contrasted with the situation in 7 IM tonsils where overall viral loads were on average 100-fold higher than above. Here most viral DNA was consistently found in the CD38+ fraction, with lower loads in the class-switched memory and the smaller non-class switched memory subsets. We then investigated further the phenotype of the CD38+ population and found that, in IM tonsils, this not only included cells with the CD10 and CD77 markers of germinal centre origin but also other cells lacking these markers. Adopting a different FACS sorting strategy based on CD10, CD77 and IgD staining, we found that the virus was not present at significant levels in either the centrocyte (CD10+ CD77neg) or centroblast (CD77+) fractions. Thus EBV appeared to be concentrated in B cells expressing CD38 as a marker of activation rather than of germinal centre origin; this concurred with the results of immunohistochemical staining for the EBER RNAs showing an extrafollicular distribution of virus-infected cells in these IM tonsils. Our findings suggest that CD38 expression is a useful marker for B cells recently activated by EBV infection in vivo and show that, during primary EBV infection, the tonsillar CD38+ B cells lacking germinal centre markers harbour the highest viral load. If EBV does exploit the germinal centre pathway as an entry route into B cell memory, where primary infection is manifest as IM this pathway must be taken by a very small fraction of infected B cells and/or must have been completed before disease symptoms appear.

Epstein-Barr virus (EBV) infects >90% of population world wide and, in healthy virus carriers, establishes life long persistence in the immunoglobulin (Ig)Dneg, CD27+ (“class-switched”) memory B cell compartment normally produced by antigen stimulation and transit through germinal centres. Patients with the X-linked lymphoproliferative disease (XLP) cannot make such class-switched memory B cells due to an inherited mutation in the slam-associated protein (SAP) gene involved in the maturation of antibody responses. Interestingly, XLP patients are highly susceptible to severe primary EBV infection and develop a fulminant infectious mononucleosis (IM) which is often fatal and where the symptoms progress to resemble those of a different (non-familial) disease, EBV-associated haemophagocytosis syndrome (EBV-AHS), caused by virus entry into the NK or T cell system. Some XLP patients survive their primary infection, but the nature of EBV carriage in these individuals (lacking conventional memory B cells) remains unresolved. To investigate this further we obtained blood samples from 8 such XLP patients. EBV load in total peripheral blood mononuclear cells (PBMC), determined by quantitative PCR, occupied a broad range but on average was 2 to 3-fold fold higher than that of healthy controls. The virus was concentrated within the B cell (CD19+) compartment, as in healthy carriers, and not within T or NK cells, as typically seen in EBV-AHS. We then confirmed that these XLP patients indeed lacked conventional class-switched memory B cells but did carry a small population of IgM+, IgD+, CD27+ (“non-switched”) memory cells; however their circulating B cell pool was dominated by naïve (IgM+, IgD+, CD27neg) cells and by expanded numbers of immature “transitional” (CD10+ CD27neg) cells. In each of 4 cases studied by cell sorting, EBV was concentrated in this small subset of “non-switched” memory B cells. To see if the high virus load detected in these patients indicated true virus persistence as opposed to recent or recurrent infection, serial samples obtained over a 3 year period from two XLP patients were assayed. Virus load was stable and, in one case with the highest load, screening with markers of virus polymorphism detected the same resident strain over time. Our results in XLP patients make it clear that EBV can persist in the absence of a conventional class-switched memory B cell compartment. Instead, the virus is sequestered in a small population of “non-class-switched memory” cells with Ig gene mutations. The origin of such cells, which are also detectable in the blood of normal donors, is uncertain; however their existence in XLP patients suggests that such cells arise independently of germinal centre activity and hence that EBV may be able to colonise its host without exploiting germinal centre transit.

Epstein-Barr virus (EBV) is aetiologically linked with Burkitt Lymphoma (BL) but its contribution to lymphomagenesis, versus that of the chromosomal translocation activating c-myc expression, remains unclear. This is in part because the full virus growth transforming programme that is expressed when EBV infects normal resting B cells, is not expressed in BL. Instead EBV in BL normally exhibits a restricted Latency I form of infection characterised by expression of only one latent antigen EBNA1 from the BamHI Q promoter. Here we describe an endemic BL, Awia, which uniquely is heterogeneous at the single cell level for EBV gene expression. Analysis of single cell clones of Awia-BL revealed cells displaying three forms of restricted EBV latency: classical Latency I, ’Wp-restricted latency’ expressing EBNAs 1, 3A, 3B, 3C and -LP, and a novel ’EBNA2+/LMP1- latency’ in which all EBNAs including EBNA2 are expressed without the latent membrane proteins LMPs 1 and 2. Comparison with rare EBV-negative clones from the same tumour showed that each form of infection provides the c-myc-expressing BL cells with a specific degree of protection from apoptosis. Microarray analysis was carried out on the isogenic Awia-BL clones to determine the influence of EBV gene expression on cellular transcription. Interestingly it was found that expression of the pro-apoptotic Bcl2 family protein BIM, which has previously been implicated in c-myc induced apoptosis, was downregulated in the apoptosis resistant BL cells. Our work suggests that EBV acts as an anti-apoptotic rather than a growth-promoting agent in BL by downregulating expression of the cellular BIM protein. In addition the microarray analysis on these isogenic Awia-BL clones showed that the EBNA profile influences the differentiation status of the BL cell on the germinal centre to plasmacytoid differentiation pathway. These findings may reflect viral functions that are important not just for BL pathogenesis but also for EBV’s normal strategy of persistence in the B cell system.

Following primary infection, Epstein-Barr virus (EBV) establishes life long persistence in the host IgD− CD27+ memory B cell compartment rather than the IgD+ CD27+ marginal zone (MZ)-like or the IgD+ CD27− naïve B cell compartments. One possible explanation for such exclusive persistence in memory B cells is that EBV preferentially infects memory B cells. Alternatively, the virus may infect all B cell subsets but then drive MZ and naïve B cells to acquire the Ig isotype-switched phenotype and hypermutated Ig genotype of memory cells. Here we ask whether there is any evidence for one or other hypothesis from in vitro experiments. B cells from healthy donor blood samples were FACS sorted on the basis of IgD/CD27 expression into naïve, MZ, and memory B cell subsets with purities of >99%, >97% and >98% respectively. Analysis of the IgVH sequence further confirmed purity of the FACS sorted B cell subsets. Accordingly, 102 of 105 IgVH sequences amplified from purified naïve B cells were germ-line where as the vast majority of sequences amplified from MZ and memory B cells were mutated. All three B cell subsets expressed equal amounts of CD21 (EBV receptor on B cells), bound similar amounts of virus, and transformed with equal efficiency to establish B lymphoblastoid cell lines (LCLs) in vitro. Naïve B cell transformants upregulated CD27 expression but retained the IgM+, IgD+ phenotype as determined by FACS analysis and RT-PCR; MZ-B derived LCLs likewise were IgM+, IgD+, CD27+; and memory-B derived LCLs were consistently CD27+, IgD− and expressed either IgG, IgA or in some cases IgM. Therefore, EBV infection per se did not induce class switching. However, both naïve and MZ-B derived LCLs could still be induced to switch to IgG in the presence of CD40 ligand and IL-4; signals that are normally provided by T cells in vivo. To assess if EBV infection might drive Ig hypermutation, we carried out IgVH sequence analysis on the naïve-B derived LCL clones. Interestingly, 42 of 114 clonal IgVH sequences amplified from naïve-B derived LCLs had 3 or more mutations and the patterns of mutation seen were consistent with that produced by somatic hypermutation (SHM). Furthermore, within some naïve-B cell derived LCL clones, there were both germ-line and mutated sequences all sharing the same VDJ rearrangement (CDR3 sequence), again implying sequence diversification following EBV transformation of a single naïve B cell. Some intraclonal variation of the already hypermutated IgVH sequence was also noted in memory and MZ-B derived LCLs further suggesting ongoing mutational activity. Consistent with this, activation-induced cytidine deaminase (AID) expression was upregulated in transformants as assessed by real time RT-PCR. Our in vitro data is therefore compatible with a model of EBV persistence where the virus infects all mature B cell subsets but then drives infected naïve B cells to acquire a memory genotype by inducing SHM. In addition, EBV infected naïve and MZ-B cells may undergo Ig class switching to acquire the IgD− CD27+ memory phenotype in the presence of T cell help in vivo. EBV’s ability to induce SHM may also contribute to the lymphomagenic potential of the virus in addition to its B cell transforming and growth promoting properties.

Somatic hypermutation of immunoglobulin (Ig) gene sequences in the germinal centres of lymphoid tissues is necessary for affinity maturation of B cell responses to antigen challenge. This process generates a few clones with improved affinity that are selected into B cell memory and many clones with other non favourable Ig mutations, including some cells with functionally inactivated Ig gene that normally die by apoptosis. It is postulated that infection with Epstein-Barr virus (EBV), a B lymphotropic agent linked to several types of B cell lymphoma, can rescue germinal centre cells with unfavourable mutations. This creates a pool of infected cells at greater risk of developing into lymphomas. In the present work, CD38+ germinal centre B cells were separated from tonsil by negative selection for IgD and CD39. Peripheral blood naïve and memory B cell subpopulations were FACS sorted as IgD+, CD27− and IgD−, CD27+ fractions respectively. These cells were infected with EBV (B95.8 strain) in vitro and seeded at limiting dilutions onto fibroblast feeders. EBV transformed lymphoblastoid cell lines (LCLs) from such cultures were analysed for surface Ig phenotype. Naïve B cell transformants were consistently IgM+, IgD+. Memory B cell transformants were IgM+ in some cases but more frequently IgG+ or IgA+. Germinal centre transformants showed the same spectrum of surface Ig phenotypes as memory cell transformants but in addition we identified six germinal centre derived LCLs which were consistently surface Ig negative. Sequencing from these lines confirmed that in at least three cases EBV had rescued cells with functionally inactivated Ig heavy chain gene.

CD8+ T cells dominate the lymphocyte population insynovial fluid in chronic inflammatory arthritis. It is known that theseCD8+ T cells are often clonally or oligoclonally expanded, but theirspecificity and their relevance to the pathogenesis of joint disease hasremained unclear. We found that as many as 15.5% of synovial CD8+ Tcells may be specific for a single epitope from an Epstein-Barr virus lyticcycle protein. The virus-specific T cells within the joint showed increasedexpression of markers of activation and differentiation compared with those inthe periphery, and retained their functional capacity to secreteproinflammatory cytokines on stimulation. These activated, virus-specificCD8+ T cells could therefore interact with synoviocytes, either bycell-cell contact or by a cytokine network, and play a 'bystander'role in the maintenance of inflammation in patients with arthritis.

This work examines the basis for our earlier observation that certain monoclonal antibodies (MAbs) specific for the B-cell-associated C3d/Epstein-Barr virus (EBV) receptor molecule CD21 also react with the surface of some epithelial cells. Of 9 proven anti-CD21 MAbs now examined on frozen sections of human nasopharynx, tonsil and ecto-cervix, only 3 (HB5, anti-B2, AB1) showed staining of stratified epithelium; 2 of these (HB5, anti-B2) also reacted with the surface of epithelial cells freshly dispersed from these sites. The proportion of HB5- and anti-B2-reactive cells in primary epithelial cultures fell to a low but stable level within days of explantation, while almost all permanently established epithelial cell lines, whether SV40 virus-transformed or of malignant origin, were not reactive with either MAb. This contrasts with the pattern of expression of another surface marker also found selectively on cells of the lymphoid and epithelioid lineages, the CDw40 antigen. Staining with CDw40 MAbs on epithelial sections was usually restricted to the basal (proliferating) layer, but the proportion of CDw40-positive cells increased to a relatively high level in normal epithelial cultures; furthermore, most epithelial cell lines expressed this antigen. Immunoprecipitation from the surface of metabolically labelled epithelial cells with the anti-CD21 MAb HB5 yielded a protein of approximate MW 200 kDa, clearly different in size from the 145 kDa CD21 molecule on B cells. This 200 kDa protein was identified on fresh ecto-cervical epithelium, on primary cultures of a laryngeal carcinoma and on one unusual SV40-transformed epithelial cell line. We conclude that stratified human epithelial cells express a 200 kDa surface molecule which is antigenically related to, but not identical with, the CD21 antigen on B cells. It remains to be seen whether this epithelial cell protein can function as an EBV receptor.

Epstein-Barr virus (EBV) isolates show sequence divergence in the BamHI YH region of the genome which encodes the nuclear antigen EBNA 2, a protein thought to be involved in the initiation of virus-induced B-cell transformation; type A isolates (such as B95-8 EBV) encode a 82- to 87-kilodalton EBNA 2A protein, whereas type B isolates (such as AG876 EBV) encode an antigenically distinct 75-kilodalton EBNA 2B protein. In the present work 12 type A isolates and 8 type B isolates have been compared for their ability to transform resting human B cells in vitro into permanent lymphoblastoid cell lines. Although the kinetics of initial focus formation was not markedly dependent upon the EBNA 2 type of the transforming virus, on subsequent passage type A virus-transformed cells (type A transformants) yielded cell lines much more readily than did type B transformants. Direct comparison between the two types of transformant revealed clear differences in several aspects of growth phenotype. Compared with type A transformants, cell lines established with type B virus isolates consistently displayed an unusual growth pattern with poor survival of individual cells shed from lymphoblastoid clumps, a lower growth rate and a greater sensitivity to seeding at limiting dilutions, and a significantly lower saturation density that could not be corrected by supplementation of the medium with culture supernatant containing B-cell growth factors. This is the first direct evidence that, in EBV-transformed B-cell lines, the EBNA 2 protein plays a continuing role in determining the cellular growth phenotype.

Some EBV+ BL cell lines continue to grow as single cells on in vitro passage, show an unusually restricted expression of EBV-latent genes and retain a BL biopsy-like cell surface phenotype (group I/II lines); others change to growth in aggregates, show a broader pattern of virus latent gene expression, and develop a cell surface phenotype more characteristic of EBV-transformed LCL (group III lines). Here we show that the cell surface adhesion molecules LFA-1, ICAM-1, and LFA-3 are expressed at very low levels, if at all, on group I/II lines and are coordinately upregulated as BL lines move towards group III. The change to growth in aggregates reflects the increasing availability of LFA-1 and ICAM-1, the two ligands whose mutual interaction underlies homotypic BL cell adhesion in vitro. The low levels of ICAM-1 and LFA-3 on group I/II BL cell lines are also associated with an impaired ability to interact with EBV-specific CTL in the antigen-independent phase of effector/target conjugation. mAb blocking studies show that the small number of conjugates that are formed with group I/II BL targets involve the LFA-1/ICAM-1 adhesion pathway but not the LFA-3 pathway; in contrast, both pathways contribute to the efficient conjugate formation shown by group III BL or LCL targets. Earlier work identified one group III line, WW1 BL, as unusual since is expressed the full spectrum of EBV-latent proteins yet remained insensitive to lysis by EBV-specific CTL. Here we show that this line has an anomalous pattern of adhesion molecule expression with high levels of LFA-1 and ICAM-1 in the absence of detectable LFA-3. The WW1 BL cells form conjugates with EBV-specific CTL through the LFA-1/ICAM-1 pathway, but in the absence of a target LFA-3/effector CD2 interaction these conjugates do not achieve target cell lysis. This may reflect an important role for target LFA-3 molecules in activating EBV-specific CTL function. From these in vitro studies, we postulate that downregulation of the adhesion molecules LFA-3 and ICAM-1 on EBV+ BL underlies the ability of the malignant clone to evade EBV-specific T cell surveillance in vivo.

This report describes the first of 2 investigations studying mechanisms of Epstein-Barr virus (EBV) persistence in the infected host; specifically, we wish to determine the extent to which virus carriage within the B-cell system is dependent upon continued replication of the virus in permissive oropharyngeal epithelium. Levels of EBV infection at these 2 sites have been monitored in 21 acute infectious mononucleosis (IM) patients before, during and after treatment with high doses of acyclovir (ACV). Twelve patients received oral ACV for 10 days and 9 patients received i.v. ACV for 5 days before the 10-day oral course; all were followed prospectively for 28 days. Infectious EBV, detectable at high initial levels in the patients' throat washings, disappeared almost completely during ACV treatment, then returned again to high levels post treatment. In contrast, levels of virus-infected B cells in the blood showed no reduction linked to the period of ACV treatment nor any increase with resumption of EBV shedding. During IM, therefore, maintenance of high levels of virus carriage within the B-cell pool is not dependent upon the continual recruitment of newly infected B cells. This might reflect an inability of the immune T-cell response in acute IM patients to prevent continued expansion of the existing EBV-infected B-cell pool. Alternatively, it raises the possibility that EBV carriage in B cells in vivo is maintained through a virus: cell interaction which is not sensitive to virus-specific T-cell surveillance.

All Epstein-Barr virus (EBV) isolates can be classified as type A or type B depending upon the identity of their EBV nuclear antigen (EBNA) 2 protein. The great majority of isolates examined to date encode an EBNA 2A protein like that of the reference type A strain B95-8. Type B virus strains, encoding an antigenically distinct EBNA 2B protein, have as yet only been rescued from rare Burkitt's lymphoma (BL) cell lines of African origin (Jijoye, AG876). Our recent finding that type B isolates are less efficient than type A in in vitro transformation assays prompted us to determine (i) the relative contribution the two types of virus make to the incidence of BL in endemic areas of Africa (Kenya) and New Guinea and (ii) the relative incidence of infection with these two types in the normal population in these same areas. On the first point, EBNA 2 gene typing using specific DNA probes showed that four of ten recently established Kenyan BL cell lines and two of four BL cell lines from New Guinea carried type B virus isolates. To address the second point, spontaneous lymphoblastoid cell lines were established from the blood of normal virus carriers and typed for EBNA 2 at the protein level; a significant proportion (greater than 20%) of the normal population in both the above BL-endemic areas were infected with type B isolates. This is the first indication of the widespread nature of type B virus infection in any community and the first isolation of such viruses from a non-BL source. The reproducible size of the EBNA 2B protein encoded by all type B isolates irrespective of their geographical origin, and of the EBNA 1 protein encoded by all type B isolates from one area, contrasted markedly with the extreme variability in the size both of EBNA 2A and of EBNA 1 seen generally among type A isolates. This suggests that the number of type B virus strains in existence worldwide could be quite limited. Most importantly, the data suggest that type B viruses, despite their relatively poor performance in in vitro transformation assays, can contribute at least as efficiently as can type A viruses to the pathogenesis of BL.

Primary Epstein-Barr virus (EBV) infection, when manifest as infectious mononucleosis (IM), induces a broad-ranging and apparently non-HLA-restricted cytotoxic response whose nature has not been resolved. In the present experiments the ability to cryo-preserve IM mononuclear cell preparations, after depletion of CD16+ natural killer cells, has allowed detailed analysis of the response on appropriately constructed target cell panels. The results show that IM effector preparations are polyclonal with separate HLA class I antigen-dependent reactivities against (a) the autologous EBV-transformed lymphoblastoid cell line (LCL) and (b) particular HLA class I-mis-matched LCLs. The autologous LCL-directed response shows the hallmarks of immunologically specific T cell cytotoxicity; only EBV+ B cell blasts are recognized and the interaction can be blocked by monoclonal antibodies to CD3 and CD8 on the effector cell surface and to HLA class I antigens on the target cell. Such findings demonstrate, for the first time, that the primary cytotoxic response to EBV infection includes a virus-specific HLA-restricted component like that found in the T cell memory of persistently infected individuals. Separate components of the response are preferentially active against some (but not all) HLA-mismatched LCLs, the pattern of reactivity being distinct for each individual IM patient and reproducible on repeated testing. Monoclonal antibody blocking experiments show that these HLA-mismatched interactions also involve CD3 and CD8 antigens on the effector cell and HLA class I antigens on the target cell. Where tested, such lysis affected both EBV+ and EBV− B cell blasts from the relevant HLA-mismatched donors. We postulate that a widespread primary infection of the B cell system by EBV leads to a generalized expansion not just of the virus-specific response but also of other T cell responses coincidentally active at the time. The unusual activity of IM effector preparations against HLA-mismatched LCLs arises from fortuitous cross-recognition of allogeneic cells by immunologically specific cytotoxic T cell clones coincidentally expanded in vivo alongside the EBV-specific response.

Certain newly established Epstein-Barr virus-containing Burkitt's lymphoma cell lines do not express the cytotoxic T-lymphocyte-detected membrane antigen (LYDMA) through which EBV infection is normally controlled by the host. When the EB virus recovered from these BL lines was used to transform peripheral blood lymphocytes from seronegative donors, the lymphoblastoid cell lines (LCLs) that arose were all LYDMA positive. This indicates that the LYDMA-negative nature of the BLs is not the result of a mutation in the resident viral genome but is rather a specific adaptation in those cells, perhaps permitting evasion of the host immune surveillance in tumour development. A comparison of the EBV gene expression in six LYDMA-negative and two LYDMA-positive BL lines and in their corresponding LCLs revealed that several of the BL lines did not express all of the viral gene products classically associated with latent transformation by EBV. Four out of eight cell lines showed restricted expression of the latent membrane protein (LMP) and/or the EB nuclear antigen, EBNA 2. A new level of EBV gene regulation therefore appears to be operating in some of the BL cell lines. The patterns of expression of EBV genes in the cell lines did not show any correlation with the known susceptibility of the lines to T cell killing.

Tumor cells from 10 patients with Epstein-Barr virus-positive endemic Burkitt's lymphoma (BL) have been examined for cell surface phenotype, both at the biopsy stage and during BL cell line outgrowth in vitro, the cultures being followed for up to 150 passages. In all 10 cases, the biopsy cells showed coexpression of the common acute lymphoblastic leukemia antigen (CALLA) and of the BL-associated glycolipid antigen (BLA) with no accompanying expression of several "lymphoblastoid" cell surface markers defined by selected monoclonal antibodies. During cell line establishment and in vitro passage, the individual BL cell lines showed different degrees of progression toward a more "lymphoblastoid" cell surface phenotype, some even losing CALLA and BLA expression while retaining the chromosomal translocations indicative of their malignant origin. This differential capacity for phenotypic progression in vitro explains much, if not all, of the heterogeneity of the BL cell phenotype apparent from many previous studies with panels of long-established lines. Such heterogeneity in vitro belies the true homogeneity of the tumor cell phenotype in vivo.

Summary Cultured human epithelium bound and internalized radiolabelled Epstein-Barr virus (EBV) within 1 h of exposure. A similar percentage of cultured cells also were reactive with monoclonal antibodies to the EBV/C3d receptor of B lymphocytes. In cross-sections of fresh frozen, stratified epithelium, receptor expression seemed limited to the less differentiated subpopulation of cells. These findings support the notion of direct infection of epithelial cells by EBV and suggest a viral life cycle in epithelium dependent on the stage of cell differentiation.

Inoculation of cottontop tamarins with a large dose of Epstein-Barr virus (EBV) leads to the induction of multiple EBV genome-positive lymphomas. These tumors have been characterized as oligoclonal or monoclonal large-cell malignant lymphomas that closely resemble the EBV genome-positive B-cell lymphomas that arise in human allograft recipients. The expression of latent and lytic EBV-encoded proteins was investigated in these virus-induced tamarin lymphomas and in derived cell lines. The tamarin tumors were found to express EBV nuclear antigen 1 (EBNA 1), EBNA 2, EBNA leader protein, and the latent membrane protein (LMP) as determined both by immunohistochemical staining and by immunoblotting. However, within the limits of the immunoblotting assays, no expression of the EBNA 3a protein family could be detected. Assays for lytic-cycle proteins by using both polyclonal human sera and monoclonal antibodies against viral capsid antigen, early antigen, and membrane antigen (gp340/220) showed minimal, if any, expression of these antigens in the lymphoma biopsies. In contrast, the cell lines derived from these lymphomas, even in early passage, expressed abundant levels of the lytic-cycle antigens and also expressed the EBNA 3a protein as well as EBNA 1, EBNA 2, EBNA leader protein, and LMP. This finding suggests that the virus-lymphoma cell interaction, in particular the switch to lytic cycle, is subject to some form of host control in vivo. The expression of EBNA 2 and LMP in these tamarin lymphomas strengthens their resemblance to posttransplant lymphomas in humans, since these human tumors are also EBNA 2 and LMP positive (L. S. Young, C. Alfieri, K. Hennessy, H. Evans, C. O'Hara, K. Anderson, A. Rickinson, E. Kieff, and J. I. Cohen, submitted for publication). Since both proteins are known to be important effector molecules of virus-induced B-cell growth transformation in vitro, their expression in these lymphomas constitutes the best evidence for a direct oncogenic role for EBV in vivo.

Summary Monoclonal antibodies specific for the ‘latent membrane protein’ (LMP) of Epstein-Barr virus (EBV), one of the effector proteins of EBV-induced B cell transformation, have been generated from mice immunized with a β-galactosidase fusion protein containing the carboxyl half of the B95.8 strain LMP sequence. Four monoclonal IgG1 antibodies, designated CS.1, CS.2, CS.3 and CS.4, which together recognized at least three different epitopes on the molecule, were used to examine various aspects of LMP expression in B cell lines transformed in vitro. The pooled CS.1 to 4 reagent detected the LMPs encoded by each of 20 geographically distinct EBV isolates, despite a degree of inter-isolate heterogeneity in the size and antigenicity of the protein. In cell lines carrying the prototype B95.8 virus strain, particularly if these were virus producers, an additional lower molecular weight LMP was also detected; this appeared to correspond to the truncated form of the protein already predicted to exist from the analysis of B95.8 lytic cycle mRNAs. Attempts were made to identify an analogous truncated form of LMP in cell lines carrying other virus isolates after treatment with phorbol ester and/or sodium butyrate to induce virus production. Surprisingly these experiments showed that expression of the full length LMP molecule was itself strongly inducible by these agents; when monitored at the single cell level, this was a generalized response and was not restricted to cells entering a lytic cycle. Expression of LMP in EBV-transformed B cells therefore appears to be subject to a distinct type of regulation.

Tumour cell lines were established in vitro from 16 cases of Epstein-Barr (EB) virus genome-positive Burkitt's lymphoma (BL), 7 of "endemic" origin (i.e. from holoendemic malarial areas of Africa and of New Guinea) and 9 of "sporadic" origin (i.e. from outside such high-incidence areas). All the BL cell lines thus established were monoclonal by immunoglobulin isotype expression and displayed a characteristic chromosomal translocation, t(8:14) or t(8:22), confirming their malignant origin. Clear differences observed between the individual BL cell lines appeared to be related to their endemic or sporadic status. All 7 endemic cell lines began growth as a carpet of single cells, often with small, loose clumps appearing in later passage. Whilst 3 lines of sporadic origin displayed a similar pattern to the above, the majority of sporadic lines grew as large, tight clumps of cells from the first passage onwards. These differences in growth pattern were reflected by differences in cell surface phenotype, as defined in indirect immunofluorescence tests using a panel of monoclonal antibodies (MAbs) specific for B-lineage-associated antigens. BL cell lines could be classified into 3 separate groups on the basis of their reactivity with 6 particular antibodies (MHM6, AC2, Ki-1, Ki-24, J5 and 38.13). All 7 endemic BL cell lines and 2 of the 3 sporadic BL cell lines which began growth as single cells showed a group-I cell-surface phenotype (MHM6, AC2, Ki-1, Ki-24 negative; J5, 38.13 positive) in early passage. In contrast, all 6 sporadic BL cell lines which began growth in large clumps displayed a distinct group-II phenotype (MHM6, AC2, Ki-1 positive/negative; Ki-24, J5, 38.13 positive); in later passage most of these sporadic lines progressed to a group-III phenotype (MHM6, AC2, Ki-1, Ki-24 positive; J5, 38.13 negative) without loss of those immunoglobulin and chromosomal markers identifying the cells' malignant origin. These clear differences between endemic BL cell lines on the one hand and the majority of sporadic BL cell lines on the other suggest that endemic BL arises from a more restricted range of progenitor B cells than does the sporadic form of the disease.

Epstein-Barr (EB) virus-specific effector T cell lines were established from nine virus-immune donors positive for the serologically defined HLA-A2 antigen; of these, four lines contained a demonstrable A2-restricted cytotoxic component. When these four effector populations were each tested on the same panel of EB virus-transformed lines from 20 HLA-A2-positive individuals, 16 of the target cell lines were consistently killed at levels above 25% of the relevant autologous cell lysis. Cytotoxicity appeared to be mediated through a restricting determinant associated with the 'common A2' antigen that these lines shared; indeed the lysis could be specifically blocked by high concentrations of an HLA-A2-specific monoclonal antibody. In contrast, 4 out of 20 target cell lines were not killed by HLA-A2-restricted effector cells, even though they did express the serologically defined A2 antigen and were found in other tests to be susceptible to EB virus-specific cytolysis restricted through other HLA-A or -B antigens on their surface. These results suggest that EB virus-specific cytotoxic T cells can distinguish between serologically identical HLA-A2 molecules via the heterogeneity of their T cell-restricting determinants. Data from one of the effector cell populations further suggested that a serologically defined cross-reaction between the otherwise distinct HLA-A2 and -Bw57 antigens might also be reflected in a cross-reactivity of T cell-restricting determinants.

Epstein-Barr (EB) virus-positive Burkitt's lymphoma (BL) cell lines, recently established from tumour biopsies and displaying chromosomal translocations indicative of their malignant origin, can be classified into two broad sets: (i) lines growing predominantly as single cells/small clumps whose cell surface markers remain close to those of the original tumour cells, and (ii) lines whose growth pattern and cell surface markers have progressed closer to the more “lympho-blastoid” phenotype displayed by all in vitro transformed B-cell lines (LCLs) of normal origin. When compared to LCLs derived from normal B cells of the same patient, BL-cell lines in set (i) generally showed a lower expression of HLA class I and class II antigens and a reduced ability to activate both allospecific and nonspecific (natural killer-like) cytotoxic responses when cocultured with peripheral blood lymphocytes. By contrast, the HLA antigen expression and in vitro stimulatory capacity of most BL-cell lines in set (ii) were much closer to the values displayed by their corresponding LCLs. Since set (i) rather than set (ii) BL cell lines are phenotypically representative of the malignant cells as they exist in vivo , this work suggests that successful out-growth of the virus-carrying tumour cells in the affected host may be facilitated by the inability of these cells to stimulate strong cytotoxic responses.

Summary Epstein—Barr virus (EBV), an agent with growth transforming potential for human B cells, is associated with certain B cell lymphomas in man and also with an epithelial tumour, undifferentiated nasopharyngeal carcinoma (NPC). Since B cell growth transformation is associated with the constitutive expression of a small number of EBV-coded latent proteins, the nuclear antigens EBNA 1, EBNA 2, EBNA 3 and EBNA-LP and the latent membrane protein (LMP), the present work sought to determine whether this same pattern of virus gene expression occurred in NPC. Tumour biopsies were taken from NPC patients from three areas of differing tumour incidence (Kenya, Algeria, Britain) and immediately snap-frozen, as were biopsies of non-EBV-related carcinomas for controls. Immunoblotting of PAGE-separated proteins with selected human sera identified 24 NPC biopsies clearly expressing EBNA 1. When the analysis was extended using selected human sera with antibodies against the other EBNAs, there was no detectable expression of EBNA 2, EBNA 3 or EBNA-LP in any of these 24 biopsies; their EBNA 2-negative status was confirmed using a monoclonal antibody (MAb) PE2 which was reactive in immunoblotting and in immunoprecipitation with EBNA 2A and EBNA 2B proteins. Similar experiments with two different LMP-specific MAbs, CS1 to 4 and S12, revealed heterogeneity between NPC biopsies; 9/24 biopsies were demonstrably LMP-positive, the degree of expression varying considerably between individual tumours in a manner which was not related to the level of EBNA 1 expression. None of the 24 NPC biopsies expressed detectable amounts of EBV lytic cycle antigens. A nude mouse-passaged NPC cell line, C15, likewise expressed EBNA 1 and LMP but none of the other EBV latent proteins nor lytic cycle antigens. This work identifies a novel type of EBV-cell interaction in NPC cells which is distinct from that seen in in vitro transformed B cell lines and from that seen to date in EBV-positive B cell lymphomas.

Interleukin 2-dependent alloreactive cytotoxic T cell lines, with activity predominantly directed against the HLA-A2 antigen, have been generated in vitro by stimulating blood mononuclear cells from donors nonimmune to the Epstein-Barr (EB) virus with appropriate numbers of EB virus-transformed B cells from A2-homozygous individuals. Such effector cells were tested against a panel of EB virus-transformed target cell lines all expressing the serologically defined A2 antigen but typed into "common A2" and "variant A2" subgroups on the basis of their recognition by A2-restricted EB virus-specific cytotoxic T cells. "Variant A2" responder cells cocultivated with "common A2"-bearing stimulators gave rise to effector T cell lines which recognized only the "common A2"-bearing subgroup of targets. By contrast, responder cells from A2-negative donors stimulated with "common A2"-bearing cells produced effector T cell lines in which the strong lysis of "common A2"-bearing targets was accompanied by a lower, but still significant, lysis directed against all targets within the "variant A2" subgroup. In both cases, lysis of the target cells was blocked equally well by the anti-A2-specific monoclonal antibody MA2.1 as by the monoclonal antibody W6/32 specific for HLA-A, -B, and -C determinants. This suggests that HLA-A2 molecules possess at least two distinct sets of epitopes capable of inducing alloreactive T cell cytotoxicity: first, epitopes probably associated with T cell-restricting sites, which generate subgroup-specific responses, and second, epitopes shared by all A2 molecules, and perhaps associated with serologically defined sites, which generate "pan A2" group-specific responses.

Epstein-Barr virus (EBV) is causally linked with endemic Burkitt's lymphoma (BL), a tumour whose homogeneous cell surface phenotype suggests derivation from a particular subset of activated germinal centre B cells in vivo. Endemic BL also shows an unusual form of EBV in fection with downregulation of certain of the virus latent proteins which are constitutively expressed when EBV infects and transforms normal resting B cells in vitro. Here we question whether this virus:cell interaction is unique to malignant BL cells or whether it might be reproduced by in vitro infection of those particular germinal centre cells displaying the BL-like phenotype. Firstly, we show by biochemical meanns that a subset of normal tonsillar B cells does indeed express the globotriaosylceramide glycolipid BLA and the common acute lymphoblastic leukaemia antigen CALLA, 2 important markers of the BL phenotype. Secondly, using 2-colour immunofluorescence labelling with anti-BLA and anti-CALLA monoclonal antibodies (MAbs), 4 subsets of low buoyant density tonsillar B cells (BLA+ CALLA+, BLA+ CALLA−, BLA− CALLA+, BLA− CALLA−) have been separated by means of a FACS and tested for their susceptibility to EBV-induced growth transfromation in a limiting dilution assay. The BLA+ CALLA+ (i. e., BL-like) subset contained the highest proportion of cells already actively in cycle in vivo and gave the lowest yield of transformants, perhaps reflecting the greater efficiency with which EBV transforms resting target cells. Of the cell lines established from the BLA+ CALLA+ population, a significant number retained BLA expression but CALLA was always lost. In 2 further respects, these lines resembled conventional in vitro transformants rather than lines of BL type; thus the cells expressed cellular “activation” antigens (CD23, CD39, CD30, Ki-24) characteristic of the lymphoblastoid phenotype and contained the full spectrum of EBV latent proteins.

Epstein–Barr virus, a human gammaherpesvirus, possesses a unique set of latent genes whose constitutive expression in B cells leads to cell growth transformation. The initiation of this growth transforming infection depends on a viral promoter in Bam HI W (Wp) whose regulation is poorly understood. Using Wp reporter constructs in in vitro transfection assays, we found that Wp was 11- to 190-fold more active in B cell than in non-B cell lines and that three regions of the promoter (termed UAS1, UAS2, and UAS3) contributed to transcriptional activation. The upstream regions UAS3 (−1168 to −440) and UAS2 (−352 to −264) both functioned in a cell lineage-independent manner and were together responsible for the bulk of Wp activity in non-B cells; mutational analysis indicated the importance of a YY1 binding site in UAS2 in that context. By contrast, UAS1 (−140 to −87) was B cell specific and was the key determinant of the promoter's increased activity in B cell lines. Mutational analysis of UAS1 sequences combined with in vitro bandshift assays revealed the presence of three binding sites for cellular factors in this region. When mutations that abolished factor binding in bandshift assays were introduced into a Wp reporter construct, the loss of any one of the three UAS1 binding sites was sufficient to reduce promoter activity by 10- to 30-fold in B cells. From sequence analysis, two of these appear to be novel transcription factor binding sites, whereas the third was identified as a cyclic AMP response element (CRE). Our data indicate that this CRE interacts with CREB and ATF1 proteins present in B cell nuclear extracts and that this interaction is important for Wp activity.

Epstein-Barr virus (EBV)–induced cytotoxic T lymphocyte (CTL) responses have been detected against many EBV antigens but not the nuclear antigen EBNA1; this has been attributed to the presence of a glycine-alanine repeat (GAr) domain in the protein. Here we describe the isolation of human CD8+ CTL clones recognizing EBNA1-specific peptides in the context of HLA-B35.01 and HLA-A2.03. Using these clones, we show that full-length EBNA1 is not presented when expressed endogenously in target cells, whereas the GAr-deleted form is presented efficiently. However, when supplied as an exogenous antigen, the full-length protein can be presented on HLA class I molecules by a TAP-independent pathway; this may explain how EBNA1-specific CTLs are primed in vivo.

Using recombinant retrovirus-mediated transfer, the coding region for the Epstein-Barr virus-encoded EBNA 2A gene was selectively introduced and stably expressed in the Louckes Burkitt lymphoma cell line. Transfected cells displayed the following altered growth characteristics when compared with the parental line: (i) maintenance of proliferation at reduced serum concentrations, (ii) improved seeding efficiency, (iii) production of a unique growth-enhancing factor, (iv) increased sensitivity to the unique factor. These growth-related changes were accompanied by the appearance of large amounts of soluble CD23 fragments in medium conditioned by Louckes cells expressing EBNA 2A. Tumor-promoting phorbol esters were found to induce similar phenotypic change in the Louckes parental line as seen on the introduction of the EBNA 2A gene. Possible mechanisms by which EBNA 2A provides a growth advantage to B lymphoma cells are discussed.

A series of lymphoblastoid cell lines (LCLs) have been established by in vitro infection of fetal bone marrow and fetal liver cells with Epstein-Barr virus (EBV). While most lines showed the usual mature B cell phenotype, a small proportion were cytoplasmic and surface immunoglobulin (Ig) heavy and light chain negative. Analysis of gene rearrangements indicated that the Ig- lines were either germ-line or nonproductively rearranged when probed for JH and were in germ-line configuration for C chi; no mu or chi mRNA could be detected in such cells. Precursor B cell lines were indistinguishable from their normal Ig+ counterparts in their expression of a wide variety of cell surface markers including "activation" antigens usually associated with the lymphoblastoid state; even the single LCL showing germ-line heavy and light chain genes expressed B lineage-specific cell surface antigens. However, the Ig- lines were distinct from their Ig+ counterparts in three important respects: (a) they grew much more slowly and achieved lower saturation densities, (b) they showed unusually high proportions (8-16%) of cells in EBV-productive cycle, and (c) they contained unusually high proportions (up to 40%) of cells expressing free joining (J) chain. These results suggest that precursor B cells differ in their response to the growth-transforming effects of EBV such that the virus-cell interaction in precursor B cell lines is inherently less stable than in conventional LCL. In particular there may be a greater movement of cells out of cycle and along the B cell maturation pathway. It is possible that such movement leads in individual cells either to virus replication or to a "sterile" plasmacytoid differentiation with J chain expression in the absence of Ig synthesis.

In this study, infectious Epstein-Barr virus (EBV) shedding in the oropharynx and numbers of virus-infected B cells in the blood have been monitored in long-term virus carriers receiving acyclovir (ACV) therapy for herpes zoster. Eleven patients on oral ACV were followed prospectively before, during and for 2 weeks after treatment. As expected, the low levels of EBV shedding in these virus carriers (measured as cord-blood lymphocyte transforming activity in throat washings) were eliminated during the period of ACV treatment and returned at later times. Over the same period, however, the frequency of virus-infected B cells in the blood (measured by spontaneous transformation in limiting dilution assay) remained completely unchanged. Regression assays showed that these same patients had normal levels of EBV-specific cytotoxic T-cell immunity, so that the in vivo persistence of virus-infected B cells could not be ascribed to a defect in T-cell surveillance. We infer that the in vivo half-life of the virus-infected B-cell pool in long-term virus carriers is measured in months rather than days. We further suggest that such persistence requires a novel form of virus: B-cell interaction distinct from the type of “latent” infection displayed by in vitro-transformed cells.

Four indices of the EB virus carrier state, for which quantitative in vitro assays now exist, have been monitored in 55 renal allograft recipients under long-term immunosuppression, each patient being tested on a single occasion. By comparison with parallel data from healthy control donors, the results indicate the extent to which virus replication in the throat and virus-infected B cells in the blood are increased as a result of immunosuppression; the concordance between these two independent indices of the level of EB virus infection in vivo, first noted with healthy donors, was again observed within this large group of patients. Immunosuppression also leads to an impairment of virus-specific memory T-cell responsiveness and to an increase in anti-viral antibody titres, but the results show that the level of virus infection prevailing in any one individual cannot be inferred directly from these immunological indices of the virus:host balance. In allograft patients on stable levels of immunosuppression, virus and host appear to establish a new equilibrium. Limited prospective studies suggest that the position of this new equilibrium depends critically upon the virus:host balance prevailing in the same individuals before immunosuppression began. This may be an important consideration in identifying patients for whom immunosuppression may carry a particularly high risk of developing EB virus genome-positive lymphoma.

Five blood donors were identified whose HLA-A2 is different from the common HLA-A2. Their A2 molecule (A2.2) had a more basic isoelectric point than normal A2 (A2.1). Cytotoxic T lymphocytes (CTL) restricted by HLA-A2.1, specific for influenza A and Epstein-Barr viruses, failed to lyse virus-infected target cells with HLA-A2.2. Identical patterns were obtained with both viruses. CTL from four of the A2.2-positive donors recognized target cells prepared from others in the group that shared only the HLA-A2.2 antigen. The A2.2 antigen from one donor seemed to be different in that target cells were not recognized by CTL from donors with the normal A2.1 nor with basic A2.2. There seems, therefore, to be heterogeneity within the HLA-A2.2 subtype.

The Epstein-Barr (EB) virus is the best-known member of a particular class of genetically restricted herpesviruses which are found in several species of ape and of Old World monkey (Deinhardt and Deinhardt, 1979) and which display a unique tropism for host cells of the B lymphocyte lineage. In each case, virus and host appear to have co-evolved such that the natural infection, both in the primary and in the persistent phase, remains largely asymptomatic. This is a remarkable testament to the efficiency of host control mechanisms since these agents clearly have the potential to induce uncontrolled proliferation in infected B cells, a capacity which is manifest in vitro as the virus-induced transformation of such cells into permanent virus genome- positive lymphoblastoid cell lines (Pope, 1979; Rabin et al., 1978). It seems that an analogous sequence of events can occur in vivo but only in very rare circumstances, the classic example being the human X-linked lymphoproliferative syndrome, where primary EB virus infection of boys with a genetically-determined deficiency of cellular immune functions leads to a fatal B lymphoproliterative disease (Purtilo et al., 1982).

A 3.0-kilobase-pair Epstein-Barr virus (EBV) DNA segment necessary for lymphocyte immortalization encodes at least part of a nuclear protein (EBNA2) which is characteristically expressed in latently infected, immortalized cells. A 1.5-kilobase open reading frame within this DNA segment has now been inserted into a murine leukemia virus (MuLV)-derived expression vector (pZIP-NEO-SV(X)1) which provides for transcription of heterologous DNA but not for translational start. Transfection of the recombinant DNA into NIH 3T3 cells resulted in expression of a full-sized EBNA2 which localized to the cell nucleus. Significant new evidence is thereby provided that this 1.5 kilobase open reading frame includes a translational start site and encodes the entire EBNA2 protein. Transfection of the recombinant DNA into a helper cell line (psi am22b) providing amphotropic MuLV-packaging functions resulted in the release of a recombinant MuLV carrying the EBNA2 gene. This recombinant virus can infect rodent cells and convert them to stable EBNA2 expression. Rat-1 cells infected with the MuLV EBNA2 recombinant expressed EBNA2 and grew more rapidly in medium supplemented with 1 or 0.5% fetal calf serum than did Rat-1 cells infected with MuLV vector lacking EBNA2. The Rat-1 cells expressing EBNA2 remained contact inhibited, anchorage dependent, and nontumorigenic in nude mice. Different EBV isolates have one of at least two EBNA2 alleles. Despite divergence between the two alleles, a human serum recognized the prototype EBNA2 allele (EBNA2A) as well as the variant EBNA2B allele characteristic of some Burkitt tumor EBV isolates. The EBNA2B allele was also expressed from the MuLV-derived vector. The reproducible expression of EBNA2A or EBNA2B from these recombinant vectors will facilitate analysis of the EBNA2A and EBNA2B phenotypes.

Epstein-Barr (EB) virus-specific cytotoxic T cells, prepared from virus-immune donors by reactivation in vitro and maintained thereafter as IL-2-dependent T cell lines, have been tested against large panels of EB virus-transformed lymphoblastoid cell lines of known HLA type. Whilst the pattern of lysis of the majority of targets was always consistent with HLA-A and HLA-B antigen restriction of effector function, in several cases it was noticed that certain HLA-mismatched targets were also reproducibly lysed. When this "anomalous" lysis was investigated in detail, it was found to be directed against allodeterminants on class I HLA antigens; thus, mitogen-stimulated as well as EB virus-transformed lymphoblasts from the relevant target cell donors were sensitive to the killing, and in each case the lysis could be specifically blocked by monoclonal antibodies to class I HLA antigens. In one example the target for this alloreactive lysis could be identified as a single serologically defined antigen, HLA-Bw57, while in another example lysis was directed against a "public" epitope common to HLA-Bw35, -Bw62, and a subset of -B12 antigens. Both cold target inhibition experiments and limiting dilution analysis strongly suggested that this alloreactive lysis was being mediated by the same effector T cells that recognize EB viral antigens in the context of self-HLA. This is the first demonstration in man that alloreactive responses can be derived from within the antigen-specific, self MHC-restricted T cell repertoire.

Summary Experiments have been performed to identify the type and size of cell infected by EB virus in the blood of acute infectious mononucleosis (IM) patients, and to investigate the nature of the infection. Virus-infected cells, recognized by their ability to give rise to lymphoblastoid cell lines when co-cultivated with foetal lymphocytes, were shown to be restricted to the B-lymphocyte population. Samples of this population from each of eight IM patients were found to be negative for EB nuclear antigen (EBNA) staining. Thereafter, fractions of IM B-lymphocytes prepared on the basis of cell size were assayed either by co-cultivation, for the incidence of virus-infected cells, or by immunofluorescence staining for the presence of cells expressing EBNA. The great majority of virus-infected cells were found in the fractions of normal sized B-lymphocytes and yet these fractions were unequivocally EBNA-negative. The finding of EB virus-infected, EBNA-negative B-cell populations in IM blood is discussed in terms of the type of infection established by EB virus in the circulation of IM patients.

The Epstein-Barr (EB) virus shows a striking association with two human tumours, Burkitt’s lymphoma (BL) and nasopharyngeal carcinoma (NPC), both of which exhibit an unusual geographical distribution. This is particularly interesting because the virus itself is ubiquitous, being found as a widespread and largely asymptomatic infection in all human communities. To date, there is little evidence to support the view that these different consequences of EB virus infection reflect the existence of different viral strains. Indeed virus isolates from different regions of the world, or more importantly from patients with different virus-associated diseases, have proved to be remarkably similar when their genomes have been compared by restriction enzyme analysis (1). Accordingly much more attention has been given to the identification of non-viral co-factors, which are clearly involved in the pathogenesis of both BL and NPC, as a means of explaining the unusual distribution of these tumours.

Specific T cell proliferation was observed in short-term blood mononuclear cell cultures set up from Epstein-Barr virus (EBV)-immune individuals and challenged either with UV-irradiated EB virions or with a candidate subunit vaccine preparation, the purified envelope glycoprotein gp340 incorporated into immune stimulating complexes (gp340 iscoms). Limiting dilution culture of the activated T lymphoblasts in interleukin 2-containing medium generated stable CD3+CD4+CD8- T cell clones. Particular clones showing virus-specific proliferation in preliminary screening assays were selected for more detailed study. Three gp340 iscoms-induced clones from EBV-immune donor CG responded specifically to restimulation either with UV-EBV or with purified gp340 iscoms in the presence of autologous antigen-presenting cells (APC). Both T cell-depleted blood mononuclear cells and the EBV-transformed B cell line (treated with Acyclovir to block endogenous gp340 production) could be used for presentation, the latter being the more efficient when gp340 iscoms was the source of antigen. Blocking studies with monoclonal antibodies to HLA class II antigens and experiments using HLA-typed allogeneic APC indicated that all three gp340-specific CG clones were restricted through the HLA-DR2 antigen. One gp340 iscoms-induced clone from another EBV-immune donor, MR, likewise showed gp340-specific proliferation, in this case restricted through a HLA-DR4 antigen. Using HLA-DR-homozygous B cell lines representing the five known DR4 subtypes, efficient presentation of gp340 to this T cell clone was observed with both DR4 Dw4 and DR4 Dw14 antigens. Parallel experiments on one UV-EBV-induced T cell clone from donor MR gave a different pattern of results; these cells appeared to be specific for a virus structural component other than gp340 and to be restricted through an HLA-DP determinant.

Several facets of the biology of Epstein-Barr virus (EBV)make this agent particularly interesting to immunolo-gists. Firstly, the virus selectively infects human B lymphocytes and can act as a polyclonal B cell activator both in vivo (1) and in vitro (2). Secondly, primary infection with this agent can elicit an extremely vigorous T cell response, as reflected in the numbers of reactive T lymphoblasts seen in the blood of acute infectious mononucleosis (IM) patients (3). Thirdly, following primary infection, the virus persists for life in B lymphoid tissues where the numbers of infected cells are controlled by long-term T cell surveillance (4). Finally the appearance of EBV genome-positive tumours as a rare accompaniment of the persistent infection suggests some escape of the tumour cells from T cell surveillance (5). Many of the immunological phenomena outlined above have been discussed in some detail in two recent reviews (6,7) and so the present paper will confine itself to the most recent developments in our understanding of cellular controls over primary and persistent EBV infection.

Cytotoxic T cells from Epstein-Barr virus (EBV)-immune individuals specifically kill EBV-transformed B cells from HLA class I antigen-matched donors even though the latently infected cells express only a restricted set of virus genes. The virus-induced target antigens recognized by these immune T cells have not been identified. In our experiments, EBV DNA sequences encoding the virus latent gene products Epstein-Barr nuclear antigen (EBNA)1, EBNA 2, and EBNA-LP and the latent membrane protein (LMP) were individually expressed in a virus-negative human B-lymphoma cell line, Louckes. Transfected clones expressing LMP were killed by EBV-specific cytotoxic T-cell preparations from each of three virus-immune donors HLA matched with Louckes through HLA-A2, B44 antigens; control transfectants or clones expressing one of the EBNA proteins were not recognized. Expression of LMP in a second virus-negative B-cell line, BL41, sensitized these cells to EBV-specific cytolysis restricted through the HLA-A11 antigen. To distinguish between the viral protein and an induced human B-cell activation antigen as the target for T-cell recognition, LMP was then expressed in a murine mastocytoma cell line, P815-A11-restricted human T cells. The LMP-expressing P815-A11 transfectants were susceptible to lysis by EBV-specific cytotoxic T cells from three HLA-A11-positive individuals. Both Louckes and P815-A11 cells were also transfected with constructs capable of encoding a truncated form of LMP (Tr-LMP) which lacks the N-terminal 128 amino acids of the full-length protein. Tr-LMP-expressing transfectants were not recognized by the above T-cell preparations. The results suggest that LMP, and, in particular, epitopes derived from the N-terminal region of the protein, provides one of the target antigens for the EBV-induced human cytotoxic T-cell response.

The Epstein-Barr virus (EBV) nuclear antigens EBNA 3a, 3b, and 3c have recently been mapped to adjacent reading frames in the BamHI L and E fragments of the B95.8 EBV genome. We studied by immunoblotting the expression of the family of EBNA 3 proteins in a panel of 20 EBV-transformed lymphoblastoid cell lines (LCLs) carrying either type A (EBNA 2A-encoding) or type B (EBNA 2B-encoding) virus isolates. Certain human sera from donors naturally infected with type A isolates detected the EBNA 3a, 3b, and 3c proteins in all type A virus-transformed LCLs (with a single exception in which EBNA 3b was not detected) but detected only EBNA 3a in LCLs carrying type B isolates. These results were confirmed with human and murine antibodies with specific reactivity against sequences of the type A EBNA 3a, 3b, or 3c expressed in bacterial fusion proteins. Conversely, selected human sera from donors naturally infected with type B strains of EBV identified the EBNA 3a encoded by both types of isolates plus two novel EBNAs present only in type B, and not in type A, virus-transformed LCLs; these novel proteins appear to be the type B homologs of EBNA 3b and 3c. The distinction between type A and type B EBV isolates therefore extends beyond the EBNA 2 gene to the EBNA 3 family of proteins. This has important implications with respect to the evolutionary origin of these two EBV types and also places in a new light recent studies which identified differences between type A and type B transformants in terms of growth phenotype (A. B. Rickinson, L. S. Young, and M. Rowe, J. Virol. 61:1310-1317, 1987) and of detection by EBV-specific cytotoxic T cells (D. J. Moss, I. S. Misko, S. R. Burrows, K. Burman, R. McCarthy, and T. B. Sculley, Nature [London] 331:719-721, 1988).

Epstein-Barr virus (EBV) infection of EBV-negative Burkitt lymphoma (BL) cells induces some changes similar to those seen in normal B lymphocytes that have been growth transformed by EBV. The role of individual EBV genes in this process was evaluated by introducing each of the viral genes that are normally expressed in EBV growth-transformed and latently infected lymphoblasts into an EBV-negative BL cell line, using recombinant retrovirus-mediated transfer. Clones of cells were derived that stably express the EBV nuclear antigen 1 (EBNA-1), EBNA-2, EBNA-3, EBNA-leader protein, or EBV latent membrane protein (LMP). These were compared with control clones infected with the retrovirus vector. All 10 clones converted to EBNA-2 expression differed from control clones or clones expressing other EBV proteins by growth in tight clumps and by markedly increased expression of one particular surface marker of B-cell activation, CD23. Other activation antigens were unaffected by EBNA-2 expression, as were markers already expressed on the parent BL cell line, including BL markers (cALLA and BLA), proliferation markers (transferrin receptor and BK19.9), and cell adhesion-related molecules (LFA-1 and LFA-3). Increased CD23 expression in cells expressing EBNA-2 was apparent from monoclonal anti-CD23 antibody binding to the cell surface, from immunoprecipitation of the 45-kDa and 90-kDa CD23 proteins with monoclonal antibody, and from RNA blots probed with labeled CD23 DNA. The results indicate that EBNA-2 is a specific direct or indirect trans-activator of CD23. This establishes a link between an EBV gene and cell gene expression. Since CD23 has been implicated in the transduction of B-cell growth signals, its specific induction by EBNA-2 could be important in EBV induction of B-lymphocyte transformation.

Burkitt’s lymphoma (BL) is a tumour which is endemic in regions of Africa and New Guinea where temperature and rainfall are high and malaria is holoendemic. BL also occurs worldwide (sporadic BL), but at a much lower incidence, and showing no association with specific climatic or geographical features. Some 98% of endemic tumours carry multiple copies of the Epstein-Barr (EB) virus genome, and it is possible that this virus has a causal relationship with BL. However, only 20% of sporadic cases are associated with the EB virus, and this would suggest that the endemic and sporadic forms of the disease may differ in pathogenesis.

It is clear that in man, just as in the mouse, virus-specific cytotoxic T cells are genetically restricted in their function, recognising a viral target antigen on the surface of infected cells not as an isolated structure but in some form of association with the polymorphic determinants of HLA class I gene products (1–3). Thus virus-specific cytotoxic T cells offer a means of probing HLA class I antigen polymorphism which is independent of that provided by allo-specific cytotoxic T cells raised in mixed lymphocyte cultures or of that afforded by serological reagents. There is now growing evidence to suggest that a divergence exists between serologically-defined and T cell-defined polymorphisms (4–7). However, relatively little is known about the relationship between those sites on HLA class I molecules recognized by virus-specific cytotoxic T cells and those recognised by allospecific cytotoxic T cells. The work reported here, seeking to distinguish between T cell restricting determinants and allo-specific T cell recognition sites on HLA class I molecules, has taken advantage of the fact that Epstein-Barr (EB) virus-transformed B cells can act as in vitro stimulators of either of EB virus-specific cytotoxic T cells (showing classical HLA class I antigen restriction) or of allo-specific cytotoxic T cells (directed against foreign HLA class I antigens) depending upon the particular in vitro responder: stimulator combination employed (8,9).

Our understanding of the pathogenesis of Burkitt’s Lymphoma (BL) and, in particular, the role played by Epstein-Barr (EB) virus has been hampered by the hitherto unsuccessful attempts to identify the normal B cell subpopulation from which this tumour arises. Previous studies relying upon long established tumour-derived cell lines have reported marked heterogeneity in surface phenotype between individual tumours (1–3), leading to the suggestion that BL might be derived from a relatively broad “window” of target cells on the B cell differentiation pathway. Our recent work has shown that this is not the case (4). Analysis of fresh biopsy cells from EB-virus genome-positive cases of BL has shown that the tumour cells display a consistent and homogeneous surface phenotype with co-expression of the pan B cell marker B1 (CD20), the common acute lymphoblastic leukaemia antigen, CALLA (CD10) and the glycolipid BL-associated antigen, BLA.

Epstein-Barr virus (EBV)-negative Burkitt's lymphoma (BL) cell lines have been converted to EBV genome positivity by in vitro infection with the transforming EBV strain B95.8 and with the nontransforming mutant strain P3HR1, which has a deletion in the gene encoding the nuclear antigen EBNA2. These B95.8- and P3HR1-converted lines have been compared for their patterns of expression of EBV latent genes (i.e., those viral genes constitutively expressed in all EBV-transformed lines of normal B-cell origin) and for their recognition by EBV-specific cytotoxic T lymphocytes (CTLs), in an effort to identify which latent gene products provide target antigens for the T-cell response. B95.8-converted lines on several different EBV-negative BL-cell backgrounds all showed detectable expression of the nuclear antigens EBNA1, EBNA2, and EBNA3 and of the latent membrane protein (LMP); such converts were also clearly recognized by EBV-specific CTL preparations with restriction through selected human leukocyte antigen (HLA) class I antigens on the target cell surface. The corresponding P3HR1-converted lines (lacking an EBNA2 gene) expressed EBNA1 and EBNA3 but, surprisingly, showed no detectable LMP; furthermore, these converts were not recognized by EBV-specific CTLs. Such differences in T-cell recognition were not due to any differences in expression of the relevant HLA-restricting determinants between the two types of convert, as shown by binding of specific monoclonal antibodies and by the susceptibility of both B95.8 and P3HR1 converts to allospecific CTLs directed against these same HLA molecules. The results suggest that in the normal infectious cycle, EBNA2 may be required for subsequent expression of LMP and that both EBNA2 and LMP (but not EBNA1 or EBNA3) may provide target antigens for the EBV-specific T-cell response.

The pathogenesis of Epstein–Barr (EB) virus-positive Burkitt's lymphoma (BL) appears to involve the combined actions of (1) virus-induced B-cell proliferation, and (2) a rare chromosomal translocation juxtaposing c-myc and immunoglobulin gene loci in a single B cell; holoendemic malarial infection in some way facilitates the oncogenic process1–3. Outgrowth of the EB virus-positive tumour suggests either breakdown or evasion of those immune controls, in particular cytotoxic T-cell responses against the virus-induced lymphocyte-detected membrane antigen LYDMA, which limit virus-infected B-cell numbers in healthy virus carriers4. Immunosuppression, such as that which malarial infection may induce5,6, cannot itself be a sufficient explanation in this regard since our studies have identified a number of BL patients who retain detectable LYDMA-specific T-cell surveillance7. The present work shows that in many cases of virus-associated BL, the emerging malignant clone is insensitive to such surveillance. Several EB virus-positive BL cell lines, recently established in vitro and expressing the class I histocompatibility locus antigens (HLAs) which restrict cytotoxic T-cell function, were not killed by HLA-matched LYDMA-specific effector populations in assays where the EB virus-positive lymphoblastoid cell line (LCL), derived from normal B cells of the same patient, sustained high levels of lysis.

Burkitt's lymphoma (BL) biopsy cells and derived cell lines can be grouped according to their patterns of reactivity with 6 selected monoclonal antibodies (MAbs) against B cell-associated surface antigens. Group I cells react only with MAbs J5 and 38.13, recognising the common acute lymphoblastic leukaemia antigen and a BL-associated antigen respectively; group II cells react with J5 and 38.13 and with one or more of a set of MAbs (Ki-24, MHM6, AC2, Ki-1) against "lymphoblastoid" antigens; group III cells react only with these anti-"lymphoblastoid" MAbs. Tumour biopsy cells from 17 cases of sporadic BL, 9 positive for the Epstein-Barr (EB) virus genome and 8 negative, have been analysed during the process of cell line establishment in vitro. In early passage the EB virus-negative BL cells showed either a group I phenotype or gave an additional reactivity with MAb Ki-24 which placed them in group II; these phenotypes remained essentially stable with continued growth of the cell lines for up to 50 passages. By contrast the EB virus-positive BL cells were much more susceptible to phenotypic change in vitro. Although such cells displayed a group I or group II phenotype in early passage, many of the lines soon moved into group III whilst retaining the karyotypic markers indicative of their malignant origin. These observations suggest that a resident EB virus genome can drive the in vitro progression of BL cells towards a more "lymphoblastoid" phenotype. This was confirmed in subsequent experiments where virus-negative BL cell lines were converted to EB virus positivity by in vitro infection. Clearly, therefore, phenotypic analysis of long-established lines can lead to false distinctions being drawn between the EB virus-positive and -negative forms of sporadic BL; both may derive from the same sub-population of target B cells in vivo.

Inoculation of cottontop tamarins (Saguinus oedipus oedipus) with a large standard dose of EBV regularly gives rise to multiple EBV genome-positive B cell lymphomas which appear within 14 to 21 days. In the majority of cases the disease is fatal; in the few surviving animals the tumours regress over a period of 8 to 14 weeks (1). These lymphomas are very similar to the human B cell malignancies which arise in organ transplant patients by the following criteria. Both types of lesion are described histologically as large cell malignant lymphomas with well circumscribed “geographical necrosis”, neither lesion shows a consistent chromosomal abnormality, individual tumours are mono- or oligoclonal in origin and both types of lesion have a capacity to regress(l). Material from human “post-transplant” lymphomas is rarely available in sufficient quantity for detailed protein analysis. For this reason we sought to examine EBV gene expression in the tamarin animal model system.