Double the Trouble When Herpesviruses Join Hands

Abstract

KSHV is the etiologic agent of PEL—an aggressive lymphoma. Interestingly, EBV concurrently exists in nearly 70% of PEL cases. In this issue of Cell Host & Microbe, McHugh et al. (2017) develop humanized mouse models for EBV/KSHV co-infection and identify their complementary effect on in vivo tumor formation.

Virus-induced cancer accounts for 15%–20% of human cancer worldwide. Oncogenic viruses, especially herpesviruses, establish persistent infections that last for long periods without killing the host cells. Kaposi’s sarcoma-associated herpesvirus (KSHV) is the etiologic agent of primary effusion lymphoma (PEL)—a rare, aggressive body cavity-based lymphoma with poor prognosis. Epstein-Barr virus (EBV) was the first discovered human tumor virus, and it frequently causes B cell lymphomas. Interestingly, EBV concurrently exists in nearly 70% of PEL cases, indicating that co-infection plays an active role in the development of PELs (Cesarman et al., 1995). In fact, a previous study indicates that KSHV/EBV co-infected PEL cells are more tumorigenic compared to KSHV-infected PEL cells (Boshoff et al., 1998). To date, investigators have mainly relied on patient-derived PEL cells to address many questions about viral pathogenesis. How KSHV and EBV cooperate to promote tumorigenesis remains unclear, mainly due to the lack of relevant animal models. Since both herpesviruses are primate specific, non-human primates serve as susceptible animal models for EBV and KSHV persistent infections (Moghaddam et al., 1997; Chang et al., 2009). However, they rarely develop tumors in vivo. Since primate models have restricted availability and high costs, small animal models were initially developed by using NOD-scid immunodeficient mice transplanted with KSHV-infected cells (Wu et al., 2006). Later, investigators developed humanized mouse models by engrafting human immune cells into NOD-scid IL2Rgamma^null (NSG) immunodeficient mice and then injecting virus to better mimic the natural settings of viral transmission. These studies have shown that EBV infection of humanized NSG (HuNSG) mice establishes persistent infection, resulting in the induction of B cell- and T cell-mediated immune responses and B cell proliferative disorder (Ma et al., 2011). In addition, KSHV has been shown to establish a robust infection in humanized mouse models (Wang et al., 2014), yet the in vivo mechanism of lymphomagenesis has not been explored.

Co-infection is of particular human health importance because pathogens can interact within the host and have synergistic effects on transmission and disease progression. For instance, human immunodeficiency virus (HIV) and tuberculosis form a lethal combination, each speeding the other’s progression; EBV and malarial co-infection is a major risk factor for endemic Burkitt lymphoma; and KSHV and HIV co-infected patients develop Kaposi’s sarcoma at a higher rate. In this issue of Cell Host & Microbe, McHugh and colleagues introduce a reproducible humanized mouse model of EBV/KSHV co-infection (Figure 1A). In previous studies of humanized mice with KSHV infection alone, only a small population of cells displayed latent and lytic viral gene expression and no tumor formation was detected (Wang et al., 2014). Thus, the co-infection approach of McHugh et al. could provide an appropriate mouse model to study KSHV persistency and tumorigenesis. Indeed, the authors discovered that co-infection increased KSHV persistency and frequency of tumor formation (Figure 1A).

Figure 1. Humanized Mouse Models for KSHV/EBV Co-infection.

(A) Humanized mice (HuNSG) injected with KSHV alone do not form tumors, EBV infection alone induces some tumors, and KSHV and EBV co-infection dramatically increases the number of tumors.

(B) Virus-infected tumor cells are collected from EBV-infected humanized mice or KSHV/EBV co-infected humanized mice. B cells infected with EBV alone show latency type 3 infection. B cells co-infected with KSHV and EBV show plasma cell-like phenotype along with various EBV latency types and lytic gene expression. The dotted circle in the nucleus represents the latent viral episome, and the full viral structure represents virion particle.

The authors demonstrated that in their model both viruses frequently co-infected individual B cells in humanized mice. Notably, these cells showed EBV/KSHV copy loads as well as viral gene transcription profiles similar to human PEL cells. Moreover, the host gene expression profile of infected cells also correlated with that of the primary PEL cells. Specifically, the authors showed that host genes such as IRF4 and BLIMP1 were upregulated in EBV/KSHV co-infected lymphomas in humanized mice, while CD40 and CD19 were downregulated (Figure 1B), corroborating the previous report that patient PEL cells display a plasma cell-like phenotype (Jenner et al., 2003). These results demonstrate that the authors developed an in vivo co-infection system in humanized mice that supports KSHV persistency and lymphoma induction, potentially recapitulating some features of human PELs.

To define the contribution of KSHV to co-infection-induced lymphomagenesis, McHugh et al. (2017) compared host gene expression profiles of EBV+ and KSHV+ lymphoma cells from co-infected mice to those of EBV+ lymphoma cells of singly infected mice. While KSHV co-infection reduced expression of innate immune response and cytokine signaling genes, it induced expression of genes involved in mitochondria function, mitotic cell cycle, and anti-apoptotic process, which are the hallmarks of tumorigenesis. Another interesting result was uncovered when EBV expression profiles were compared between the two infection conditions. It is known that EBV remains mostly latent in infected human B cells and establishes four different forms (0, 1, 2, and 3) of latency depending on associated disease settings. Each latency pattern with different latent gene expression profiles is correlated with B cell differentiation and activation. When EBV-infected B cells undergo germinal center reaction, they express a limited set of latent genes (EBNA-1, LMP-1, and LMP-2), which is called type 2 latency. Infected B cells then differentiate into memory B cells where they express only EBV-encoded RNA, become a long-lived reservoir, and enter type 0 latency. When some of these cells are differentiated into plasma cells by specific stimuli, EBV switches from latency to lytic replication, producing viral progeny for the next round of infection. Most previous studies have been focused primarily on EBV latency for its critical role in viral oncogenesis. Still, lytic EBV-infected cells are frequently detected in EBV-associated lymphomas, and EBV with the BZLF1 lytic gene deletion develops fewer lymphomas, suggesting an oncogenic role for the EBV lytic cycle (Ma et al., 2011). However, the contribution of EBV lytic replication to lymphomagenesis has not been extensively investigated in sophisticated animal models. McHugh et al. (2017) found that tumor cells from KSHV/EBV co-infected humanized mice surprisingly displayed enhanced expression of EBV lytic genes and a mixed population of latency types 1, 2, and 3 (Figure 1B). This is an interesting finding since EBV single infection of humanized mice displays a latency type 3 expression pattern. Consistently, BZLF1 was expressed at a higher level in EBV/KSHV co-infected lymphoma cells than EBV singly infected lymphoma cells. Moreover, KSHV/EBV(BZLF–) co-infection failed to develop tumor formation as frequently as KSHV/EBV co-infection, suggesting that enhanced EBV lytic gene expression upon KSHV co-infection contributes to efficient lymphomagenesis. Based on these findings, the authors conclude that KSHV co-infection has a critical role in PEL development through modulating host and viral gene expression.

Since most PEL patients are co-infected with EBV and KSHV, numerous studies have explored the interactions between these two viruses for their replication and latency in an in vitro setting. The prevailing view of these studies has been that EBV and KSHV effectively inhibit each other’s lytic replication. In contrast to this in vitro effect, McHugh et al. (2017) found that co-infection of humanized mice with EBV and KSHV led to a significant synergic effect on in vivo lymphomagenesis. Is this simply a difference between in vitro and in vivo studies? Does this co-infection model of humanized mice fully recapitulate EBV/KSHV-induced PELs in patients? Future studies should be directed to answer how EBV/KSHV co-infection affects each virus’s life cycle and how they cooperate to induce efficient lymphomagenesis. As the KSHV RTA-induced expression of CD21 glycoprotein, which is an EBV receptor, efficiently facilitates EBV infection (Chang et al., 2005), does this suggest KSHV infection occurs first, followed by EBV infection, or vice versa? While KSHV is the major causative agent for PEL, the KSHV single infection of humanized mice still does not lead to tumor formation, which is a limitation of this model. Nevertheless, we hope that this co-infection model using humanized mice is further advanced and used to provide detailed mechanisms of gammaherpesvirus-induced lymphomagenesis that could lead to the development of PEL treatments such as immunotherapy.