|Home | About | Journals | Submit | Contact Us | Français|
Epstein-Barr virus (EBV) infection is associated with B cell lymphomas in humans. The ability of EBV to convert human B cells into long-lived lymphoblastoid cell lines (LCLs) in vitro requires the collaborative effects of EBNA2 (which hijacks Notch signaling), latent membrane protein 1 (LMP1) (which mimics CD40 signaling), and EBV-encoded nuclear antigen 3A (EBNA3A) and EBNA3C (which inhibit oncogene-induced senescence and apoptosis). However, we recently showed that an LMP1-deleted EBV mutant induces B cell lymphomas in a newly developed cord blood-humanized mouse model that allows EBV-infected B cells to interact with CD4 T cells (the major source of CD40 ligand). Here we examined whether the EBV LMP2A protein, which mimics constitutively active B cell receptor signaling, is required for EBV-induced lymphomas in this model. We find that the deletion of LMP2A delays the onset of EBV-induced lymphomas but does not affect the tumor phenotype or the number of tumors. The simultaneous deletion of both LMP1 and LMP2A results in fewer tumors and a further delay in tumor onset. Nevertheless, the LMP1/LMP2A double mutant induces lymphomas in approximately half of the infected animals. These results indicate that neither LMP1 nor LMP2A is absolutely essential for the ability of EBV to induce B cell lymphomas in the cord blood-humanized mouse model, although the simultaneous loss of both LMP1 and LMP2A decreases the proportion of animals developing tumors and increases the time to tumor onset. Thus, the expression of either LMP1 or LMP2A may be sufficient to promote early-onset EBV-induced tumors in this model.
IMPORTANCE EBV causes human lymphomas, but few models are available for dissecting how EBV causes lymphomas in vivo in the context of a host immune response. We recently used a newly developed cord blood-humanized mouse model to show that EBV can cooperate with human CD4 T cells to cause B cell lymphomas even when a major viral transforming protein, LMP1, is deleted. Here we examined whether the EBV protein LMP2A, which mimics B cell receptor signaling, is required for EBV-induced lymphomas in this model. We find that the deletion of LMP2A alone has little effect on the ability of EBV to cause lymphomas but delays tumor onset. The deletion of both LMP1 and LMP2A results in a smaller number of lymphomas in infected animals, with an even more delayed time to tumor onset. These results suggest that LMP1 and LMP2A collaborate to promote early-onset lymphomas in this model, but neither protein is absolutely essential.
Epstein-Barr virus (EBV) is a human herpesvirus that contributes to a variety of human B cell lymphomas, including lymphoproliferative disease (LPD) in immunocompromised hosts, diffuse large B cell lymphoma (DLBCL), Burkitt lymphoma (BL), and Hodgkin lymphoma (HL) (1, 2). EBV-infected tumors are composed largely of latently infected cells, in which the virus persists as a nuclear episome and is replicated by the host cell DNA polymerase (3). Several different forms of viral latency, which differ in regard to the number of viral genes expressed, can occur in EBV-positive human lymphomas (3). However, only type III EBV latency can convert primary B cells in vitro into long-term lymphoblastoid cell lines (LCLs). Nevertheless, this form of latency, which allows the expression of each of the nine viral latency proteins (plus the small EBV-encoded nuclear RNAs [EBERs] and virally encoded microRNAs), is also the most immunogenic form and thus is usually restricted to tumors of immunosuppressed patients.
The generation of EBV-transformed LCLs in vitro requires both EBV-encoded nuclear antigens (EBNAs), including EBNA1, EBNA2, EBNA3A, and EBNA3C, and latent membrane protein 1 (LMP1) (3). The cellular gene expression pattern in EBV-driven LCLs largely reflects the transcriptional effects of the EBNA2 and LMP1 proteins (4). EBNA2 interacts directly with the cellular protein RBP-Jκ (CBF1) to mimic the effect of constitutive Notch signaling and promote B cell proliferation (5, 6). EBNA2 (directly or indirectly) activates the expression of c-Myc, cyclin D2, and E2F1 in B cells, and c-Myc expression is required for the proliferation of LCLs (7, 8). LMP1 mimics the effect of constitutively active CD40 signaling, thereby activating the NF-κB, AP1, and ATF2 transcription factors and inhibiting apoptosis (9,–12).
Although the establishment of long-term LCLs in vitro requires LMP1 expression, the rapid proliferation of B cells during the first week of EBV infection in vitro is driven largely by EBNA2 (13). During this initial “proliferative” period, EBV-infected cells replicate more rapidly (dividing every 12 h) than at later times (dividing every 24 h) and do not express appreciable amounts of LMP1 or NF-κB (13). Thus, EBNA2 can drive B cell proliferation in the absence of LMP1. The EBNA3A and EBNA3C proteins, which collaboratively turn off the expression of the tumor suppressor protein p16 (14, 15) and the proapoptotic protein BIM1 (16, 17), are also required for long-term LCL outgrowth, as is EBNA1, which mediates the replication of the latent viral genome (3).
Another EBV-encoded protein, LMP2A, could potentially be required for EBV-induced lymphomas in humans, even though it is largely dispensable for EBV-induced B cell transformation in vitro. LMP2A mimics the effect of low-level constitutive B cell receptor (BCR) signaling (18,–22), and EBV-negative B cell lymphomas often have cellular mutations or other abnormalities that lead to constitutive BCR signaling (23, 24). Furthermore, in transgenic mouse models, LMP2A promotes the survival of B cells in the periphery that have not undergone a productive BCR rearrangement (21) and cooperates with c-Myc to induce Burkitt-like B cell lymphomas (25). LMP2A is also required for the ability of EBV to transform germinal center (GC)-derived B cells in vitro that have not undergone a productive BCR rearrangement (26).
Although EBV efficiently infects many different types of B cells in vitro and in vivo, including naive B cells, GC B cells, and post-germinal center memory B cells (3, 27, 28), long-term latent EBV infection appears to be restricted to the memory B cell compartment in immunocompetent humans (3, 29). One model to explain this phenomenon proposes that EBV infection of naive B cells promotes a GC-like reaction (via the effects of LMP1 and LMP2A) that converts naive B cells into memory-like B cells even in the absence of a true (i.e., helper T cell-dependent) GC reaction (3, 27). Another model is based upon in vitro studies showing that EBV infection of naive B cells induces T cell-independent somatic hypermutation (SHM) (but not class switching) by inducing the expression of activation-induced cytosine deaminase (AID) (27). This model proposes that EBV-infected B cells that have undergone GC-independent SHM have a selective survival advantage in vitro and in vivo (27). Nevertheless, a subset of EBV-positive posttransplant lymphoproliferative disease (PTLD) lesions are derived from naive B cells (30), indicating that EBV can also successfully establish long-term latent infection in naive B cells of immunocompromised hosts. In addition to preferentially establishing latency in long-lived memory B cells, another potential mechanism by which EBV may ensure the prolonged survival of latently infected B cells is to inhibit terminal plasma cell differentiation (31,–33).
Animal and cell models that can examine how EBV infection affects different aspects of normal B cell biology have been difficult to develop and often give contradictory results. For example, while LMP2A expression in transgenic mice has been reported to promote autoimmunity and plasma cell differentiation (34, 35), and LMP1 expression can lead to lymphomas (11, 36), the combination of LMP1 and LMP2A together in transgenic mice normalizes the phenotypes induced by the expression of LMP1 and LMP2A alone (37). Thus, the effects of LMP1 and LMP2A on EBV pathogenesis and B cell biology may be most biologically relevant when studied in the context of the intact virus using models that allow B cells and T cells to interact.
Here we have used a newly developed cord blood-humanized mouse model to examine the effects of the loss of LMP2A (in the presence and absence of the concomitant loss of LMP1) on EBV-induced lymphoma formation and to ask whether EBV infection of naive (cord blood-derived) B cells regulates their ability to undergo plasma cell differentiation. Using this model, we recently showed that LMP1 is unexpectedly dispensable for EBV-induced lymphoma formation, since CD40L-expressing human CD4 T cells can partially substitute for LMP1 (38). Our results here suggest that neither LMP1 nor LMP2A is absolutely required for EBV-induced lymphomas in this model, even when both proteins are simultaneously deleted. Nevertheless, the deletion of LMP2A alone delays the onset of lymphomas, and the deletion of both LMP1 and LMP2A together reduces the number of EBV-infected animals that develop lymphomas and further delays the onset of tumor development. Furthermore, we find that EBV infection strongly inhibits plasma cell differentiation in this model and show that this effect does not require the expression of either LMP1 or LMP2A. Together, these findings suggest that the cord blood-humanized mouse model may be useful for dissecting how EBV proteins (including the nuclear EBNA proteins) and/or virally encoded small RNAs contribute to EBV-induced lymphomas in humans and for understanding how EBV infection regulates B cell biology and differentiation.
We used a cord blood-humanized mouse model to determine if LMP2A is required for EBV-induced lymphomas in the presence, or absence, of LMP1 expression. CD34-depleted human cord blood cells were either mock infected or infected for 1.5 h in vitro with 10,000 infectious particles (green Raji cell units [GRUs]) of wild-type (WT) EBV (B95.8 strain bacmid), an EBV mutant (B95.8 strain) deleted for the LMP2A gene (LMP2A-KO), or an EBV mutant (B95.8 strain) deleted for both the LMP2A and LMP1 genes (LMP1/2A-KO) and then injected intraperitoneally (i.p.) into NSG (NOD/LtSz-scid/IL2Rnull) mice. We previously showed that almost all WT EBV-infected mice in this model develop activated DLBCLs with type III EBV latency, which often invade the pancreas, liver, and gallbladder (38, 39). In contrast, mice injected with uninfected cord blood consistently engraft human T cells, but only a portion of mice sustain long-term coengraftment of human B cells, and they do not develop B cell lymphomas.
As shown in Fig. 1A, as expected, almost all mice injected with WT EBV-infected cord blood developed tumors, whereas injection of uninfected cord blood did not result in any tumors. Mice injected with LMP2A-KO virus-infected cord blood cells developed lymphomas at a frequency similar to that for WT virus-infected animals (Fig. 1A), although the tumors derived from this mutant often occurred at a later time point than did the WT virus-induced tumors (Fig. 1B). Somewhat surprisingly, a substantial subset of mice (47%) infected with an EBV mutant missing both LMP2A and LMP1 (LMP1/2A-KO virus) also developed lymphomas, although the number of animals developing lymphomas was significantly smaller than the number of animals developing lymphomas following infection with the WT or LMP2A-KO virus (Fig. 1A). Tumors derived from the LMP1/2A-KO double mutant virus also occurred at later time points than did tumors derived from the LMP2A mutant or the WT virus (Fig. 1B). Animals injected with either LMP2A-KO virus-infected or LMP1/2A-KO virus-infected cord blood were also substantially more likely to sustain long-term B cell engraftment (assessed by CD20 immunohistochemistry [IHC] at days 50 to 60 post-cord blood injection) than were mice injected with uninfected B cells (Fig. 1C).
To confirm that the LMP1/2A-KO virus used in these studies is profoundly defective for transforming B cells in vitro, the outgrowths of peripheral B cells infected with two different amounts of the WT, LMP2A-KO, or LMP1/2A-KO virus were compared. Immunoblot analysis of infected B cells in vitro confirmed that WT virus-infected cells, but not LMP2A-KO or LMP1/2A-KO virus-infected cells, expressed LMP2A and that WT virus- and LMP2A-KO virus-infected cells, but not LMP1/2A-KO virus-infected cells, expressed LMP1 (data not shown). As expected, the LMP1/2A-KO virus was unable to produce long-term LCLs, and cells infected with this virus had much decreased proliferation compared to that of WT virus-infected cells (Fig. 2A). Interestingly, the LMP2A-KO virus was also partially defective for inducing B cell proliferation and transformation in vitro in comparison to equal titers of the WT virus (Fig. 2A).
To demonstrate that tumors obtained in vivo with the LMP2A-KO and LMP1/2A-KO viruses were not contaminated with the WT virus, tumor DNA was isolated from paraffin-fixed slides, and PCR analysis was used to determine if either the LMP1 or LMP2A gene was present (Fig. 2B). These results showed that no detectable WT virus contaminated the tumors derived from the mutant viruses. Thus, neither LMP1 nor LMP2A is absolutely essential for the ability of EBV-infected B cells to stably engraft, or cause invasive lymphomas, in the cord blood-humanized mouse model.
To determine if the loss of either LMP2A alone or LMP1 and LMP2A together affects the phenotype of EBV-induced lymphomas in the cord blood-humanized mouse model, tumors were stained with hematoxylin and eosin (H&E) as well as a variety of different antibodies to assess the B cell differentiation state and viral expression pattern. Results are summarized in Table 1. All of the tumors obtained with the WT virus or the LMP2A mutant virus had aggressive DLBCL-like lymphomas, showing diffuse sheets of large, atypical lymphoid cells with readily apparent mitotic figures (Fig. 3). In contrast, while many of the lymphomas derived from the LMP1/2A-KO mutant also appeared similar to aggressive DLBCLs, some of the lesions derived from this mutant were more similar to polymorphic PTLD lesions, displaying a mixture of large- and medium-sized atypical lymphoid cells interspersed with plasma cells (Fig. 3, bottom right, and Table 1). The majority of tumors were intraparenchymal in the pancreas, often with extension into adjacent liver tissue and forming masses in the mesentery/mesothelium. Some tumors also involved the white pulp of the spleen and perivascular areas of the lung.
The WT EBV-infected, LMP2A-KO virus-infected, and LMP1/2A-KO virus-infected tumors each expressed CD20 (confirming B cell identity) and the EBV latency protein EBNA2 (confirming type III latency) (Fig. 4A and andB).B). All tumors (regardless of whether they were infected with WT EBV or mutant viruses) also expressed interferon regulatory factor 4 (IRF4) (Fig. 4A), a marker for early plasma cell differentiation and a surrogate immunohistochemical marker for the “activated B cell type” of DLBCLs (40, 41). IRF4 is also an essential survival factor for activated B cell lymphomas in humans (41) and for LCLs in vitro (42, 43). In contrast to lymphoma cells, T cells infiltrating the tumors had very little IRF4 expression (Fig. 4A). WT and LMP2A-KO virus-infected lymphomas also expressed similar levels of LMP1 (Fig. 4B and andCC).
Since EBNA2 activates the expression of the potent cellular oncogene c-Myc in LCLs in vitro, we next examined whether c-Myc is expressed in WT or mutant virus-infected DLBCLs in the cord blood-humanized mouse model. As shown in Fig. 5 (left), high c-Myc levels were found in a subset WT, LMP2A-KO, and LMP1/2A-KO virus-infected lymphoma cells. Given the ability of EBNA3A and EBNA3C to inhibit the expression of the proapoptotic protein BIM1 (which plays a critical role in preventing c-Myc-induced B cell lymphomas) (17, 44) in vitro, we next determined if B cell lymphoma cells infected with the WT virus, LMP2A-KO virus, or LMP1/2A-KO virus lost the expression of BIM1. As shown in Fig. 5 (right), both WT virus-infected and LMP2A-KO virus-infected lymphomas contained no cells that costained for both EBNA2 and BIM1, although BIM1 expression was observed in the surrounding uninfected cells (presumably T cells). In the case of the LMP1/2A-KO virus-infected lymphomas, most EBNA2-positive cells likewise contained no detectable BIM1, although a minority of cells that had a lower level of EBNA2 expression expressed BIM1. These results suggest that LMP2A-KO and LMP1/2A-KO virus-induced lymphomas in cord blood-humanized mice may be promoted and sustained by both increased c-Myc expression and decreased BIM1 expression, two properties that are known to be conferred in vitro by the EBNA2 and EBNA3A/3C proteins, respectively.
Whether LMP2A primarily promotes or inhibits lytic EBV reactivation in infected human B cells in vivo is not totally clear, since LMP2A has been reported to inhibit BCR-mediated lytic viral reactivation of LCLs in vitro (45) but to enhance constitutive lytic gene expression when expressed in certain EBV-infected Burkitt lines (46). To determine if constitutive LMP2A expression (with or without concomitant LMP1 expression) in the context of the intact viral genome modulates the level of lytic EBV gene expression in the cord blood-humanized mouse model, we stained tumors with an antibody (Ab) directed against the viral immediate early lytic protein BZLF1. In the context of the B95.8 EBV strain, which we previously showed has very little lytic protein expression in humanized mice (38, 39, 47, 48), the WT-, LMP2A-KO-, and LMP1/2A-KO-induced lymphomas had few, if any, cells expressing BZLF1 (Fig. 6A and andC).C). This result suggests that LMP2A is not required for the inhibition of lytic viral reactivation in the cord blood-humanized mouse model.
Since the M81 strain has been shown to be much more lytic than the B95.8 strain when injected into humanized mice (49), we also examined the expression of two different lytic EBV proteins (BZLF1 and BMRF1) in lymphomas infected by either the WT or LMP2A-deleted M81 strain of EBV. Although we confirmed that WT M81 EBV-infected lymphomas contain numerous lytically infected cells, we did not find that a loss of LMP2A expression in the context of the EBV M81 strain altered the amount of BZLF1 or BMRF1 expression in lymphomas (Fig. 6B, ,D,D, and andE).E). Thus, in contrast to previously reported in vitro results, endogenous LMP2A expression in the context of the intact viral genome does not appreciably increase, or decrease, the level of lytic EBV infection in the cord blood-humanized mouse model.
We previously showed that T cells infiltrate WT EBV-infected lymphomas in the cord blood-humanized mouse model and demonstrated that immune checkpoint blockade increases the ability of cord blood-derived T cells to infiltrate these lymphomas and slow their growth (39). To determine if the loss of expression of LMP2A and/or the loss of expression of both LMP1 and LMP2A alters the amount of T cell infiltration in EBV-infected lymphomas, lymphomas were stained with antibodies that detect total T cells (CD3), cytotoxic T cells (CD8), or helper T cells (CD4). WT EBV-infected, LMP2A-KO virus-infected, and LMP1/2A-KO virus-infected lymphomas were similarly infiltrated by CD3-positive, CD8-positive, and CD4-positive T cells (Fig. 7). These results suggest that neither LMP2A nor LMP1 grossly alters T cell infiltration of EBV-infected lymphomas in this particular humanized mouse model.
We previously showed that coinjected CD4 positive T cells are required for the ability of LMP1-deleted EBV, but not wild-type EBV, to cause lymphomas in the cord blood-humanized mouse model and demonstrated that CD40L-producing CD4 T cells can substitute for LMP1 in this model (38). To determine if T cells are required for the development LMP2A-KO-induced lymphomas in this model, mice were injected with LMP2A-KO virus-infected cord blood and then treated with or without a T cell-depleting antibody (OKT3) starting at day 4 after injection. As shown in Fig. 8, the T cell-depleting antibody did not inhibit the ability of the LMP2A-KO virus to establish lymphomas in cord blood-humanized mice, presumably because LMP1 expression substitutes for growth-promoting signals derived from helper CD4 T cells. In contrast, consistent with our previously reported results using the single LMP1-KO virus (38), none of the 10 animals infected with the LMP1/2A-KO virus developed lymphomas when treated with the OKT3 Ab (data not shown).
Finally, given the previously reported ability of EBV to inhibit plasma cell differentiation in vitro (31,–33), we asked whether EBV infection prevents plasma cell differentiation in the cord blood-humanized mouse model. Splenic B cell follicles from cord blood-humanized mice that were mock infected or infected with the WT, LMP2A-KO, LMP1-KO, or LMP1/2A-KO virus were stained with CD20 antibody, CD138 or BLIMP1 antibodies (plasma cell markers), or (in the case of EBV-infected animals) EBERs (EBV infection markers). As shown in Fig. 9, splenic follicles containing uninfected B cells expressed high levels of CD138 and had occasional BLIMP1-expressing cells, whereas splenic follicles that were composed largely of WT, LMP2A-KO, LMP1-KO (not shown), or LMP1/2A-KO virus-infected B cells expressed little or no CD138 or BLIMP1. These results suggest that EBV infection inhibits plasma cell differentiation in the cord blood-humanized mouse model, even in the absence of LMP1 and LMP2A expression.
In this paper, we used a cord blood-humanized mouse model to examine whether EBV-induced lymphomas in this model require the expression of the EBV LMP2A latency protein. We recently showed that an LMP1-deleted EBV mutant can induce invasive lymphomas in this model, as long as CD4 T cells are present to provide an alternative source of CD40 signaling (38). Since LMP2A mimics the effect of constitutive BCR signaling, and constitutive BCR signaling is commonly selected for in EBV-negative human B cell lymphomas, we expected that the loss of LMP2A expression might inhibit the formation of EBV-induced lymphomas in this model. Instead, we show here that the loss of LMP2A alone has little effect on the ability of EBV to form B cell lymphomas, other than increasing the time to tumor onset, and demonstrate that even an EBV mutant simultaneously deleted for both the LMP1 and LMP2A genes still induces invasive lymphomas in a subset of animals. These results suggest that EBV-infected B cells may obtain growth and/or survival signals in vivo that allow the virus to induce invasive DLBCLs in the absence of the expression of both LMP1 and LMP2A. Nevertheless, our finding that the simultaneous deletion of both LMP1 and LMP2A decreases the percentage of animals developing EBV-positive tumors, and substantially increases the time to tumor onset, suggests that the expression of either LMP1 or LMP2A accelerates the onset of virally associated lymphomas and increases the efficiency of this process.
The ability of EBV to transform primary B cells in vitro into long-lived LCLs has proven to be an invaluable tool for dissecting mechanisms by which various different EBV proteins might contribute to EBV-associated human lymphomas. Nevertheless, in vitro transformation studies may not recapitulate certain critical aspects of EBV-associated malignancies in humans. For example, the supporting roles of the in vivo tumor microenvironment, and the various different types of interactions between EBV-infected B cells and T cells, are difficult to model by using in vitro systems. Furthermore, although EBV infection of primary B cells in vitro initially induces the rapid proliferation of essentially all infected cells (13), most of these infected cells die after a few weeks via poorly understood mechanisms, and only a small minority of EBV-infected cells actually have the capacity to become long-term LCLs. Whether a similar massive die-off of EBV-infected B cells occurs in vivo (in the absence of immune-mediated killing) is not known; if not, then certain viral proteins required to sustain long-term LCLs in vitro may not be required for the ability of EBV to form lymphomas in vivo. Finally, the ability of EBV-infected B cells to invade organs also cannot be easily studied in vitro.
We initiated these studies hoping to develop an in vivo system in which the contributions of LMP2A (within the context of the intact viral genome) to the establishment of persistent EBV infection, regulation of EBV lytic reactivation, and/or EBV-induced transformation of B cells might be readily apparent. A number of potential roles for LMP2A in enhancing EBV persistence and/or EBV-associated lymphomas have been proposed, which include regulating the latent-to-lytic switch (45, 46), inducing plasma cell differentiation (35), promoting the survival of B cells that have undergone nonproductive BCR rearrangements (21), cooperating with c-Myc overexpression to induce Burkitt-like lymphomas (50, 51), and inhibiting the host immune response (52, 53). However, many of those previous studies expressed LMP2A at nonphysiological levels and/or were performed outside the context of the intact viral genome. In contrast, our studies here examined the effects of endogenous LMP2A expression in the context of the intact viral genome and in the presence and absence of human T cells.
We did not find that the loss of LMP2A expression alone substantially alters the phenotype of EBV-infected lymphomas in this cord blood-humanized mouse model. Almost all EBV-induced lymphomas in this model, in the presence or absence of LMP1 and/or LMP2A expression, were activated DLBCLs that expressed IRF4 but little if any CD138 (a marker for more differentiated plasma cells). Thus, in contrast to the results of transgenic mouse studies, we did not find that LMP2A enhances plasma cell differentiation in the context of the intact viral genome in this cord blood-humanized mouse model. Of note, however, in contrast to the single LMP1- and LMP2A-deleted lymphomas, a subset of the LMP1/2A-KO double mutant virus-induced lymphomas were more similar to PTLD-like lesions than invasive DLBCLs. These results suggest that the expression of both LMP1 and LMP2A promotes the more invasive DLBCL phenotype in this model.
In addition, in contrast to some previously reported in vitro studies, we did not find that endogenous LMP2A expression in the context of the intact viral genome significantly impacts the amount of lytic viral gene expression in the cord blood-humanized mouse model. We examined the effect of the loss of LMP2A on lytic viral protein expression using two different viral strains (the largely latent B95.8 strain and the highly lytic M81 strain), but in both cases, we did not observe any effect on lytic viral protein expression. It remains possible that other viral proteins (such as LMP1 and/or one or more EBNA proteins) have stronger effects on lytic reactivation than LMP2A and that the effect(s) of LMP2A on lytic viral reactivation is restricted to cells that have stringent (type I) viral latency. In addition, EBV-infected B cells in this model may have constitutive BCR signaling in response to foreign tissue antigens, in which case the ability of LMP2A to induce further BCR stimulation might be minimal.
Although LMP2A has been reported to activate survival pathways in vitro, including the phosphatidylinositol 3-kinase (PI3K) and NF-κB pathways (54, 55), we did not find that the deletion of LMP2A alone substantially affected the number of EBV-induced lymphomas in this model, although the onset of lymphomas was somewhat delayed. Since LMP1 is also a potent activator of both canonical and noncanonical NF-κB, as well as PI3K (56), it may at least partially substitute for the survival effects of LMP2A. In addition, EBV-infected B cells can be activated by CD40L expressed on CD4 T cells in this model, which would likewise result in NF-κB and PI3K signaling. Although the LMP2A-KO virus induced lymphomas in T cell-depleted animals, as expected, the LMP1/2A-KO virus could not induce lymphomas in T cell-depleted animals. These results suggest that CD40L signaling from CD4-positive T cells is not required to sustain LMP2A-KO-induced tumors unless LMP1 is also deleted.
Although LMP2A has been reported to decrease the expression of major histocompatibility complex (MHC) class II (53) as well as the expression of the NKG2D receptor ligands MICA and ULBP4 (52), the ability of T cells to control WT virus- versus LMP2A-KO virus-induced tumors was not obviously different in the cord blood-humanized mouse model. In both cases, CD4 and CD8 T cells were able to infiltrate the tumors but did not ultimately control them. Although we have recently shown that T cells have some ability to control EBV-induced lymphomas in the cord blood-humanized mouse model, particularly in the presence of immune checkpoint blockade (39), the potential immune-evasive roles of LMP2A might be more clearly observed by using more immunocompetent humanized models containing coengrafted human thymic tissue.
As suggested by several different in vitro studies (31,–33), we confirm here that EBV infection in the cord blood-humanized mouse model inhibits the ability of B cells to differentiate into plasma cells. Although overexpressed LMP1 has been reported to inhibit plasma cell differentiation in vitro by inhibiting BLIMP1 activity (32), we found here that the LMP1/2A-KO virus also inhibited plasma cell differentiation. This result may reflect the recent finding that EBV-encoded BHRF1 microRNAs (which are expected to be expressed in the LMP1/2A-KO mutant) also inhibit plasma cell differentiation in vitro (33). It is possible that one or more EBNA proteins also contribute to this phenotype. BLIMP1 is the master regulator of plasma cell differentiation, and BLIMP1-inactivating mutations were recently shown to be very common in activated DLBCLs in humans (57). Thus, the ability of EBV to inhibit terminal plasma cell differentiation likely plays an important oncogenic role in this cord blood-humanized model for EBV-induced DLBCLs. In addition, since plasma cell differentiation induces the lytic form of EBV infection (58), inhibition of plasma cell differentiation may be required to maintain long-term viral latency in humans.
Finally, the fact that the LMP1/2A-KO virus causes highly aggressive and invasive lymphomas in a portion of cord blood-humanized mice potentially makes this an attractive model to dissect the roles of specific EBV EBNA proteins (or EBV-encoded EBERs and/or viral microRNAs) for EBV-induced lymphoma formation in vivo. Our findings that lymphomas induced by the LMP1/2A-KO virus have high c-Myc expression levels as well as decreased BIM1 expression (similar to the phenotype of WT-induced tumors) suggest likely mechanisms by which this mutant virus promotes B cell lymphomas in vivo and are consistent with the known ability of EBNA2 and EBNA3A/3C, respectively, to activate c-Myc and inhibit BIM1 expression in vitro.
Wild-type EBV strains B95.8 (GenBank accession number NC_007605.1) and M81 (GenBank accession number KF373730.1) are available as recombinant bacmids. The LMP1-KO virus (missing amino acids 1 to 384) used in these studies was previously described (59) and was a gift from Wolfgang Hammerschmidt. The single LMP2A-deleted B95.8 strain mutant used in most of these studies (a gift from Wolfgang Hammerschmidt) was created as previously described (18) and deletes the first exon of LMP2A (nucleotides [nt] 166170 to 166938 of the genome of the prototype EBV strain B95.8). The M81 strain LMP2A mutant was created by first deleting a region beginning 61 nucleotides before the LMP2A open reading frame and ending 2 nucleotides before the end of LMP2A exon 1 (in B95.8, deletion from nt 166042 to nt 166457; in M81, deletion from nt 165906 to nt 166321). The deletion was achieved by homologous recombination of the recombinant virus with a linear DNA fragment that harbors the kanamycin resistance gene and Flp recombination sites flanked by short DNA regions homologous to LMP2A-specific sequences. The oligonucleotides used were 5′-AGCATCACAGGTTATTTTGCCTGAAGCTTGCTGGGGCGTAAACAGCTATGACCATGATTACGCC-3′ and 5′-TGATGCAATAAATAAAAGTACAGATAGATGGCACTCTTACCCAGTCACGACGTTGTAAAACGAC-3′. The kanamycin cassette was subsequently excised with the Flp recombinase. The LMP1/LMP2A double-knockout virus used in these studies was created from the second LMP2A deletion described above (in B95.8) by introducing deletions of the first two LMP1 exons and of most of the third LMP1 exon. The deletion reaches from nt 167745 to nt 169148 and was obtained by linear targeting with a PCR product that harbors the tetracycline resistance gene and sequences specific for the LMP1 gene. The oligonucleotides used are 5′-GTCATAGTAGCTTAGCTGAACTGGGCCGTGGGGGTCGTCAGCGTGTTTAGATTGGAGTGAACG-3′ and 5′-TACATAAGCCTCTCACACTGCTCTGCCCCCTTCTTTCCTCGCGGAATAACATCATTTGGTGACG-3′.
Infectious viral particles were produced from 293 cell lines stably infected with the WT or mutant viruses following transfection with EBV BZLF1 and Gp110 expression vectors as previously described (47). The titer of EBV was determined on Raji cells by using the Green Raji cell assay as previously described (47).
Immunodeficient NSG (NOD/LtSz-scid/IL2Rγnull) mice were purchased from Jackson Laboratory (catalogue number 005557). Commercially purchased CD34-depleted human cord blood mononuclear cells (CB117; AllCells, LLC) were mock infected or infected with WT or mutant viruses in vitro for 1.5 h, and 12 million to 25 million cells were then injected i.p. into 3- to 5-week-old NSG mice. Experiments using B95.8-derived bacmids were done by using 10,000 infectious viral particles (titers were determined by the Green Raji cell assay); experiments using bacmids derived from the M81 strain virus were done by using 2,000 infectious units. Mice were sacrificed at day 70 postinfection unless they developed symptoms requiring euthanasia prior to day 70. Both WT and mutant virus-infected animals were included in many of the cord blood experiments so that results could be compared by using the same sets of cord blood. In some experiments, animals were treated with the T cell-depleting OKT3 antibody (50 μg of antibody three times per week i.p.) starting 4 days after EBV infection and continuing for the remainder of the experiment.
Following euthanasia, multiple different organs (including the lungs, spleen, pancreas, liver, gallbladder, mesenteric fat, and abdominal lymph nodes) were formalin fixed and then examined by using a variety of techniques to determine if animals had persistent EBV infection and/or EBV-positive lymphomas and to assess the viral protein expression pattern. Samples from all EBV-infected animals (infected with either the WT virus, the LMP2A-KO virus, or the LMP1/2A-KO virus) were examined by H&E staining to determine if tumors were present and to assess the types of tumors in each animal. Tumors from at least 7 different animals infected with the WT virus, the LMP2A-KO virus, or the LMP1/2A-KO virus (B95.8 strain) also underwent IHC staining by using the antibodies listed in Table 2, as previously described (47, 48). IHC staining was also performed on tumors from at least 4 different animals infected with the WT or LMP2A-KO virus in the context of the M81 strain of EBV. Coauthor Erik A. Ranheim, a board-certified hematopathologist, performed the pathological analysis of the tumors described in Table 1. In some animals, EBER in situ hybridization studies were performed by using the PNA ISH detection kit (DakoCytomation) as previously described (47). For quantification of IHC results, at least 2 random fields of view were selected per animal, photographed, and then counted by 2 independent observers. The counts of positively staining cells were averaged across observers. For each ratio (i.e., CD3 to CD20), stains were done on adjacent slides, and counts were completed on the same field of view.
To confirm that tumors derived from EBV mutants contained the expected deletions, in some animals, DNA was isolated from paraffin-fixed slides by using the QIAamp DNA FFPE (formalin-fixed paraffin-embedded) tissue kit and then PCR amplified by using primers specific for the EBV LMP2A gene (left, CCCTAGAAATGGTGCCAATG; right, ATGAGTCATCCCGTGGAGAG [315 bp]), the LMP1 gene (left, AGTCATCGTGGTGGTGTTCA; right, TTACCACACCCCCACTTTTC [291 bp]), or the BZLF1 gene promoter (left, ACCAGCCTCCTCTGTGATGT; right, TTTGGACGAACTGACCACAA [298 bp]) to confirm that LMP2A-KO- or LMP1/2A-KO-induced tumors were not contaminated by WT EBV. PCR was performed in a 50-μl reaction mixture volume containing 0.2 μmol/liter primers and 1 U Taq DNA polymerase, under the following conditions: 95°C for 2 min; 30 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 40 s; and 72°C for 5 min. PCR products were visualized with ethidium bromide on a 1% agarose gel.
Mstat software (http://mcardle.wisc.edu/mstat/download/index.html) was used to statistically analyze tumor incidence and B cell engraftment data at study completion. Two-tailed Fisher's exact test was used to compare tumor formation and B cell engraftment between different viruses. Mann-Whitney tests were performed on cell ratios for different viruses by using GraphPad Prism version 6.00 for Windows (GraphPad Software, La Jolla, CA, USA). A Kaplan-Meier curve was employed (GraphPad Prism) to determine differences in rates of tumor development postinjection by using a log rank test. Animals that did not develop tumors were censored at study completion. Differences were considered significant at a P value of <0.05.
All animal work experiments were approved by the University of Wisconsin—Madison Institutional Animal Care and Use Committee (IACUC) and conducted in accordance with the Guide for the Care and Use of Laboratory Animals (60).
We thank Andrea Bilger and Liz Barlow for help in animal experiments and Eric Johannsen for reviewing the manuscript.
This research was supported by grants P01CA022443 and R01-CA174462 from the National Institutes of Health and University of Wisconsin Cancer Center support grant P30 CA014520.