Immunocompromised individuals, including those infected with human immunodeficiency virus (HIV), are at increased risk of Epstein–Barr virus (EBV)-associated aggressive B cell malignancies such as Burkitt’s lymphoma (BL) or diffuse large B cell lymphoma (DLBCL). Differential diagnosis of these lymphomas requires histopathological, immunohistochemical and cytogenetic assessments. Rapid, less invasive approaches to the diagnosis of EBV-associated B cell lymphomas are needed. Here, high-throughput cytokine profiling of BL cell lines and EBV-transformed B lymphoblastoid cell lines (B-LCL), representing DLBCL, was carried out. By monitoring the production of 42 different cytokines, unique cytokine signatures were identified for BL and B-LCL/DLBCL. The BL cells produced interleukin (IL)-10, 10 kDa interferon gamma-induced protein (IP-10)/CXCL10, macrophage-derived chemokine (MDC)/CCL22, macrophage inflammatory protein (MIP)-1α/CCL3 and MIP-1β/CCL4. In addition to these five cytokines, the cytokine signature of B-LCL/DLBCL cells included IL-8/CXCL8, IL-13, platelet-derived growth factor (PDGF)-AA, and regulated upon activation, normal T cell expressed and secreted (RANTES)/CCL5. Epstein–Barr virus latency was responsible for the increased production of IL-10, MDC/CCL22 and MIP-1α/CCL3 in BL cells, suggesting that EBV-mediated BL-genesis involves these three cytokines. These results suggest that high-throughput cytokine profiling might be a valuable tool for the differential diagnosis and might deepen our understanding of the pathogenesis of EBV-associated B cell malignancies. (Cancer Sci 2011; 102: 1236–1241)
Epstein–Barr virus (EBV) is a human gamma herpes virus that has been linked to the development of Burkitt’s lymphoma (BL) and diffuse large B cell lymphoma (DLBCL) in immunocompromised individuals, including those with acquired immunodeficiency syndrome (AIDS).(1–4) B lymphoblastoid cell lines (B-LCL), which arise from the EBV-transformed peripheral B cells, serve as a model of EBV-associated DLBCL.(5) The EBV-positive BL and EBV-transformed B-LCL and some BL cell lines produce an array of cytokines at varying levels, including interleukin (IL)-6, IL-8/CXCL8, IL-10, monocyte chemoattractant protein (MCP)-1/CCL2, macrophage-derived chemokine (MDC)/CCL22, macrophage inflammatory protein (MIP)-1α/CCL3, MIP-1β/CCL4, regulated upon activation, normal T cell expressed and secreted (RANTES)/CCL5, thymus and activation-regulated chemokine (TARC)/CCL17, tumor necrosis factor (TNF)α and TNFβ.(6–12) Certain cytokines can be detected in the sera of individuals with EBV-associated malignancies, which suggests an active role for these and perhaps other cytokines in disease pathogenesis.(7,13,14) Certain cytokines, such as IL-6 or IL-10, have been proposed as diagnostic and/or prognostic markers of EBV-associated B cell malignancies.(7,9,15) However, these cytokines are not widely used clinically primarily due to their lack of specificity and sensitivity. One way to demonstrate the utility of cytokine profiles as a specific diagnostic marker is to analyze a number of different cytokines in parallel.
Epstein–Barr virus establishes latent infection in the infected cells in which only a limited number of viral latency-associated genes are expressed, including EBNA and LMP. In this latency, EBV does not produce progeny viruses actively. There are three types of EBV latency, type I, II and III,(5) each of which is associated with the expression of different sets of latent genes. In BL cells, EBV type I latency in vivo involves the expression of only a few viral latent genes, including EBNA1, BARF0, EBER.(5) In DLBCL, EBV establishes type III latency, in which all of the viral latent genes are expressed, including EBNA2, EBNA3, EBNA-LP, LMP, in addition to the type I latency associated genes.(5) Cytokines produced by EBV-infected B cells during type III latency are well documented because the genes expressed during type III latency, including LMP-1, are strong inducers of cytokine production. In contrast, other than IL-10, the cytokines produced by BL cells in type I latency are less well characterized.
Historically, the dysregulation of cytokine production by latent EBV infection has been studied by comparing the cytokine profiles of EBV-positive and EBV-negative BL cell lines, or by monitoring cytokine production of primary B cells after EBV infection, or EBV-negative BL cell lines transfected with EBV genes.(4,11,16,17) However, there are several caveats to these methods. Comparisons using cell lines might not be sensitive enough to detect correlations between EBV status and cytokine production because baseline cytokine production varies among cell lines. The analysis of cytokine dysregulation induced by type I latency has been problematic because EBV infection of primary B cells and EBV-negative BL cells in vitro results in type III latency. Finally, transfection protocols typically use cells that are non-B cell in origin, or are limited to a few EBV-negative B cell lines, and the results from such experiments might not be relevant to cytokine production by BL cells with type I EBV latency.
To achieve better sensitivity and accuracy in detecting altered cytokine production associated with latent EBV infection, especially type I latency, the preferred method is a comparison of EBV-positive cells with the EBV negative from the same origin. Recently, EBV-negative cell clones were successfully isolated from the EBV-positive BL cell lines Akata, Daudi and Mutu.(18,19) These cells have previously been used as a model system for elucidating the oncogenic role of EBV in type I BL cells.(20–22) Dysregulation of cytokine production induced by type I EBV latency should be possible through systematic cytokine profiling of these EBV-positive and EBV-negative BL cells.
In the current study, we assessed the production of 42 cytokines by EBV-positive type I BL cell lines and their corresponding EBV-negative counterparts at the protein level using a high-throughput microbead-based system. We also analyzed cytokine production of B-LCL, a BL-like cell line, as well as T lymphoblastoid cell lines and primary blood mononuclear cells (PBMC). We identified cytokine signatures that are characteristic of type I BL and B-LCL that might be useful in the differential diagnosis and prognosis of EBV-associated B cell malignancies.
Materials and Methods
Cells. Cells were maintained in RPMI 1640 medium (Sigma, St. Louis, MA, USA) supplemented with 10% fetal bovine serum (Japan Bioserum, Tokyo, Japan), 50 U/mL penicillin and 50 μg/mL streptomycin (Invitrogen, Tokyo, Japan) at 37°C in a humidified 5% CO2 atmosphere. The following cells lines were used: Akata, a Japanese EBV-positive BL cell line (non-AIDS related);(23) Daudi, an African EBV-positive BL cell line (non-AIDS related);(6) BJAB, an African EBV-negative Burkitt-like lymphoma cell line that lacks the c-myc translocation (non-AIDS related);(24) and Mutu, an African EBV-positive BL cell line.(25) The B-LCL were established by infecting healthy donor-derived B cells with EBV of Akata(26) or B95-8 origin.(27) B95-8-derived recombinant EBV carrying a G418-resistance gene has been developed by Shimizu et al.(28) as Akata virus. The B-LCL were generated as oligoclonal pools. The CEM, Jurkat and MOLT-4 cell lines were derived from acute T cell lymphoblastic leukemias. MT-4 is a T cell line transformed with human T-cell lymphotropic virus-1 (HTLV-1).(29) An epithelial cell line MKN28 is derived from well-differentiated tubular adenocarcinoma,(30) and the recombinant Akata(28) was used to convert MKN28.(31)
Cytokine assay. Cytokine assays were carried out using a human cytokine/chemokine kit (#MPXHCYTO60KPMX42; Millipore, Tokyo, Japan). Tissue culture supernatants were collected 3–4 days post-cultivation when the cell density was approximately 4–7.5 × 105 cells/mL. Signals were detected using a Luminex 200 system (Luminex, Austin, TX, USA) operated with xponent 3.1 software (Luminex).
Cytokine production profile of BL cells during different types of EBV latency. We monitored the production of 42 cytokines by three type I BL cell lines (Akata, Daudi and Mutu) and their virus-negative counterparts. The data are summarized in Figure 1 and color-coded according to cytokine concentration (Fig. 1). There were five cytokines that were common to all type I BL cell lines: IL-10, IP-10/CXCL10, MDC/CCL22, MIP-1α/CCL3 and MIP-1β/CCL4, albeit at varying levels (Table 1). This suggested that these cytokines might be involved in BL pathogenesis and that this cytokine signature might be characteristic of type I EBV latency in BL (Fig. 1). To investigate whether this pattern of cytokine production was specifically induced by EBV, we compared the types and concentrations of cytokines produced by type I BL cells with their corresponding EBV-negative counterparts (Fig. 2). The color-coded scale in Figure 1 was sufficient to represent overall trends but was not suitable for analyzing changes in cytokine concentrations. Thus, a new color-coded scale was applied to analyze the magnitude of the differences in cytokine concentrations between EBV-positive and EBV-negative cells in more detail. The levels of IL-10, MDC/CCL22 and MIP-1α/CCL3 were 11.9-, 185.1- and 2.4-fold higher, respectively (average values for all three cell lines), in type I BL cells than in their EBV-negative counterparts (Fig. 2), suggesting that these three cytokines are upregulated by type I EBV latency and might mediate in part the pathogenic potential of EBV. Other cytokines, such as IP-10/CXCL10 and MIP-1β/CCL4, were upregulated in a cell line-dependent manner, which suggests that their expression does not have to be upregulated by EBV during the evolution of BL (Fig. 2).
Table 1. Summary of cytokines characteristic to transformed B cells with distinct EBV latency
BL, Burkitt’s lymphoma; B-LCL, B lymphoblastoid cell lines; DLBCL, diffuse large B cell lymphoma; EBV, Epstein–Barr virus.
Somewhat counterintuitively, the production of certain cytokines in EBV-negative BL cells was downregulated by type I EBV latency. The profile of downregulated cytokines was unique to each cell line. In Akata cells, fibroblast growth factor-2 (FGF-2), granulocyte colony-stimulating factor (G-CSF), IL-1α, IL-2, IL-3 and IL-4 were downregulated 2.5- to 3.3-fold (Fig. 2). These cytokines were upregulated during type III EBV latency, which indicates that Akata cells are able to express them. One possible explanation for these results is that EBV expresses a repressor of these cytokines during type I latency in Akata cells, whereas during type III latency, EBV expresses a strong inducer of these cytokines that can counteract the repressor function. In Daudi cells, type I EBV latency resulted in a 3.3-fold decrease in IP-10/CXCL10, whereas in the other two cell lines (Akata and Mutu), IP-10/CXCL10 was upregulated by type I EBV latency (Fig. 2).
Akata and Mutu cells support type III EBV latency. Infection of EBV-negative Akata cells with B95-8 EBV results in type III latency. Mutu cells with type III EBV latency have been isolated in vitro, which was a spontaneous shift from type I–III latency. Type III Akata and Mutu cells produced IL-10, IP-10/CXCL10, MDC/CCL22, MIP-1α/CCL3, MIP-1β/CCL4 and RANTES/CCL5 at high levels, a signature that, with the exception of RANTES/CCL5, overlapped that of BL cells (Fig. 1). Type III Mutu cells were unique in that they also produced high levels of IFN-α2 (Fig. 1). Cytokine production in type III BL cells was more active than in type I cells (Fig. 1). The magnitude of IL-10, MDC/CCL22 and MIP-1α/CCL3 induction during type III EBV latency was 48.6-, 5441.8- and 65.1-fold, respectively (average values for two cell lines), in Akata and Mutu cells (Fig. 2). These results suggest that type III latency-associated viral genes are more potent cytokine inducers than those expressed during type I latency. Although the production levels were modest, the many other following cytokines were upregulated in type III BL cells compared with their EBV-negative counterparts: FGF-2, Flt-3L, fractalkine/CX3CL1, IFNγ, IL-1α, IL-1β, IL-1ra, IL-6, IL-8/CXCL8, IL-10, IL-12(p70), IL-15, IL-17, IP-10/CXCL10, MCP-1/CCL2, MCP-3/CCL7, MDC/CCL22, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, sCD40L, TNFα, TNFβ and vascular endothelial growth factor (VEGF) (Fig. 2).
Previously, we reported that EBV-encoded small RNA (EBER) play a role in the induction of IL-10 in Akata cells and contribute to their malignant phenotype.(22,32) We investigated whether the expression of EBER was responsible for the induction of type I BL cytokines. The expression of EBER in EBV-negative Akata cells resulted in the upregulation of IL-10 production by approximately 2.5-fold (Fig. 1), consistent with a previous report.(22) However, the levels of MDC/CCL22 and MIP-1α/CCL3 were unaffected by EBER (Fig. 1). These results suggest that an as-yet unidentified viral factor is responsible for the upregulation of MDC/CCL22 and MIP-1α/CCL3. Ectopic expression of EBER also failed to downregulate the levels of FGF-2, G-CSF, IL-1β, IL-2, IL-3 and IL-4 in Akata cells, which suggests that EBER do not function as a repressor of these cytokines during type I EBV latency (Fig. 1).
Cytokine signature of the BL-like cell line BJAB. The BL-like cell line BJAB exhibited a similar cytokine profile to BL cells but with some key differences. Both types of cells produced IL-10, IP-10/CXCL10, MDC/CCL22, MIP-1α/CCL3 and MIP-1β/CCL4 (Fig. 1), which was the BL cytokine signature. This is noteworthy because the histology of BL-like lymphoma is similar to BL, even though BL-like lymphoma does not harbor an Ig/c-myc translocation and is negative for EBV, strongly suggesting the role of BL signature cytokines in the pathogenesis. Unlike the BL cell lines, BJAB cells produced high levels of fractalkaine/CX3CL1, RANTES/CCL5, TNFβ and VEGF (Fig. 1). The level of MDC/CCL22 was substantially lower than the BL cell lines. Thus, the BJAB cytokine signature was distinct from the BL cytokine signature. Epstein–Barr virus infection of BJAB cells, which results in type III latency, was associated with the upregulation of IL-4, IL-8/CXCL8, MDC, sIL-2Ra, TNFα and TNFβ, and downregulation of eotaxin/CCL11, FGF-2, Flt-3L, IL-1ra, IL-6, IL-12(p70), IL-13, IP-10/CXCL10, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5 and VEGF (Fig. 2). The levels of MIP-1α/CCL3 and MIP-1β/CCL4 production in EBV-infected BJAB cells were markedly reduced to 0.6% and 2.3%, respectively, relative to the uninfected cells (Fig. 2). This strong repression of cytokine production by EBV was unique to BJAB cells. Similar to BJAB cells, EBV also establishes type III latency in Akata and Mutu cells. Type III EBV latency in Akata and Mutu cells did not result in the downregulation of any cytokines (Fig. 2). These results indicated that the cellular genetic background of BJAB, a BL-like lymphoma, differs from that of BL cells. To our knowledge, this is the first report of differences in cytokine production associated with type III EBV latency in BL and BL-like cells.
Previously, it was reported that MDC/CCL22 levels are upregulated upon EBV infection in BJAB cells, and that LMP-1 induces the expression of IL-6, IL-8/CXCL8, IL-10, IP-10/CXCL10, RANTES/CCL5, TNFα and TNFβ.(12,16,17,33–35) In Akata cells, we reproduced the upregulation of these cytokines by LMP-1, although the color code did not show modest upregulation of IL-6, IL-8/CXCL8, TNFα and TNFβ for their low expression levels (2.9-, 6.9-, 8.3-, and 11.0-fold, respectively; Fig. 1). Our data were consistent with these earlier results in terms of induction of IL-8/CXCL8, MDC/CCL22, TNFα and TNFβ by type III EBV latency in BJAB cells, which express LMP-1. However, unlike previous results, type III EBV latency in BJAB cells was not associated with the upregulation of IL-10 and resulted in the downregulation of IL-6, IP-10/CXCL10 and RANTES/CCL5. One explanation for these seemingly controversial results is that there might be viral genes expressed during type III EBV latency in BJAB cells that can counteract the function of LMP-1. It is worth noting that the analysis of viral gene function using transfection assays might not accurately reflect events that are relevant to viral infection.
Cytokine production profile of B-LCL. There were more cytokines produced by B-LCL than BL cells, and the levels of the cytokines produced were higher than in BL cells (Fig. 1). The cytokine profiles varied among B-LCL from different donors, and the cytokine profiles of independent cultures of B-LCL from the same donor also exhibited some variation (Fig. 1). However, almost all the B-LCL produced the following cytokines: IL-8/CXCL8, IL-10, IL-13, IP-10/CXCL10, MDC/CCL22, MIP-1α/CCL3, MIP-1β/CCL4, PDGF-AA, RANTES/CCL5, TNFα, TNFβ and VEGF (Fig. 1). The cytokine signature of the B-LCL was distinct from the BL and BL-like cell lines in that the levels of certain cytokines, namely MIP-1β/CCL4, PDGF-AA, RANTES/CCL5 and TNFβ, were substantially higher (Fig. 1). There was no evidence of virus strain-specific cytokine signatures in B-LCL, in that the cytokine profiles induced by EBV of B95-8 and Akata origin were similar.
Cytokine production footprints for other cell types. To determine the specificity of the BL and B-LCL cytokine signatures, we investigated the cytokine production profiles of PBMC from a healthy donor and several different T cell lines. Cytokine production by naïve PBMC was active and the following cytokines were produced at high levels: granulocyte–macrophage colony-stimulating factor (GM-CSF), growth-related oncogene (GRO), IL-1ra, IL-8/CXCL8, MCP-1/CCL2, MDC/CCL22, sIL-2Ra and TNFα (Fig. 1). Unlike the B-LCL, PBMC were poor producers of IL-10, MIP-1α/CCL3 and TNFβ (Fig. 1). These results suggest that the cytokine signature of PBMC is distinct from that of BL and B-LCL. We also analyzed three acute T cell lymphoblastic leukemia-derived cell lines (CEM, Jurkat and MOLT-4) and the HTLV-1-transformed MT-4 T cell line. The T cell lines produced only a few cytokines, namely VEGF (by MOLT-4 cells), MIP-1α/CCL3, RANTES/CCL5 and sIL-2Ra (by MT-4 cells) (Fig. 1). Importantly, none of the T cell lines produced detectable levels of IL-10, IL-12 (p40), IL-13, MDC/CCL22 and MIP-1β/CCL4, a cytokine phenotype that could theoretically be used to distinguish between tumor cells of T cell and B cell origin (Fig. 1). All four T cell lines exhibited unique cytokine signatures, which suggests that these cell lines are genetically divergent and that the cytokines produced might play distinct roles in the pathogenesis of T cell malignancies. Additionally, we analyzed an epithelial cell line, MKN28, in which EBV establishes type I latency. According to the comparison between EBV-positive and EBV-negative counterparts (Fig. 2), EBV type I latency in MKN28 cells upregulated expressions of G-CSF, GRO, IL-1α, IL-4, IL-8/CXCL8, IL-10 and IP-10/CXCL10, whereas it downregulated expressions of MDC/CCL22, PDGF-AA/BB, sLI-2Ra and TNFα. This profile was unique to MKN28 cells, as highlighted by the reduction of MDC/CCL22, which was increased in all the BL cell lines upon type I EBV latency. Interestingly, however, the induction of IL-10 was consistently seen among all cell lines with type I EBV latency. Overall, the results indicate that the cytokine signatures of BL and B-LCL are unique and can be differentiated from those of other malignancies.
In the present study we focused on BL and DLBCL, which occur with relatively high frequency in AIDS patients. Cytokine production profiling revealed that BL cell lines and B-LCL, which represent DLBCL, have unique cytokine signatures (Table 1). Serum IL-6 or IL-10 concentration has been proposed as a diagnostic and prognostic marker for EBV-associated B cell malignancies,(8,15) and profiling more than one cytokine (i.e. IL-6, IL-10 and/or TNFβ) might be more informative than a single cytokine measurement in terms of a differential diagnosis.(9) The 42 cytokine profiles investigated herein represent a potent diagnostic and prognostic tool. Currently, invasive approaches such as needle biopsy are required to reach a differential diagnosis for B cell malignancies. Serum cytokine profiling is less invasive than current approaches, can be performed rapidly using a microbead-based assay and does not require special expertise to reach a diagnosis, as a pathological examination does. These aspects represent clear advantages to cytokine profiling over current invasive diagnostic procedures. High-throughput cytokine profiling of multiple cytokines represents a potentially useful tool for the differential diagnosis of EBV-associated B cell malignancies with distinct latencies. However, EBV-associated B cell malignancies represent just a subset of all lymphomas. To establish cytokine profiling as a diagnostic tool for all malignant lymphomas, a systematic cytokine analysis of a broad and diverse set of lymphomas is needed.
Cytokines play important roles in the pathogenesis of cancer, including the enhancement of cancer cell proliferation, metastasis, angiogenesis and immune disturbance.(36,37) The cytokines that were upregulated in almost all of the B-LCL and in all of the type I cells are most likely involved in B cell lymphoma pathogenesis. In the case of BL, IP-10/CXCL10 might be involved in part in tissue necrosis and vascular damage.(38) MDC/CCL22 recruits Th2 and regulatory T cells, and might help tumor cells escape Th1-mediated immune surveillance.(38) Although the major function of MIP-1α/CCL3 and MIP-1β/CCL4 is the recruitment of Th1 cells,(38) they might also attract regulatory T cells in vivo that help tumor cells counteract host immune defenses.(39) In B-LCL, cytokines such as IL-8/CXCL8, PDGF-AA and VEGF might be involved in local angiogenesis, IL-10 could function as an autocrine/paracrine cell growth factor, and MDC/CCL22, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, TNFα and TNFβ might be involved in immune disturbance. The precise roles of these cytokines in the pathogenesis of EBV-associated B cell malignancies remain to be elucidated. Furthermore, the viral gene product(s) responsible for the upregulation of these cytokines should be clarified, as it might lead to novel molecular therapeutic approaches to EBV-associated malignancies.
The BL, BL-like and B-LCL cell lines each differed from the others in their response to type I and III EBV latency. What then is the molecular mechanism underlying these distinct B cell responses to EBV latency? The BL cells are assumed to have originated from the germinal center at late centrocyte stage.(40) The B-LCL are generated from immature-to-mature B cells that express CD21/CR2, a cell surface receptor for EBV. The origin of BL-like lymphoma is not fully understood. The BL-like lymphoma cells and B-LCL do not harbor an Ig/c-myc translocation, as do BL cells. The distinct responses to EBV latency exhibited by each of these cells might be due to differences in their cellular genetic backgrounds that result in differential epigenetic regulation of gene expression, reflected in the induction of certain cytokines by viral genes. Different BL cells exhibited distinct patterns of cytokine regulation during type I or III EBV latency, suggesting that their genetic backgrounds are also different. The unique response of Daudi cells might also be due to the fact that Daudi EBV lacks a part of the viral genome.(41) It remains to be determined whether the peculiarity of BJAB cells to EBV latency is unique to BL-like cells or to BJAB cells.
Using EBV-positive and EBV-negative cells of the same origin, we showed that type I EBV latency results in the upregulation of three cytokines, two of which have been overlooked to date because of the limitations of traditional approaches to cytokine profiling of type I latency. Future studies should be designed to investigate the pathophysiological roles of the BL and B-LCL signature cytokines to better understand the pathogenesis of BL and DLBCL. These cytokines might also represent potential molecular therapeutic targets.
The authors thank Kenji Ota and Toru Utsunomiya (Luminex, Japan) for technical assistance. K.M. is a research resident of the Japan Foundation for AIDS Prevention. E.U. is a research resident of Japan–North America Medical Exchange Foundation.
The authors declare no financial or commercial conflict of interest.