Epstein-Barr virus: biology, pathogenesis and therapy of lymphomas

Abstract

Epstein-Barr virus (EBV) remains dormant in host B cells for a long time, and more than 90% of patients are asymptomatically infected. However, some internal factors (such as immune deficiency) or exogenous factors (such as HIV infection, inflammation and hypoxia) can destroy this underlying property and lead to disease development. EBV was originally discovered in association with Burkitt lymphomas and is now etiologically associated with malignant lymphomas of B, T, and NK cell origins. Previous relevant studies have reported that EBV could promote the occurrence and development of lymphomas and affect the prognosis. In fact, the pathogenesis of EBV-associated lymphomas involves different viral gene expression patterns and complex cytogenetic changes. Here, the present study summarizes viral two life cycles, latent modes, the establishment of latency and viral reactivation. We also focus on the occurrence of EBV-related lymphomas, generalize and put forward novel perspectives or innovative solutions for treatments according to these pathogenic mechanisms.

Introduction

Epstein-Barr virus (EBV), also known as Human herpesvirus 4, is a linear double-stranded DNA virus with a genome of 172 kb, which encodes about 100 genes. EBV is transmitted by saliva, keeps dormant in the lymphocytes of hosts for a long time and presents with asymptomatic infection among more than 90% patients [1–3]. However, some internal (e.g. immune deficiencies) or exogenous factors (e.g. HIV infection or DNA damage) destroy this latent nature and cause the disease to develop. Therefore, parts of population with EBV infection are prone to develop tumors under certain conditions [4].

EBV was found to be related with Burkitt lymphomas (BLs) firstly as well as strongly [5, 6] and was identified in other various lymphomas subsequently, such as Hodgkin lymphomas (HLs) [7], diffuse large B-cell lymphomas (DLBCLs) [8], NK/T-cell lymphomas (NKTCLs) [9] and peripheral T-cell lymphomas (PTCLs) [10]. The incidence of EBV infection among lymphomas varies by areas, ages and types of disease. More details are shown at Table 1 and will talk at the part of “EBV-associated lymphomas”.

Table 1.

EBV-associated lymphomas

EBV-associated Lymphomas		EBV association	Latency type	Viral gene expression	Molecular changes about EBV
HLs	developed countries	30–50%	II	EBER, EBNA1, LMP1, LMP2, miRNAs-BARTs	Genetic variants within Human Leukocyte Antigen (HLA)-A*01 and HLA-A*02 risk alleles
	developing countries	> 90%
BLs	Endemic	> 95%	I	EBER, EBNA1, miRNAs-BARTs	Mutations in FOXO1 and BCR
	Sporadic	10–20%
DLBCLs		Varies by countries	II	EBER, EBNA1 LMP1, LMP2, miRNAs-BARTs	Mutations in MYC, RHOA, PIM1, MEF2B, MYD88 and CD79B; activation of the JAK/STAT, NOTCH and NF-kB pathways; amplification of 9p24.1
			III	EBER, EBNA1, 2, 3 A, 3B, 3 C, LP, LMP1, LMP2, miRNAs-BARTs, miRNAs-BHRF1

EBV Epstein-Barr virus, HLs Hodgkin lymphomas, BLs Burkitt lymphomas, DLBCLs diffuse large B-cell lymphomas, NKTCLs NK/T-cell lymphomas, PTCLs peripheral T-cell lymphomas; HIV human immunodeficiency virus, EBER EBV-encoded small RNA, EBNA Epstein-Barr nuclear antigen, LMP latent membrane protein, BART BamH1 A rightward transcripts, BHRF BamH1 H rightward open reading frame, NA non-applicable

Here, we focus on EBV biphasic infection mode, selective expression of the latent gene, the reactivation of the virus, discuss how to treat patients with EBV-positive lymphomas from tumor development and viral reactivation pathways, summarize therapeutic interventions and put forward innovative perspectives.

Biology of EBV

Two life cycle

EBV has a biphasic mode (latent and lytic cycle). Generally, EBV contributes to lytic infection in epithelial cells without the establishment of viral latency. However, both latent and lytic viral genes can be expressed during the primary infection for B cells.

During latent infection, the virus cannot copy itself but relies on the proliferation of the host cells. The virus utilizes EBNA1 [11] to directly binds with origin of plasmid replication (OriP) for the replication in host S-phase, connects with genome chromosomes in host and retains as a nuclear episome to passively achieve long-term existence [12]. There are several latent infection genes, including 6 nuclear antigens (Epstein-Barr virus nuclear antigens [EBNA] 1, 2, 3 A, 3B, 3 C and leader proteins [LP]), 3 latent membrane proteins (LMP1, 2 A, 2B) and also express some non-coding RNAs, such as EBV-encoded small RNA (EBER)-1, EBER-2, miRNAs-BamH1 A rightward transcripts (BARTs) and miRNAs-BamH1 H rightward open reading frame (BHRF1) [13, 14]. The EBERs are expressed at all latent time so applied as a marker for EBV infection. The only expression of EBER, miRNAs-BARTs, and EBNA1 is known as latency 0. Latency I is present with the restricted expression of EBNA1 and is relate with BLs [15]. Latency II is the relative elaborate latency programme (EBER1, EBER2, EBNA1, LMP1, LMP2A) and is associated with HL, NKTCL and PTCL [16–18]. Latency III has extensive viral antigen expression such as EBER1, EBER2, EBNA1–6, LMP1 and LPM2, and is more common in immunocompromised patients [2, 19].

During lytic infection, viral DNA undergoes three temporal phases: immediate early (IE) gene expression, early (E) gene expression, and late (L) gene expression [20]. The first two stages with expression of IE and E genes together are also called as prelatency or abortive-lytic phase [21]. Viral DNA is replicated independently. Then virally encoded structural proteins such as viral capsid antigen (VCA), major capsid protein (MCP) as well as gp350 are translated.

We observe that latent infections are essential for parallel transmission as well as long-term persistence related with immune escape. These latent patterns depending on the expression of proteins vary according to cell types, infected time, and cellular environment. We infer that treatments for the same latent pattern in different diseases may have connections; the same disease having different latent patterns like DLBCLs may have different responses to the same treatment. Lytic infection is linked with viral replication and reactivation so inhibiting lytic cycle may improve survival outcomes for infected patients.

Latent patterns

Taking EBV infection of primary B cells as an example to illustrate what latent proteins are selectively expressed (Fig. 1). The expression of latent genes is regulated by methylations [22] on viral gene promoters, such as Wp, Cp, Qp and LMPp. The initial infectious response initiates the expression of partial IE and E genes and commences a temporary lytic cycle without producing viral particles. This special episode with sporadic replication is called as prelatency or abortive-lytic phase [23]. Several scientists thought it probably was due to by leaky expression of many uncontrolled viral gene [2, 24].

Fig. 1.

The process of EBV invasion

①EBV can invade epidermal cells and B cells. ②When EBV enters B cells through cell-to-cell interaction, sporadic replication will occur. ③Subsequently, a large number of viral antigens will be replicated and expressed, causing the body’s immune response. ④Virus escapes immune defenses, travels to the germinal center, and begins to differentiate. ⑤After further differentiation, centroblasts transform into centrocytes. Centrocytes further differentiate and leave the germinal center toward the capsule. ⑥When infected cells released from germinal centers, the expression of relevant viral antigens is suppressed on memory B cells (proliferating status presents as latency I; resting status presents as latency 0). The virus can sleep in memory B cells to achieve virus latency. ⑦When B cells successfully differentiate into plasma cells, the virus begins to replicate and cleave again.

The stage of latent phase III is followed after transient reproduce. Three latent membrane proteins, six EBV nuclear antigens and viral non-coding RNAs are expressed [25, 26]. Abundant expressions of viral antigens activate the immune function, participating in the elimination of virus [27]. However, partial viruses escape this immune attack, move into germinal centers (GC) successfully and come into differentiation [28]. In GC, viruses enter the latent phase I/II and express latency-associated genes. After release from GC, relative viral antigens are suppressed in EBV-infected memory B cells and only EBER can be seen (latency 0) [29]. Viruses rest in the memory B cells and usually result in asymptomatic performance in patients. However, these quiescent cells could proliferate readily after the expression of EBNA1 by specific signal stimulation. This pattern is defined as latency I [12]. In this latency, memory B cells recruit to GC, and differentiate into plasma cells where start EBV reactivation [30, 31]. Both the differentiation into plasma cells and activation of B-cell receptor (BCR) promote EBV reactivation [32, 33].

We can see that latency III is associated with immunocompromised populations. When defense system is crippled, a large amount of viral protein will be expressed constantly, like HIV-associated patients and elderly DLBCL patients, but the older in other lymphomas show inconsistent profiles. On the other hand, patterns of latency I or II may correlate with the origin of tumors or specific signal stimulation, implying that latent genes drive oncogenesis of EBV-positive malignancies to some extent.

Epigenetic lifestyle

Before the virus infects B cells, host genome is characterized as repressive CpG methylation while virion is present with no histone or no CpG methylation. When primary B cells are invaded by viruses, sporadic replication will occur due to leaky expression during prelatency. Subsequently, a series of modifications, such as methylations of viral DNA, and the trimethylation of histone H3 lysine 27 (H3K27me3) give rise to severe suppression of lytic gene expression to establish latent infection, which influence the expression of host genes in return. The heterochromatinized status by CpG methylation in the host genome is partially replaced although the mechanism is elusive. Genomic changes are depicted at Table 2 [2, 22, 29, 34].

Table 2.

Genomic changes when viruses infect B cells

Status	No EBV infection		Latency
Host genome	Massive methylation	[–]	Repressive histone marks and CpG methylation are disrupted	[+]
Viral genome	No methylation	[+]	Methylation selectively and heavily	[–]

[+] means activation and [–] means suppression.

EBV Epstein-Barr virus

Totally, this inhibited status assists in latent infection and is considered as a type of innate immunity of the host. Expression of latent genes and suppression of lytic genes are manipulated by methylations and so we speculate that epigenetic modifications such as methylation and histone acetylation (H3K27ac) [34] have great significances for EBV-associated lymphomas. Hypomethylating agents and histone deacetylase inhibitors (HDACi) probably have a function in treating EBV-positive diseases.

Mechanism of viral reactivation

IE expression

Lytic EBV genes are dormant due to host-driven extensive methylation but viral replication will be reactivated due to decreased immune function and specific exogenous stimulation [11], such as inflammation, hypoxia and DNA damage. The process of viral reactivation is depicted at Fig. 2. Following reactivation, two promoters, the BZLF1 promoter (Zp) and BRLF1 promoter (Rp), are activated, then two IE genes [35], the BZLF1 and BRLF1, are transcribed at first and encode the transcription factors, Z (aka Z, Zta, ZEBRA) and R (aka R, Rta, RTA) respectively.

Fig. 2.

The process of viral reactivation

Z-mediated reactivation

The silence of the Zp is mainly caused by histone methylation rather than CpG methylation [36, 37]. It is easy for the Zp to activate the transcription of BZLF1 by activated histone modifications. The above conditions lay the foundation for the switch from silencing to activated BZLF1. Intriguingly, Z actually prefers the methylated form [38] so binds with a CpG-methylated BZLF1-responsive element (meZRE) for the transcription of early lytic viral genes. Moreover, Z interacts with the origin of lytic replication (ori-Lyt) where ZRE is located and with BHLF1 promoter which enhances the ori-Lyt’s activity, forming a stable DNA-RNA hybrid. The complicated interaction provokes viral DNA synthesis [39]. Additionally, BZLF1 [40] prevents NF-κB activation to suppress proinflammatory factors, which is an antiviral and anticancer strategy.

R-mediated reactivation

Different from Z, R prefers unmethylated promoters although it is capable to activate both methylated and unmethylated ones. Therefore, the methylation status of the viral genome determines whether Z or R initiates transcription of early cleavage genes firstly. In normal oral keratinocytes, there is much less methylated. In this case, BRLF1 directly binds with BRLF1-responsive element (RRE) and induces transcription of relative viral genome [41]. R fails to identify oriLyt, but combines with oriLyt’s enhancer region and regulates the expression of replication proteins by BHLF1 then participates in viral DNA replication [42]. The knockout of either genes can result in the block of lytic cycle [43]. Although Z protein is sufficient to induce reactivation, Z and R proteins activate promoters individually or coordinately [4, 44], trigger waterfall expression of lytic genes and make virus quickly enter the proliferation phase.

E gene expression

Subsequently, E lytic genes such as BMRF1, BALF5, BHLF1, BHRF1 and BGLF4 indispensable for the viral DNA replication are expressed. BMRF1 function as an accessory factor of the virally-encoded DNA polymerase (BALF5) to consolidate the binding to DNA [45], as a regulator of BALF2 to act at replication forks [46] and as an activator of BHLF1 to enhance promoter activity [47]. As the DNA polymerase, BALF5 use ori-Lyt for viral DNA replication [48]. BHLF1 plays a function to promote persistent cell growth during latent infection [49] and is also active during lytic infection as the role of oriLyt’s promoter. A stable RNA-DNA hybrid from oriLyt and BHLF1 transcript is necessary for the commence of DNA replication. BHLF1 is controlled by R to regulate the expression of replication proteins and to participate in viral DNA replication [42].

Initiation of transformation also requires the expression of two BCL-2 homologue proteins named as BHRF1 and BALF1. BHRF1 not only combines with BZLF1 to improve oriLyt’s activity as a promoter of oriLyt but also exerts anti-apoptotic effects similar to BCL-2. BHRF1 [50] alters mitochondrial dynamics and subsequently inhibits type I interferon activation, an important effector cytokine against viral infection, contributing to innate immune evasion. Additionally, the expression of BHRF1 stimulates the development of MYC-induced tumors [51]. BHRF1 is also expressed at latency but the function is still elusive. BALF1 is another BCL-2 homologue protein and can promote tumor metastasis and inhibit autophagy [52, 53]. Conversely, BALF1 [54] was reported to function as a negative regulator of BHRF1. The role of BALF1 is still equivocal and needs further exploration.

IE and E staged cleavage cycle of EBV effectively separates the process of viral DNA replication (genome proliferation) from the release (propagation) of viral particles. The IE and E genes are only responsible for genome replication of the virus, but do not produce structural proteins. This means that the virus can undergo genome expansion in the host while not producing infectious virions, reducing the risk of being recognized and cleared by the immune system. Meanwhile, the phase of DNA synthesis that IE and E genes are responsible for can be seen as a critical node in the viral life cycle. Effective intervention at this stage, such as inhibition of viral DNA replication, can prevent the virus from entering a more active phase of transmission and reduce the risk of disease transmission.

L gene expression

After de novo replication of the viral genome, methylation of viral DNA is eliminated [2]. Therefore, the newly replicated DNA released from nucleus capsid is unmethylated so Z because of its preference, fails to bind with CpG islands to stimulate relative promoters and produce progeny virus. On the contrary, unmethylated DNA is beneficial to the viral preinitiation complex (vPIC) for recruiting host RNA polymerase II and activating L genes expression [55, 56]. The L viral genes encode structural proteins (such as VCA, MCP and gp350) associated with virus assembly. In short, only after DNA synthesis are L genes expression, virus particles released.

The epigenetic modification of EBV is a dynamic process with different modes at different time. Initiation of IE expression is manipulated by methylations, which proves potential efficacy of hypomethylating drugs once again; the elimination of methylation during L gene expression ensure viral launch for its next epigenetic life cycle in a newly infected cell. Lytic gene products participate in latent cycle and have regulatory effects on the host’s system. Several effects in the tumor microenvironment (TME) are mainly negative factors that make the generation and maintenance of meaningful immune responses quite difficult. Thus, constant immune attack is a key to treating EBV-associated lymphomas. Considering strict chronological order during lytic infection, blocking any steps of viral reactivation by antiviral drugs may decrease the viral load of EBV and improve prognosis for infected patients.

Pathogenic mechanism of viral gene products

EBNAs

EBNA1

EBNA1 expressed at all latency is beneficial to not only long-term tumor latency but also B-cell immortalization and neoplastic growth [57, 58]. The possible mechanism is that [1] EBNA1 [59] inhibits the canonical NF-kB pathway to evade immune system [2]. EBNA1 [60] is also linked with oncogenesis. It binds with Vav1, then suppresses the expression of bim which is a member of pro-apoptotic Bcl-2 family and promotes growth of lymphoma [3]. EBNA1 [61] can down regulate the immune system related to NK cells, escape immune surveillance and reduce apoptosis [4]. Viruses’ products regulate the expression of miRNAs to affect molecular level. For example, EBNA1 [62] up regulates the expression of hsa-miR-127, causing changes in the modulation of genes in memory B cells.

EBNA2

EBNA2 serves as a transcription factor which is anchored in relative DNA via sequence-specific binding factor. EBNA2 can not only master viral genes, such as EBNAs (including EBNA2 itself) and LMPs but also activate cellular genes [63, 64], such as CD23 to induce cellular growth transformation and MYC to initiate the viral immortalization. In addition, it increases PD-L1 expression to escape immune surveillance [65]. Moreover, EBNA2 [66] induces the expression of CCL3 and CCL4 which in turn activate Bruton’s tyrosine kinase (BTK) and NF-κB signaling pathways, and surrenders B-cell lymphomas resistant to doxorubicin. The dysregulation of microRNA influences genetic expression and is linked with tumor development. EBNA2 [67] utilizes miR24 to upregulate the expression of c-MYC and to downregulate the inducible co-stimulator ligand (ICOSL) causing immune escape. EBNA2 [68] also alter expression of miR-21 and miR-146 to induce B-cell transformation.

EBNA-LP, EBNA3 A, B, and C

EBNA-LP (also known as EBNA-5) is produced with EBNA2 and cooperates with EBNA2 to induce LMP1, which possibly influence cellular proliferation [69]. Of note, EBNA-LP [70] was found to be involved in alternative splicing to modulate cellular proliferation and apoptosis recently.

EBNA3 A, B, and C [71] play an important role in cell proliferation and vitality, which are beneficial for the development of lymphoma. EBNA 3 A [72] contributes to cell growth by restraining cyclin-dependent kinase inhibitor (p21, also call as p21WAF1/CIP1p21 or CDKN1A). A previous study [73] showed that EBNA 3B was not required for B cell transformation in vitro, but EBV-infected mice lacking EBNA 3B could develop into DLBCL-like tumors. EBNA3B might play a role of tumor suppressor. EBNA 3 C fights against apoptosis by regulating aurora kinase B and E2F1, leading to proliferation and persistence of viruses [74, 75]. Moreover, EBNA 3 C cooperates with EBNA 3 A to antagonize bim-mediated apoptosis [71].

Overall, latent gene products play a role in viral long-term persistence, neoplastic cell proliferation, cellular apoptosis and chemotherapy resistance. In this respect, latent phenotype can be potential therapeutic targets for lymphomas. Moreover, the restricted expression of EBNA1 is strongly associated with BLs so targeting on EBNA1 may have clinical benefits for BLs. We hypothesize that presentation of EBNA1 and deregulation of MYC might exert cooperating genetic alterations for lymphomagenesis.

LMPs

LMP1

In physiological conditions, B cells are activated when they contact a CD40L-bearing T cells. However, LMP1 mimics an activated tumor necrosis factor receptor such as CD40, congregates relevant factors, and produces several signal cascade reaction including the NF-κB [76, 77] JAK/STAT3 and PI3K/AKT [78] pathways, leading to sustain activation of EBV-infected B cells. LMP1 is related with treatment resistance. It stimulates the level of mitochondrial dynamin-related protein 1 (Drp1) through AMPK and cyclin B1/Cdk1 pathway to cisplatin resistance [79]. Moreover, LMP1 promotes BNIP3-induced autophagy to achieve the radioresistance [80]. LMP1 influences the immune system and survival outcomes as well. It can induce the expression of CD137 in Hodgkin and Reed-Sternberg (HRS) cell lines via PI3K-AKT-mTOR pathway [81] that contribute to immune invasion. It mediates the level of PD-L1 in cytotoxic T cells which suppress immune system by NF-κB pathway and then influences the prognosis of lymphomas [82]. It simultaneously mediates the release of cytokines to influence prognosis [83]. Just like other viral products, LMP1 regulates various miRNAs, such as miR-146a [84], to modulate the intensity and/or duration of the interferon response.

LMP2

In contrast, the function about LMP2 has not been elucidated clearly. The N-terminal domain of LMP2A is similar to the domain of BCR with an immunoreceptor tyrosine-based activation motif. It is taken for granted that LMP2 plays an alternative role in BCR-deficient cells, like HRS cells to suppress viral reproduction and promote tumor cell growth [85, 86]. Considering the opinion that activated BCR signaling commonly induces the virus lytic cycle [2, 33, 87], the BCR-like role of LMP2 has been questioned. Several studies [85, 88] explained that LMP2A was like a decoy receptor that recruited tyrosine kinase proteins, hijacked BCR signaling and blocked viral replication. Vockerodt et al. [89] reported BZLF1 expression was not induced until Egr-1 gene indispensable for viral replication could be upregulated by LMP2 and BCR, but the expressions of Egr-1 and BZLF1 were absent in HRS cells. Recently, proteomics and genetic level between LMP2 and BCR were compared [88], and scientists found that most transcription factors were LMP2-specific, with only a few overlaps. By the way, LMP2 and LMP1 have a cross-linking reaction and cooperate to promote EBV lymphoma [90, 91]. The lack of LMP2A delayed tumorigenesis while loss of LMP1 plus LMP2A resulted in fewer lymphomas after EBV infection and even later onset of tumors [92].

In conclusion, LMP1 participates in various signaling pathways and immune responses, affecting tumor occurrence and survival outcomes. LMP2 competitively inhibit normal BCR signaling in tumor cells. This suppression may be strengthened in HLs owning to the loss of BCR but other destabilizing signal remodeling derived from crippled GC B cells dominate in order to escape the physiological apoptosis program. Based on previous reports, we infer that LMP2A works to accelerate neoplastic development and interacts with LMP1 synergistically to induce tumor formation. However, there is no relevant experiments to confirm the above speculation.

Viral non-coding RNAs

There are numerous viral non-coding RNAs, such as EBER1, EBER2, miRNAs-BHRF1 and miRNAs-BARTs. The function of miRNAs is complex and they influence viral replication [32, 33], cellular proliferation and apoptosis [93, 94]. They also play a role in carcinogenesis [94].

A brief glimpse of latent proteins at Table 3 provides a good summary of the pathogenic mechanisms; gene position structure of oriLyt is shown in Fig. 3. The latent status promotes the development of EBV-associated malignant tumors, indicating that those latent genes or proteins are potential targets for relative lymphomas. Hence, understanding underlying molecular mechanisms is a pressing need for developing novel effective treatment strategies and ultimately achieving the optimal prognosis.

Table 3.

Pathogenic mechanism of viral gene products

Viral gene products		Pathogenic mechanism
EBV Nuclear Antigen	EBNA1	Viral proliferation during latency
		Inhibit canonical NF-κB pathway
		Inhibit autophagy (↓ bim)
		Regulate the expression of miRNAs (↑ hsa-miR-127)
		Inhibit immune system related to NK cells
	EBNA2	Activate viral genes (EBNAs and LMPs)
		Activate cellular genes (↑ CD23; ↑ MYC; ↑ CCL3 and CCL4 which influence BTK and NF-κB; ↑ PD-L1 expression)
		Utilize miRNAs (↑ c-MYC and ↓ ICOSL by miR24; induce B-cell transformation by miR-21 and miR-146)
	EBNA-LP	↑ LMP1 by cooperating with EBNA2
		Modulation of alternative splicing
	EBNA3A	Inhibit autophagy (↓ bim; ↓ p21)
	EBNA3B	Function as a tumor suppressor
	EBNA3C	Inhibit autophagy (↓ bim; ↓ aurora kinase B; ↓ p53; ↓ E2F1)
Latent Membrane Protein	LMP1	Mimic CD40 to activate PI3K/Akt, NF-kB, Stat3 pathways and many other cellular signaling pathways
		Treatment resistance
		Influence the immune surveillance and survival outcomes
		Regulate the expression of miRNAs
	LMP2A	Mimic and hijack BCR signaling
		Cross-linking reaction with LMP1 to promote EBV lymphoma
	LMP2B	Modulate the effects of LMP2A on BCR function
Viral non-coding RNA	EBER	A marker of infection
	miRNAs-BHRF1	Inhibit autophagy (↓ p53; ↓ PRDM1/blimp1)
		Influence the lytic cycle
	miRNAs-BART	Target various gene (↓ DKK1, ↓ DAB2, ↓ APC and ↓TP53)

↑ means activation and ↓ means suppression.

EBNA Epstein-Barr nuclear antigen, LMP latent membrane protein, EBER EBV-encoded small RNA, BART BamH1 A rightward transcripts, BHRF BamH1 H rightward open reading frame, DDK1 dickkopf WNT signaling pathway inhibitor 1, APC adenomatous polyposis coli, DAB2 disabled homolog 2

Fig. 3.

Genomic map of EBV region about oriLyt

The core domain of the origin of lytic replication (ori-Lyt) includes the promoters for and intergenic region of BHLFl and BHRFl genes. Two essential components (upstream and downstream components) of oriLyt bind BZLFl which interacts with BHLF1 promoter to enhances the ori-Lyt’s activity at BZLFl-responsive elements(ZRE). Once entering the lytic cycle, miR-BHRF1-2 and-1-3 are induced rapidly, which is consistent with the beginning of BHRF1 transcripts, while the expression of miR-BHRF1-1 is delayed until the Cp/Wp starts to transcription.

EBV-associated lymphomas

HLs

Based on the fact that infectious mononucleosis (IM) was a risk factor for HL, EBV was found to be linked with HL [31]. Since nodular lymphocyte predominant HL is rarely EBV-positive, we put spotlight on classical HL (cHL) which is characterized by HRS cells in the TME [95]. In developed countries, 30–50% HL has an association with EBV infection [96, 97]. By contrast, the proportion in developing countries is extremely higher [96], indicating that EBV may have a relationship with some environmental factors. EBV tends to infect children under 10 years and adults over 80 years. Given the association between IM and EBV-positive cHL, the occurrence of cHL in children may be attributed to primary EBV infection, while the prevalence in the elderly may be associated with a deceased immunity and an increased EBV load [95]. HRS cells derive from pre-apoptotic GC cells which experience somatic hypermutation and lose their B cell phenotype (such as BCR) [98, 99]. Some genetic variants within Human Leukocyte Antigen (HLA)-A*01 and HLA-A*02 risk alleles [100, 101] are associated with EBV-positive cHL.

Due to the lost B-cell phenotype, HLs innovate other unstable signaling pathways to avoid apoptosis, resulting in sensitivity to radiotherapy and chemotherapy. As a consequence, patients with HLs usually respond well to relative therapies.

BLs

There are three subtypes of BL sharing morphological and immunophenotypic characteristics: endemic BL (eBL), sporadic BL (sBL), and immunodeficiency-related BL. Of note, eBL is frequent in Africa while sBL can be seen in developed countries [102]. About 95% eBL, 10 to 20% sBL, and 40 to 50% HIV-infected individuals are related with EBV infection [103]. BLs exhibit the profile of latency I and is featured as reciprocal chromosomal translocation of chromosome 8 carrying the MYC ( a key factor of germinal centroblast proliferation) oncogene and chromosomes 2, 14, or 22 having the immunoglobulin locus [102]. Through genetic sequencing, Thomas et al. [104] found FOXO1 and BCR genes were mutated frequently and significantly in EBV-positive BLs. Roschewski et al. [102] concluded that EBV-positive BLs more frequently activate p53 and inactivate cyclin D3.

Although clinical management for BLs is based on risk stratification and invasion of central nervous system, it is worthy to note that we should target EBNA1-associated molecular alterations to carry out individualized treatment, like cytotoxic T lymphocytes (CTLs) and chimeric antigen receptor T (CAR-T). Moreover, VK1727, an EBNA1 inhibitor [105], can suppress EBV-positive gastric cancer cell lines. Future research on its impact on EBV-associated lymphoma is anticipated.

DLBCLs

DLBCLs are divided into two categories by cell of origin: germinal center B cells (GCB) and activated B cells (ABC). EBV-positive DLBCLs are commonly related with ABC subtypes [106, 107]. About 5% DLBCL are tested as EBV positive in Western countries and 10%-15% in Asia, Eastern Europe, and South America [8, 108]. EBV-positive DLBCLs tend to arise in elderly patients exhibiting a profile of latency III, which are suspected to be caused by weakened EBV-specific T cell immune ability while a few happen at younger age with latency II pattern [95]. EBV-positive DLBCLs show high-frequency mutations in MYC, RHOA, PIM1, MEF2B, MYD88 and CD79B [109]. The activations of the JAK/STAT, NOTCH and NF-kB pathways are more common in EBV-positive DLBCLs than the counterpart [106, 107]. Amplification of 9p24.1 which gives rise to PD-L1 over-expression is frequent as well [107].

Considering the unique genetic landscape of DLBCLs with EBV infection, the PD-1/PD-L1 inhibitors [8, 110] and NF-kB pathways inhibitors might be used to treat EBV-related DLBCLs.

However, different age groups in DLBCLs have different latent profiles so younger and older patients should be analyzed separately.

NKTCLs

NKTCLs are divided into nodal or extra-nodal NK/T-cell lymphoma (ENKTL). ENKTLs occur frequently in Asia and Latin America, and are strongly associated with EBV. NKTCLs exhibit the pattern of Latency II. The 6q21 deletion [17, 95] which causes the silence of several suppressor genes is commonly tested. Promoter hypermethylation also have similar role. TP53 and PD-L1 are also detected to promote the development of NKTCLs [17]. The signal pathways [95] such as NF-kB, JAK-STAT and NOTCH are influenced as well. Xiong et al. [111] classified NKTCL into three molecular subtypes: TSIM, MB, and HEA. The TSIM subtype is characterized as mutations in JAK-STAT pathway, TP53 and so on, parts of which drive PD-L1/2 expression. The MB subtype is enriched for MYC-associated aberrations, such as MGA mutation and LOH at the BRDT locus. The HEA subtype (HDAC9, EP300, and ARID1A mutation) shows aberrant histone acetylation.

Taking this molecular classification into consideration, specific targeted treatments are recommended. PD-1/PD-L1 inhibitors are a good choice for TSIM subtype; homoharringtonine which regulates the expression of MYC may have an inhibitory effect on the MB subtype; aberrant histone acetylation in the HEA subtype may enhance the therapeutic activity of HDACi.

PTCLs

There are many subgroups of PTCLs. About 25–40% PTCLs have association with EBV infection [112, 113]. The EBV gene expression pattern in PTCLs is mainly latent phase II; however, a few shows latency type III with lytic activation [114]. Few studies explore the association between EBV and PTCLs at the molecular level. Only Wai et al. [18] found JAK-STAT and NF-κB related with PD-L1 were upregulated in EBV-positive PTCLs. Our previous founding [115] demonstrated that EBV-related PTCLs were more likely to have DNMT3A (P = 0.002) and TET2 mutations (P = 0.032). These mutations [116, 117] were associated with DNA methylation in hematologic tumors. It is worthy to note that TET2 overexpression is commonly observed in AITL irrespective of EBV status [118]. Therefore, the relation between EBV and PTCL needs further exploration.

In light of highly heterogeneity of PTCLs and interaction between immune microenvironments, we think EBV just exerts a partial influence on disease progression.

EBV-related therapies for lymphomas

Immune checkpoint inhibitors (ICIs)

The level of PD-1/PD-L1 expression is thought to be related with immune evasion. The content we discussed above demonstrated that EBV was related with overexpression of PD-L1 [17, 18, 65, 119]. Therefore, ICIs like pembrolizumab, tislelizumab or nivolumab probably have promising clinical effects for EBV-associated lymphomas. Alterations of 9p24 [120] related with PD-L1 overexpression are expected to be a predictive factor to PD1/PD-L1 therapy. Relevant clinical trials have been in full swing. For relapsed or refractory (R/R) NKTCLs, pembrolizumab or nivolumab were highly effective as we expected [121, 122]. For R/R HLs, pembrolizumab had acceptable toxicity and high response rates with 69.0% overall response rates (ORR) and 22.4% complete remission (CR) rates, respectively [123]. Chemotherapies involved in pembrolizumab [124]^, [125] were suitable for untreated HL as well. ICIs achieve considerable remission rate for EBV-positive DLBCLs [8], resonating well with our inference. In a retrospective study of 6 refractory EBV + DLBCL patients [110], the combination of PD-1 inhibitors and chemotherapy showed promising results, with an ORR of 83% and a CRR of 67%. There is no prospective clinical trials about ICIs in patients with EBV-positive PTCL. We think that PD1/PD-L1 blockade is a good option for latency II/III-related lymphomas and alterations of 9p24 can test the effectiveness of treatments, but clinical trials will be needed to evidence this viewpoint.

BTK inhibitors

The BTK inhibitors can inhibit NF-κB signaling which is mainly active [76, 77, 106, 107] in EBV-related hematological malignancies. Additionally, LMP2 and EBNA2 influence viral reproduction and treatment response by the activation of BTK [8, 66, 126]. In this scenario, EBV-positive lymphoma can be treated with BTK inhibitors theoretically. Ibrutinib and nivolumab cooperated together to treat R/R HLs more than half of whom had progressed on prior treatment of nivolumab; the clinic program presented 29.4% CR rates [127]. Kim et al. [66] demonstrated that BTK inhibitors had synergy on adriamycin for DLBCLs with EBNA2 positive and the assessment of CCL3 and CCL4 levels could be used to select potential patients who would benefit from BTK inhibitors. ABC DLBCLs are more likely EBV infection, then the combination therapy of ibrutinib and rituximab plus lenalidomide for this entity was held and showed clinical benefits with the ORR of 65% [128]. Sadly, this study [128] contains little to no discussion of EBV status so the above combination therapy cannot be directly applied to EBV-positive DLBCL. However, a phase II study showed ibrutinib plus rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) failed to improve response rate and survival rate in comparison with R-CHOP among EBV positive DLBCLs. Interestingly, patients younger than 65 years had a better ORR (87.5% vs. 25.0%; p = 0.01) than older ones [129]. This phenomenon has an echo in our inference that different latent patterns for young and elderly patients in DLBCL present different responses for similar treatment.

Tyrosine kinase inhibitors (TKIs)

Dasatinib, a kind of TKI, targets at Philadelphia chromosome and is approved in chronic myeloid leukemia. Tyrosine kinases are necessary for the bypass of the normal BCR signal and propagation of the LMP2A-associated pathways according to previous studies [88, 130]. In this regard, Dargart et al. [131] constructed EBV-associated murine model expressing MYC and LMP2A who was treated with dasatinib and found dasatinib could effectively inhibit tumor growth. Despite the lack of related clinical trials, our team hypothesize that dasatinib may have better effect on HLs because of active the bypass of the normal BCR signal. More clinical trials are needed to keep up with the new knowledge about the association between dasatinib and EBV.

BCL-2 inhibitors

The BCL-2 inhibitors like venetoclax have been used for patients with chronic lymphocytic leukemia, acute myeloid leukemia and mantle-cell lymphoma clinically.

Considering the role of BCL-2 homologue proteins (BHRF1 and BALF1) [50–53], BCL-2 inhibitors may have a good therapeutic effect on EBV-positive lymphomas. Sejic et al. [132] used a kind of BCL-XL inhibitors named A-1,331,852 to treat EBV-positive NKTCLs and found it continuously induced cell apoptosis in vivo. Robert et al. [133] found either targeting BCL-2 alone or in combination with rituximab presents a new and hopeful way to treat post-transplant EBV-positive B lymphoproliferative disorder. BCL-2 inhibitors might be an attractive strategy against EBV positive lymphoma and subsequent clinical studies are worth exploring.

HDACi

HDACi is generally thought to function as inducers of lytic phase [8, 134] although some have different voices [25]^, [37]. Since antiviral drugs cannot work well during latency, inducing EBV into the lytic phase could enhance its susceptibility. The combination of arginine butyrate and ganciclovir had good anti-tumor activity [135]. Therapy with nanatinostat and valganciclovir also had encouraging efficacy [136]. Except that, HDACi also play a role in suppressing lymphoma growth. In light of the suppression of bim [59, 71] and activation of MYC [64, 67] by EBV, Luo et al. [137] evidenced that chidamide (a kind of Chinese HDACi) reverse the above genetic alterations to inhibit the growth of DLBCL. Xiong et al. [111] found chidamide probably have promising effect for the HEA subtype in NKTCLs. Moreover, Vorinostat [138], a kind of HDAC inhibitor, combined with rituximab plus chemotherapy, appeared to be effective in 12 HIV-related non-Hodgkin lymphoma (2 EBV+). Thus, HDACi could achieve clinical benefits in EBV positive lymphoma although the mechanism remains unclear, and combined treatment [137, 139] with other drugs is worthy of further study.

Hypomethylating drugs

Of note, some studies have shown that the typical characteristic of EBV infection is to cause extensive methylation resulting in immune escape and the progression of diseases such as gastric cancers [140], nasopharyngeal carcinoma [141] and breast cancer [142]. Hypomethylating agents, like decitabine and 5-azacytidine, have an ability to lessen the level of genomic methylation, influence immune responses and reactivate hypermethylated tumor suppressor genes. Therefore, targeting DNA methylation may be a potential novel therapeutic strategy for EBV-associated diseases [22]. Preliminary results [143, 144] proved that decitabine have an inhibitory effect on NKTCL.

Falchi et al. [145] conducted a phase II clinical trial combined 5-azacytidine with romidepsin for treating PTCLs and 48% of patients achieved a CR. Lemonnier et al. [117] reported 12 angioimmunoblastic T-cell lymphoma (AITL) patients who received 5-azacytidine for concomitant myeloid neoplasm or used as compassionate treatment in R/R AITL. However, neither of them [117, 145] elaborates on the EBV status of the patients. Notably, a patient with AITL and chronic myelomonocytic leukemia preceded by an EBER-positive DLBCL [146] had a good response to 5-azacytidine. Our previous reports [115] found that front-line treatment with CHOP plus azacytidine was possibly efficient in EBV-associated patients. These results above resonated well with our assumptions, proving that targeting DNA methylation abnormalities might be an effective strategy to these patients. It was reported that [116, 117] TET2 and DNMT3A mutations regulated the level of methylation among PTCLs. It is possible that EBV upregulates those genes leading to methylation.

The proteasome inhibitors

Mechanistically, the proteasome inhibitors [147] can suppress NF-κB pathway which is regulated positively [76, 77, 106, 107] in most times but negatively [59] sometimes in EBV-positive lymphomas. Zou et al. [148] found bortezomib eliminated B cells with latency III more easily than those with latency I and prolonged the survival in these immunodeficiency mice. In line with this argument, Hui et al. [139] proved that the combination of HDACi and proteasome inhibitors exhibited better effect on latency III BLs than latency I ones. We suspect the proteasome inhibitors have an effect on some specific targets expressed during latency III and look forward more large-sample prospective clinical studies on proteasome therapy for immunodeficiency lymphomas.

CTLs

The phenotypes displayed in EBV-associated Lymphomas like EBNA1, LMP1 and LMP2 are attractive targets for immunotherapy [3]. However, TME is anergic to these few and weak immune cells. Therefore, it is difficult to generate and keep efficient immune responses. Bollard et al. [149] created LMP-CTLs characterized by long-term defense function in TME. The majority (28/29) of high-risk or relapsed patients achieved remission at a median of about 3 years. Prockop et al. [150] found EBV-CTLs had a promising effect for rituximab-refractory EBV-associated lymphomas without severe toxicities. Over 50% cases achieved ORR and one-year OS was 88.9%. Based on the safe and potential effects, EBV-CTLs serve as antigen-presenting cells to attack neoplastic cells constantly and powerfully.

CAR-T

CAR-T therapy is the T cell immunotherapy of chimeric antigen receptor with a promising, accurate as well as rapid effect. CAR-T mainly targets at tumor-specific antigens, such as CD20 and CD19. Slabik et al. [151] developed a kind of CAR-T therapy targeted at gp350 and 75% of mice lower the frequencies of EBER-positive malignant B cells and suppressed tumor development. In line with this argument, Zhang et al. [152] had validated that gp350 CAR-T cells promoted anti-tumor responses, offering an innovative therapy for EBV-associated malignancies. From these results, we suggest that some clinical trials targeting EBV-related phenotypes can be launched. However, CAR-T preparation is expensive and patients are encouraged to participate in clinical trials.

Summary

The present study focuses on the biology, latent modes, viral reactivation, and pathogenic mechanisms about EBV. It also summarizes the clinical and molecular features of EBV-related lymphomas. EBV escapes immune surveillance, promotes lymphoma development and causes treatment resistance through various ways.

The enhanced understanding of the genetic alterations together with specific mechanisms guides novel therapeutic interventions. Strong evidences suggest that EBV infection causes genomic changes and strategies against these changes (such as hypomethylating drugs, ICIs, BTK inhibtors, TKI, and HDACi) may be effective.

The combination of lytic activators (like HDACi) and antiviral drugs can enhance the antiviral effect and achieve better clinical outcomes. Different age groups of EBV-positive DLBCL patients have different latent patterns so treatment options for young and old patients cannot make a blanket statement. It is reported that the proteasome inhibitors have an effect on patients with latency III and we look forward more large-sample prospective clinical studies on proteasome therapy for immunodeficiency lymphomas. These treatment options mentioned above do not seem to be novel or specific to latent type I lymphomas which is featured as restricted EBNA1, so we recommend using EBNA1 inhibitors [105], CTLs or CAR-T to attack neoplastic cells continually and powerfully.

There are many limitations in this study. Because of the complexity of EBV, we have not been able to systematically and clearly describe the pathogenic mechanism of the virus, nor have we been able to accurately describe the targeted treatment of drugs. We also have a lot of confusion. For example, there seems to be some contradiction regarding NF-κB pathway. EBNA1 [59] and BZLF1 [40] inhibits the NF-kB pathway to achieve anticancer effect while EBNA2 [66] and LMP1 [76, 77] activate the NF-kB pathway. Nevertheless, it is clear that the dysregulation of NF-κB pathway (overactivation or inhibition) leads to the dysregulation of related survival proteins, which in turn cause disease progression and chemotherapy resistance.

In the real world, EBV is not the only influencing factor, and other factors, such as CD30 expression [153], will affect the prognosis of lymphoma as well. The optimal treatment among EBV-infected lymphoma varies by lymphomas subtypes, latent pattern and molecular characteristics. More clinical trials are needed for further exploration. Although the mechanism about EBV infection is not clearly elucidated, treatment strategies have unlimited possibilities. It is believed that future treatments for EBV-related lymphomas will be unremittingly updated and more robust.

Author contributions

Jing Chen conceived and designed the study, and wrote the manuscript. Shan Zhang wrote part of the manuscript. Fang Zheng, Fangjun Cheng, Yi Zhao provided critical revisions to the manuscript and approved the final version. All authors read and approved the final manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

As this is a review article based on previously published studies, no ethical approval or consent to participate was required.

Competing interests

The authors declare no competing interests.