Facts and hopes in the relationship of EBV with cancer immunity and immunotherapy

Abstract

Epstein-Barr virus (EBV), the first identified human tumor virus, infects and takes up residency in almost every human. However, EBV genome-positive tumors arise in only a tiny minority of infected people, presumably when the virus-carrying tumor cells are able to evade immune surveillance. Traditional views regard viral antigens as the principal targets of host immune surveillance against virus-infected cells. However, recent findings indicate that EBV-infected/transformed B cells elicit both cytotoxic CD8⁺ and CD4⁺ T cell responses against a wide range of overexpressed cellular antigens known to function as tumor-associated antigens (TAAs), in addition to various EBV-encoded antigens. This not only broadens the ways by which the immune system controls EBV infection and prevents it from causing cancers, but also potentially extends immune protection toward EBV-unrelated cancers by targeting shared TAAs. The goal of this review is to incorporate these new findings with literature data and discuss future directions for improved understanding of EBV-induced antitumor immunity, as well as the hopes for rational immune strategies for cancer prevention and therapy.

Introduction

Epstein-Barr virus (EBV), also known as human herpesvirus 4 (HHV-4), is one of the most prevalent viruses in humans, infecting ~95% of global population by adult age. Its initial discovery in endemic Burkitt’s Lymphoma (BL) in equatorial Africa opened the era of exploration of viral etiology of human cancers (1). However, it took three decades for the scientific community to accumulate sufficient evidence to declare EBV as a class I carcinogen (2). By that time, EBV had become etiologically linked to an unexpectedly wide range of lymphoid and epithelial malignancies. Globally, not all of those tumor types showed a 100% association with the virus; however, where any individual tumor was EBV-positive, the virus genome was present and active in 100% tumor cells, indicating that the malignancy had genuinely arisen from an EBV-infected precursor (2–4). Yet this apparently dangerous virus was widespread in all human populations and, once acquired, was in most cases carried asymptomatically for life, raising the question as to why only a small minority of infected people ever develop an EBV-positive cancer (3). In that context, the pathogenesis of each of the EBV-associated tumor types is complex and multi-factorial. However, the integrity and potency of the host immune system, which normally keeps the virus at bay, is likely to be one factor determining tumor susceptibility. Multiple branches of immune cells participate in the control of EBV, among which T cells are the main players. These T cells have been shown to target various EBV antigens (5), in line with the general perception that in viral infections T cells target virus-encoded (foreign) antigens.

However, EBV is distinct from other human viruses in its capacity to hyperactivate the cellular immune response, as evidenced by the huge T-lymphocytosis seen in the blood of infectious mononucleosis (IM) patients undergoing a primary EBV infection (5). One clue as to why this should be comes from the virus’ marked B-lymphotropism and the fact that EBV-induced transformation converts resting B cells into potent antigen-presenting cells (APCs). In that regard one of the key viral effectors of growth transformation, the EBV-encoded signaling molecule latent membrane protein 1 (LMP1) (6), induces the expression of an array of costimulatory ligands (7) and upregulates the infected cell’s antigen processing and presentation machinery (8–11). To add to this panoply of LMP1-induced changes, recent research has revealed that LMP1-expressing B cells can elicit cytotoxic CD4⁺ and CD8⁺ T cell responses to a wide range of tumor-associated antigens (TAAs) (12), a group of cellular antigens often shared by multiple tumors (13).

In this review, we reflect on the more than five decades of research on EBV, with particular attention to the findings related to how the virus interacts with immune system, and discuss future directions applying this knowledge to understanding T cell immunity in various cancers (irrespective of their etiological link with EBV), in a hope to arrive at better informed immune strategies for cancer prevention or therapy.

EBV and Human Cancers: Immune System in Between

EBV was initially discovered owing to its association with endemic BL in tropical Africa (1), and subsequently found to be associated with variable fractions of other types of malignancies. These include tumors such as post-transplant lymphoproliferative disease (PTLD), acquired immunodeficiency syndrome (AIDS)-associated B-cell lymphoma, Hodgkin’s lymphoma (HL), which, like BL, arise in the virus’ natural reservoir of latent infection, the B cell system. However, they also include particular T/NK cell lymphomas, as well as two epithelial tumors nasopharyngeal carcinoma (NPC) and gastric carcinoma, indicating the dangers inherent when the virus accesses atypical target cell types (4, 14). Because EBV expresses different sets of genes—known as latency programs I, II, and III—in the different cancers (3, 15), the mechanisms whereby EBV causes or contributes to the pathogenesis of these cancers also differ, ranging from setting the stage (through enhancing the proliferation and survival of infected cells) for acquisition of oncogenic mutations to directly driving cellular transformation by EBV-encoded oncoproteins (2, 3).

The transforming potency of EBV oncoproteins is best reflected by the virus’s ability to rapidly convert human B cells in vitro into permanently growing lymphoblastoid cell lines (LCLs), which express all of the EBV latent proteins (referred to as latency III), including six Epstein-Barr nuclear antigens (EBNAs 1, 2, 3A, 3B, 3C and -LP) and three latent membrane proteins (LMPs 1, 2A and 2B) (14, 16). Such virus-induced transformation is thought to be a part of the viral strategy for colonizing the B cell system in vivo, following infection of the naïve host, but this expansion is normally contained by the immune response and only in immunocompromised individuals does it develop into pathologies such as PTLD and AIDS-associated B-cell lymphoma (3). Thus, when immune surveillance fails, EBV-driven lymphoproliferation can ensue. Other types of EBV-associated B cell lymphoma might have involved such oligoclonal lymphoproliferative lesions early in their pathogenesis but subsequent oncogenic mutation/clonal selection has allowed them to evade immune surveillance in more intricate ways. For example, chronic infection with T cell-suppressive Plasmodium falciparum malaria is a key factor predisposing to endemic BL (17–19) but the malignant BL cells that finally emerge express only one EBV latent protein EBNA1 (latency I) and have a deficiency in their class I antigen processing machinery (8). In EBV-positive HL, the malignant Reed-Sternberg cells express a few more latent proteins (EBNA1, LMPs 1, 2A and 2B; latency II) but also produces immunosuppressive molecules that foster local immune privilege (20, 21).

While it is important to further unveil how the various EBV-associated cancers breach immune control, it is even more crucial to understand how EBV turns on immune surveillance in the first place. In this regard, it is particularly stunning to realize that while EBV is the most efficient transforming agent among all cancer-associated pathogens (14), it is also the most potent at activating the immune system—it does so by turning the infected/transformed B cells into robust APCs.

EBV-LCL as APC: In Basic Research and Therapy Development

Since early 1980s it has been known that EBV LCL can function as APCs (22). These cells are able to process and present endogenous and exogenous antigens on major histocompatibility complex (MHC)/human leukocyte antigen (HLA) classes I and II, to induce CD8⁺ and CD4⁺ T cell responses, and have been widely used as an APC system in the immunology field [see discussion in (23)].

Owing to the inherent APC function and expression of all of the EBV latent proteins (latency III program), LCL cells can efficiently stimulate autologous, or HLA-matched allogeneic, T cells to produce “EBV-specific” cytotoxic T lymphocytes (CTLs). This led to the development of adoptive CTL therapy for PTLD (24) and other EBV-related malignancies. Because PTLD is a tumor that arises in immunocompromised hosts and expresses the same latency III group of viral proteins as LCLs, the LCL-elicited “EBV-specific” T cells were originally used to reconstitute the underlying immune defect that leads to the development of PTLD. This work, in a small number of patients, showed that the transferred T cells can prevent or control PTLD in transplant recipients (24). Since then, “EBV-specific” T cells have been used with considerable success in treating PTLD (25–29). These T cells usually comprise predominantly CD8⁺ cells, with variable frequencies of CD4⁺ cells. Strikingly, however, the results of one PTLD trial showed a significant correlation of the clinical response with the frequency of CD4⁺ cells in the infused T cells (27), suggesting their therapeutic importance.

In contrast to the highly immunogenic PTLD, EBV-associated cancers that develop in immunocompetent individuals, including HL and NPC, predominantly exhibit the latency II phenotype (3, 14). Crucially they lack the immunodominant EBNA3 proteins and express a more limited array of weakly immunogenic (subdominant) viral antigens (EBNA1, LMP1 and LMP2). T cells recognizing these antigens are therefore seen at low frequency (against LMP2) or are nearly undetectable (against LMP1 and EBNA1) in LCL-stimulated T cell preparations (30, 31). Nonetheless, these T cell preparations induced complete responses in some patients with refractory HL (32) or NPC (33, 34). It is noteworthy that, in patients treated with LCL-expanded T cells for latency II lymphomas, T cell responses were detected against nonviral TAAs (see discussion about their generation below); these T cell responses appear to correlate with antitumor immunity, as they were seen only in clinical responders (35).

LMP1 Signaling for APC Function

Enhancing antigen processing and presentation

Despite the wide use of EBV-LCL as APCs, the underlying mechanism for such function has been poorly understood. Some early studies suggested that LMP1 might be involved, beyond its well-known oncogenic actions. Ectopic expression of LMP1 in B cell lines with poor APC function was shown to upregulate processing and presentation of endogenous as well as exogenous antigens (8, 36). Moreover, it was found that LMP1 expression displays a cyclic pattern and varies greatly in individual cells of a LCL culture (11, 37), and its level correlates with HLA-I expression and antigen presenting function (11).

Empowering costimulation

An APC must not only present antigens, but also simultaneously provide costimulation, to induce productive T cell response. Our work in mouse models showed that expression of LMP1 in B cells induces potent T cell responses, which in turn eliminate LMP1⁺ B cells; T cell depletion leads to rapid, fatal B cell proliferation and lymphomagenesis, resembling EBV-driven malignancies in immunosuppressed patients (38, 39). This work revealed a central role for LMP1 not only in the growth transformation of B cells but also in the immune surveillance of those transformed cells in vivo. Using an in vitro culture system, we showed that LMP1⁺ B cells suffice to function as APCs to prime cytotoxic T cell responses (7). The T cell responses thus induced included not only CD8⁺ T cells but also CD4⁺ T cells displaying typical CTL phenotype and function, much like the human CD4⁺ T cells that have been seen following EBV-LCL stimulation in vitro (40–43). Our studies further revealed that LMP1⁺ B cells upregulate an array of adhesion and costimulatory molecules—including CD54 (ICAM-1), CD80 (B7), CD70 (CD27 ligand), OX40 ligand (OX40L), and 4–1BB ligand (4–1BBL)—and drive the differentiation and expansion of Eomesodermin (EOMES)-programmed CD4⁺ CTLs via CD70- and OX40L-mediated costimulation, and of CD8⁺ CTLs via CD70, and OX40L, as well as 4–1BBL (7) (see Figure 1). The relevance of these findings is corroborated by the reported expression of these costimulatory ligands in EBV-LCLs (44, 45).

Figure 1. Schema of how LMP1 signaling in B cells induces cytotoxic CD4⁺ and CD8⁺ T cell responses to TAAs.

LMP1 signaling in B cells induces massive cellular gene expression. This leads to (1) upregulation of cellular machinery involved in antigen processing and presentation, (2) upregulation of costimulatory ligands (CD70, OX40L and others), and (3) overexpression of many cellular antigens known to function as TAAs. Presentation of these TAAs and simultaneous costimulation through CD70 and OX40L drive cytotoxic CD4⁺ and CD8⁺ T cell responses. Note that EBV viral antigens and their responding T cells are not depicted. Figure adapted from ref. 12; with permission.

Of note, LMP1 has been characterized as a functional analogue of constitutively active CD40 (46, 47), a major signaling pathway for the functional maturation of APCs. Although CD40 signaling in B cells also enhances antigen presentation and costimulation through B7 molecules (48), we found that CD40-activated B cells exhibited little induction of OX40L and no induction of CD70, and co-culture with naive CD4⁺ cells led to no generation of EOMES⁺ CD4⁺ CTLs, in sharp contrast to LMP1⁺ B cells (12). Further studies are needed to understand the differences between these two signaling pathways and gain mechanistic insights into the expression regulation of these costimulatory ligands.

EBV and T cell responses to TAAs

TAAs

Tumor-associated antigens comprise a large set of non-mutated cellular antigens identified as T cell targets in human and murine cancers (49–51). TAAs can be categorized into several subclasses, including cancer germline gene-encoded antigens, differentiation antigens, and overexpressed antigens (13, 50, 51). TAA-specific T cells have been known for three decades, yet their origins remained elusive.

EBV and origins of TAA-specific T cells

While tumor cells may be an important source of TAAs for T cell priming (52), our recent findings in the LMP1 mouse model indicate that EBV, through its signaling protein LMP1, induces potent cytotoxic CD4⁺ and CD8⁺ T cell responses against a wide range of TAAs (12). Independent studies, showing that the human CD4⁺ T cell response to LCL stimulation in vitro contains a non-viral, cellular antigen-specific, component (53, 54), are consistent with this view. In particular, our data show that LMP1 signaling in B cells leads to overexpression of many cellular antigens known to function as TAAs and to their presentation via both MHC-I and MHC-II pathways. Together with LMP1-induced costimulatory ligands, in particular CD70 and OX40L, this provokes cytotoxic CD4⁺ and CD8⁺ T cell responses to these TAAs (see Figure 1). These findings raise new notions concerning tumor immunity and viral immunity: (i) tumor cells are not the sole source of TAAs, as some TAA-specific T cells can be elicited by EBV infection through the above-described mechanism; (ii) a virus, such as EBV, may induce T cell responses not only against viral antigens but also certain cellular antigens, specifically TAAs.

Further efforts are needed to identify TAAs expressed by EBV-infected/transformed B cells and to demonstrate their recognition by T cells in EBV-infected individuals. For these kinds of studies, identifying EBV-infected subjects at the right time may be critical, considering that TAA (cellular antigen)-specific T cells might be subject to much tighter control by immune tolerance mechanisms than EBV (foreign antigen)-specific T cells, and thus the former might soon be severely outnumbered by the latter. This may be particularly relevant to CD8⁺ T cells, given the known massive expansion and dominance of EBV-specific CD8 clonotypes in the most often studied donors, including IM patients and healthy carriers. In contrast, the CD4 compartment usually displays a polyclonal profile, reflecting small clonal burst sizes of CD4⁺ T cells to individual antigens, including EBV antigens (55, 56). In this regard, subjects at early stages of infection, in which mysterious “bystander T cells” are reportedly dominant over EBV-specific T cells (57, 58), may be particularly suitable for studying TAA-specific CD8⁺, as well as CD4⁺, T cell responses.

Future Directions

The latest insights discussed above should spur new efforts to look into the potential impact of EBV on human cancer immunity and immunotherapy, and ultimately the relationship of EBV with humans. In addition, the EBV’s immunostimulatory molecule LMP1 offers a tool for developing new immune approaches for cancer therapy.

EBV-induced TAA-specific T cells in cancer immune surveillance and immunotherapy

Data from human in vitro studies (53, 54) and mouse models (12, 54) have indicated the potential of EBV (LMP1)-induced TAA-specific effector T cells in killing/controlling EBV-related as well as EBV-unrelated malignant B cells (through targeting shared TAAs). These T cells, presumably a part of the LCL-stimulated T cell product for treatment of EBV-related malignancies, should also contribute to therapeutic efficacies in patients. This warrants further investigation, particularly given that one of the clinical studies reported detection of T cell responses against several TAAs, which appeared to correlate with clinical responses (35). It remains unclear whether those T cells were produced by LCL stimulation and thus present in the therapeutic T cells or generated through epitope spreading, as postulated in the study.

On the other hand, about 95% of humans have had EBV infection by adulthood, and thus should carry TAA-specific memory T cells, yet a sizable fraction develop B cell malignancies and other cancers, which likely express some numbers of shared TAAs. This would suggest that EBV-induced TAA-specific memory T cells offer little or no tumor protection or prevention. A main issue could be that the function of these T cells is suppressed by immune-checkpoint mechanisms (59), as known for other TAA-specific T cells in cancer patients (60). Therefore, any studies aimed at understanding the impact of EBV-induced TAA-specific T cell immunity on human cancers should also look at the functional states of these T cells.

Epidemiological studies probing the protective effect of EBV-induced TAA-specific effector (functionally active) T cells on cancers that co-occur with primary EBV infection is possible, but subjects with this condition are rare, and therefore large sample sizes would be needed.

Meanwhile, comparative study of TAA-specific T cells in cancer patients before and after immune-checkpoint blockade therapies may be more convenient, especially if an inventory of EBV-induced TAAs is available (discussed above) as reference. These efforts may reveal that although EBV-induced TAA-specific memory T cells lose vigor to control tumor development, they can be reactivated therapeutically for tumor treatment. If so, the findings should also lead to better understanding of the immune effectors underlying the checkpoint blockade therapies, and may inform improved strategies to reinvigorate those TAA-specific T cells for cancer therapy.

Furthermore, patient-derived xenograft (PDX) models may provide another avenue to study EBV-induced TAA-specific T cells in various human cancers. This is inspired by the following findings: (i) attempts to generate PDX lines by transplanting EBV-unrelated tumors, particularly carcinomas, into immunodeficient mice, led in ~30% of cases to the outgrowth of EBV-transformed B cells (reflecting infiltration of EBV-carrying B cells in the original biopsy) (61, 62); and (ii) detailed analyses in one such case demonstrated that the initial tumor-infiltrating (TIL) T cells also infiltrated and clonally expanded in the EBV⁺ B cell tumor, and some of the T cells seemed to target certain shared tumor-associated antigens (63).

Relationship of EBV with humans

EBV is an ancient virus, which has co-evolved with humans for millions of years (64). Thanks to five-decades of research on EBV, we know that (i) this virus infects almost every person on earth; (ii) primary infection typically occurs in early childhood and is usually asymptomatic, but when delayed into adolescence or young adulthood (due to modern lifestyle changes), may cause a self-limiting lymphoproliferative illness known as IM; (iii) after the acute phase of infection, the virus persists in the hosts throughout life in an equilibrium state, but can cause cancers when, in concert with other pre-neoplastic changes, immune surveillance fails due to immunosuppression by drugs, HIV, malaria, or other means. If future research, as discussed above, does establish a protective role of EBV-induced immunity against other cancers, we would have to re-think our relationship with this virus, and how we deal with this virus for our best benefit. Presently, several countries, including the US, are putting forward efforts to develop EBV vaccines (65). Their use in seronegative individuals at high risk for IM (such as young adults) or EBV-associated malignancies (such as transplant recipients or HIV-infected people) may be well justified. However, we caution that a thorough understanding of the impact of EBV-induced immunity on cancer immune surveillance and immunotherapy is needed before rolling out prophylactic vaccination to the general population, i.e., children at early ages (66, 67).

LMP1-based immunotherapeutic approaches

The revelation of the immunostimulatory function of LMP1 opens the possibility to exploit it to develop new immune approaches for cancer therapy. One approach, as described in our recent work, is a CD4⁺ CTL approach for treating B cell malignancies (12). In this approach, ectopic expression of LMP1 in tumor B cells will (i) enhance presentation of a broad array of endogenous antigens, including TAAs and neoantigens, on MHC-II, and (ii) concurrently upregulate costimulatory ligands CD70 and OX40L, thereby eliciting CD4⁺ CTLs against these tumor antigens. CD4⁺ CTLs generated in this fashion are able to kill unmodified tumor B cells that express the same antigens, and thus can be used for adoptive cell therapy (see Figure 2). In principle, CD8⁺ T cells targeting tumor endogenous TAAs and neoantigens can also be generated using a similar strategy. We are prioritizing developing and testing the CD4⁺ CTL approach for B cell malignancies, because (i) many B cell tumors in patients, including more than 70% of diffuse large B cell lymphomas and classical Hodgkin lymphomas, completely lose MHC-I expression (68, 69) and thus cannot be targeted by CD8⁺ CTLs; (ii) CD4⁺ CTLs have shown superior therapeutic efficacy over CD8⁺ CTLs in LMP1-driven B cell lymphoma models (7); and (iii) besides cytolytic activity, CD4⁺ CTLs maintain helper functions that support their own in vivo persistence providing long-lasting immunity. CD4⁺ T cell’s superior ability for in vivo persistence is also highlighted by the recent finding in long-surviving patients treated with CD19-directed chimeric antigen receptor (CAR)-T cells (70).

Figure 2. Schematic of the CD4⁺ CTL therapeutic strategy for B cell malignancies.

Note that LMP1 signaling in tumor B cells will lead to (1) enhanced processing and presentation of endogenous antigens on MHC-II, and (2) upregulation of costimulatory ligands, including CD70 and OX40L that drive cytotoxic CD4⁺ T cell differentiation. See text for details. Figure adapted from ref. 12; with permission.

CD4⁺ CTL-based therapy may also be applicable to other cancers that express MHC-II in an inducible manner (such as upon exposure to interferon-γ), as indicated by recent preclinical (71) and clinical studies (60, 72, 73) in several types of cancer. Besides direct cytotoxicity, CD4⁺ CTLs may exert helper functions via interaction with APCs to enlist other immune effectors—such as CD8⁺ T cells (74, 75), tumoricidal macrophages and natural killer cells (76, 77)—to tumor immunity, and thereby can protect against MHC-II–negative tumors. Furthermore, CD4⁺ cells specific for TAAs and neoantigens are often readily detectable in tumor patients (60, 78–81). These findings together encourage developing multiantigen-targeted CD4⁺ CTL approach for non-B cell cancers.