Epstein-Barr virus (EBV) has been associated with a host of hematologic malignancies since its discovery as herpesvirus-like particles in African children with lymphoma by Denis Burkitt. More than 90% of the population has been infected by this gammaherpesvirus, but only a small subset develop EBV-related malignancies despite EBV’s potent ability to transform B cells into continuously proliferating lymphoblastic cell lines (EBV-LCLs) in vitro. Effective immune surveillance of EBV in vivo is manifest by the clearance after initial infection of cells expressing an array of immunogenic viral antigens known as Latency III. Most persistently infected cells evade immune surveillance by avoiding consistent EBV protein expression. Additional profiles of restricted EBV latent protein expression occur in various disease settings. Cytotoxic T-cell lines (CTLs) as therapy for EBV-related lymphoproliferative disorders in immunocompromised individuals have historically been generated by using EBVinfected cells in Latency III. In the accompanying article, Bollard et al make significant advances in EBV-directed cellular therapy for immunocompetent individuals by expanding reactivity to less immunogenic EBV-encoded proteins expressed in the more restricted Latency II state. To do so, they target EBV proteins essential for mimicking B-cell developmental pathways and enabling viral persistence. EBV-infected cells and associated malignancies are characterized by distinct patterns of latent protein expression dictated by the differentiation state of the infected B cell. During progression from broad viral protein expression in initial infection through a restricted profile in the germinal center to minimal protein expression in the peripheral blood, EBV provides multiple signals required for naive B cells to differentiate into memory cells. This process normally occurs in response to foreign antigens and is recapitulated during the EBV latent life cycle to ensure maintenance of EBV in the host while facilitating immune escape (Figs 1A and 1B). The Latency III or “growth” program is expressed by activated B cells during initial infection, by post-transplantation lymphoproliferative disorders (PTLDs), by immunoblastic lymphomas arising in HIV infection, and by EBV-LCLs in vitro. Latency III entails expression of Epstein-Barr virus nuclear antigen 1 (EBNA1), EBNA2, EBNA3A, EBNA3B, EBNA3C, EBNA6/LP, latent membrane protein 1 (LMP1), and LMP2. Viral proteins EBNA2 and LMP1 are essential for cellular transformation and immortalization in vitro. EBNA2 is a transcriptional activator of many viral and cellular genes and pathways including CD21, CD23, PU.1, and retinol-binding protein-J / Notch. LMP1 associates with tumor necrosis factor receptor–associated factors to activate nuclear factor kappa B and upregulate expression of CD40 and bcl-2 family members. Mimicking signaling by an activated CD40 molecule, LMP1 enables germinal center B cells to escape apoptosis and enter the memory compartment. In immunocompetent carriers and using LCLs in vitro, cytotoxic CD8 T-cell responses are primarily generated against EBNA3 antigens; CD4 T-cell responses are primarily to EBNA3 and EBNA1. The default or Latency II program, consisting of EBNA1 and LMP1/LMP2 expression, exists in germinal center B cells. LMP2, although not essential for transformation, interacts with src and syk kinases to produce low, constitutive signal transduction downstream of the B-cell receptor. This desensitizes B cells to further stimulation, prevents activation of EBV into the lytic cycle, and allows B-cell survival in the absence of antigen-specific signals. The less frequent T-cell reactivity seen against LMP1 and LMP2 may, in immunocompetent hosts, enable development of Hodgkin lymphoma (HL) and diffuse large B-cell lymphoma (DLBCL), both of which express the Latency II program. EBV genomes in vivo are found in one of 10 to 10 resting memory B cells, and they evade immune surveillance by lack of consistent EBV protein expression (ie, Latency 0). The lytic cycle and viral replication occur when memory cells differentiate into plasma cells. During homeostatic division, an EBV-infected memory cell enters Latency I, expressing only EBNA1, the same program expressed in Burkitt’s lymphoma. EBNA1 binds the viral episome to mitotic chromosomes to ensure proper partitioning and persistence. Although CD4 T cells recognize EBNA1, a glycine-alanine repeat impedes proteasomal processing and presentation on major histocompatibility complex class I. CD8 T cells that recognize EBNA1 are infrequently detected. EBV-specific CTLs as therapy for PTLDs in hematopoietic stemcell transplantation recipients were originally generated by ex vivo exposure to irradiated EBV-LCLs in Latency III. Long-term follow-up of 114 recipients of donor-derived EBV-specific CTLs raised in this manner revealed 80% response rate against PTLDs and 100% efficacy as prophylaxis in high-risk individuals. This approach yielded a polyclonal mix of CD8 and CD4 T cells primarily directed against EBNA3 proteins and lytic proteins (Fig 1C). In immunocompetent individuals, CTLs generated in this manner were first used against refractory HL, and clinical efficacy was 30%; CTLs were ineffective in patients with bulky disease. These suboptimal results were potentially related to poor immunogenicity of Latency II antigens expressed in Hodgkin lymphoma. Following expansion with LCLs, 1% or fewer CD8 T cells recognize LMP2; reactivity to LMP1 is virtually undetectable. Alternate antigen-presenting cells (APCs), JOURNAL OF CLINICAL ONCOLOGY U N D E R S T A N D I N G T H E P A T H W A Y VOLUME 32 NUMBER 8 MARCH 1