Interplay between the Epigenetic Enzyme Lysine (K)-Specific Demethylase 2B and Epstein-Barr Virus Infection

In Africa, Epstein-Barr virus infection is associated with endemic Burkitt lymphoma, a pediatric cancer. The molecular events leading to its development are poorly understood compared with those leading to sporadic Burkitt lymphoma. In a previous study, by analyzing the DNA methylation changes in endemic compared with sporadic Burkitt lymphoma cell lines, we identified several differential methylated genomic positions in the proximity of genes with a potential role in cancer, and among them was the KDM2B gene. KDM2B encodes a histone H3 demethylase already shown to be involved in some hematological disorders. However, whether KDM2B plays a role in the development of Epstein-Barr virus-mediated lymphoma has not been investigated before. In this study, we show that Epstein-Barr virus deregulates KDM2B expression and describe the underlying mechanisms. We also reveal a role of the demethylase in controlling viral and B-cell gene expression, thus highlighting a novel interaction between the virus and the cellular epigenome.

ABSTRACT

The histone modifier lysine (K)-specific demethylase 2B (KDM2B) plays a role in the differentiation of hematopoietic cells, and its expression appears to be deregulated in certain cancers of hematological and lymphoid origins. We have previously found that the KDM2B gene is differentially methylated in cell lines derived from Epstein-Barr virus (EBV)-associated endemic Burkitt lymphoma (eBL) compared with that in EBV-negative sporadic Burkitt lymphoma-derived cells. However, whether KDM2B plays a role in eBL development has not been previously investigated. Oncogenic viruses have been shown to hijack the host cell epigenome to complete their life cycle and to promote the transformation process by perturbing cell chromatin organization. Here, we investigated whether EBV alters KDM2B levels to enable its life cycle and promote B-cell transformation. We show that infection of B cells with EBV leads to downregulation of KDM2B levels. We also show that LMP1, one of the main EBV transforming proteins, induces increased DNMT1 recruitment to the KDM2B gene and augments its methylation. By altering KDM2B levels and performing chromatin immunoprecipitation in EBV-infected B cells, we show that KDM2B is recruited to the EBV gene promoters and inhibits their expression. Furthermore, forced KDM2B expression in immortalized B cells led to altered mRNA levels of some differentiation-related genes. Our data show that EBV deregulates KDM2B levels through an epigenetic mechanism and provide evidence for a role of KDM2B in regulating virus and host cell gene expression, warranting further investigations to assess the role of KDM2B in the process of EBV-mediated lymphomagenesis.

IMPORTANCE In Africa, Epstein-Barr virus infection is associated with endemic Burkitt lymphoma, a pediatric cancer. The molecular events leading to its development are poorly understood compared with those leading to sporadic Burkitt lymphoma. In a previous study, by analyzing the DNA methylation changes in endemic compared with sporadic Burkitt lymphoma cell lines, we identified several differential methylated genomic positions in the proximity of genes with a potential role in cancer, and among them was the KDM2B gene. KDM2B encodes a histone H3 demethylase already shown to be involved in some hematological disorders. However, whether KDM2B plays a role in the development of Epstein-Barr virus-mediated lymphoma has not been investigated before. In this study, we show that Epstein-Barr virus deregulates KDM2B expression and describe the underlying mechanisms. We also reveal a role of the demethylase in controlling viral and B-cell gene expression, thus highlighting a novel interaction between the virus and the cellular epigenome.

INTRODUCTION

Epstein-Barr virus (EBV) is a human gammaherpesvirus that infects more than 95% of the adult population worldwide. After infection, EBV establishes a lifelong latency, often with no adverse health consequences. Despite its ubiquity, EBV infection is also associated with many human cancer types, among which is endemic Burkitt lymphoma (eBL), the most common childhood cancer in equatorial Africa (1). Although this malignancy was associated with EBV infection more than 50 years ago, the exact mechanism by which the virus contributes to the eBL pathogenic process is still not fully understood.

Many studies have highlighted a key role of epigenetic deregulations in cell transformation and cancer development. Increasing evidence indicates that different viruses may abrogate cellular defense systems by hijacking epigenetic mechanisms to deregulate the host cell gene expression program and modulate their own life cycle (2, 3). Our recent study of the methylome profiles of sporadic Burkitt lymphoma (sBL)- versus eBL-derived cell lines revealed an EBV infection-specific pattern of methylation, with aberrant methylation being detected in genes with a known role in lymphomagenesis, such as ID3, which is often found to be mutated in sBL (4). We therefore hypothesized that a virus-driven mechanism is responsible for modifying the epigenome of B cells to facilitate the lymphomagenic process, circumventing the need for mutations in lymphoma driver genes. Among the genes differentially methylated in eBL compared with sBL, we identified the lysine (K)-specific demethylase 2B (KDM2B) gene, which encodes a histone H3 demethylase known to target specific sites, such as trimethylated lysine 4 (H3K4me3) and dimethylated lysine 36 (H3K36me2). KDM2B sets the stage for DNA methylation and gene silencing by recruiting polycomb-1 proteins to unmethylated CpG regions (5) and plays a key role in somatic cell reprogramming (6). It also represses the transcription of rRNA genes, thus inhibiting cell growth and proliferation (7). KDM2B has been identified as a putative tumor suppressor by retroviral insertion analysis in mice (8). Low levels of KDM2B expression have been found in aggressive brain tumors, suggesting its potential role in cancer development. Moreover, KDM2B is involved in hematopoietic cell development and plays opposite roles in tumors of hematopoietic and lymphoid origins (9). Although high levels of KDM2B expression have been observed in different hematological malignancies, its depletion from hematopoietic cells has been reported to activate the cell cycle and reduce the activity of interferon and lymphoid-specific transcription factors, thereby contributing to myeloid transformation (9). However, whether KDM2B affects the EBV life cycle has not been determined, and its role in eBL has not been assessed. Here, using in vitro EBV infection models, we aimed to assess whether EBV can alter the expression of KDM2B by inducing methylation of its gene. Finally, we investigated how this event affects EBV infection and B-cell homeostasis. Overall, our data highlight a novel cross talk between EBV and the cellular epigenome and identify KDM2B to be a master regulator of EBV gene expression, in addition to B-cell gene expression, suggesting a role for EBV-mediated KDM2B deregulation in the lymphomagenic process.

(This article was submitted to an online preprint archive [10].)

RESULTS

KDM2B is epigenetically silenced in EBV(+) BL-derived cell lines.

Our previous comparative analysis of the whole-genome methylation profiles of a set of EBV-positive [EBV(+)] and EBV-negative [EBV(−)] Burkitt lymphoma (BL)-derived cell lines (4) led to the identification of two CpGs (CpG15695155 and CpG21423404) flanking a CpG island named CpG127 (Fig. 1A) in an intragenic putative regulatory region of KDM2B (as shown by the accumulation of the H3K27 acetylation [H3K27Ac] marker) (Fig. 1A). CpG15695155 and CpG21423404 were highly methylated in EBV(+) BL-derived cells compared with EBV(−) BL-derived cells. Here, to validate these data we performed direct pyrosequencing on DNA extracted from 10 EBV(+) BL-derived cell lines and 9 EBV(−) BL-derived cell lines (Table 1). The samples for which the pyrosequencing gave results technically suitable for analysis are displayed in the histogram in Fig. 1B. Pyrosequencing analysis confirmed that the KDM2B gene is hypermethylated at CpG15695155 and CpG21423404 in EBV(+) BL cell lines compared with EBV(−) BL cell lines (Fig. 1B). In contrast, we did not observe high methylation levels or differences between EBV(+) and EBV(−) BL cell lines when analyzing 17 positions within the CpG island 127 (Fig. 1C). Next, we assessed whether the high DNA methylation level of the KDM2B gene would affect its expression level. Treatment of 3 EBV(−) BL and 3 EBV(+) BL cell lines with the demethylating agent 5-aza-2′-deoxycytidine (Aza) for 48 h led to a significant rescue of KDM2B expression in EBV(+) BL cells, whereas this treatment had no noticeable effect on KDM2B mRNA expression in EBV(−) BL cells (Fig. 1D). Pyrosequencing analysis of DNA from EBV(+) and EBV(−) BL cell lines exposed to Aza or to dimethyl sulfoxide (DMSO) for 48 h revealed a moderate but significant reduction in the methylation level at CpG21423404 in Aza-exposed EBV(+) BL cells, whereas methylation at all the other positions analyzed remained unchanged (Fig. 1E). We then determined whether the different methylation patterns observed in EBV(+) and EBV(−) BL cells affected KDM2B protein expression. We analyzed KDM2B protein expression in 3 EBV(+) and 3 EBV(−) BL-derived cell lines and observed a significantly lower expression level of KDM2B protein in EBV-infected cells (Fig. 1F). Moreover, we analyzed 11 EBV(+) BL and 11 EBV(−) BL samples (Table 2) by immunohistochemistry for the KDM2B protein expression level; 8 of the 11 EBV(+) BL samples showed weak KDM2B staining, and 10 of the 11 EBV(−) BL samples showed a strong signal for KDM2B immunohistochemistry, suggesting that EBV infection induces reduced expression of the KDM2B protein in vivo (Fig. 1G). Of note, the EBV(+) samples with stronger KDM2B staining had fewer Epstein-Barr virus-encoded small RNA (EBER)-expressing cells, as determined by EBER in situ hybridization (ISH) (data not shown). In conclusion, these data show that two specific CpG sites in the regulatory region of KDM2B are hypermethylated in EBV(+) BL cell lines compared with EBV(−) BL cell lines, confirming our previous whole-genome methylation profiling data (4). Moreover, KDM2B expression also appears to be reduced in eBL specimens, which is probably mediated by DNA methylation at CpG21423404. These data suggest that EBV may regulate KDM2B expression by inducing the methylation of a specific position within a regulatory region of its gene.

FIG 1.

The KDM2B gene is methylated and silenced in EBV(+) BL cell lines and specimens. (A) Schematic diagram of the KDM2B gene (modified from the UCSC Genome Browser). Red lines show CpG15695155 and CpG21423404, and CpG island 127 is in light blue. ChIP data (obtained with the lymphoma cell line GM12878) for the distribution of the H3K27Ac marker within the selected region are also shown. ChIP, chromatin immunoprecipitation. (B) The histograms show the average percentage of methylation measured by pyrosequencing of CpG15695155 and CpG21423404 in the DNA of 10 EBV(+) and 9 EBV(−) BL cell lines (****, P < 0.0001; **, P < 0.01). (C) The histogram shows the average percentage of methylation at 17 positions within CpG island 127 measured in 2 EBV(+) and 2 EBV(−) BL cell lines. The difference between the two groups was not significant. (D) Three EBV(+) and 3 EBV(−) BL cell lines were cultured in the presence of dimethyl sulfoxide (DMSO; nontreated [nt]) or 5-aza-2′-deoxycytidine (Aza; 10 μM) for 48 h. The KDM2B mRNA expression level was evaluated by RT-qPCR. The pooled results of 4 independent experiments are presented in the histogram (**, P < 0.01; ns, not significant). KDM2B mRNA levels in Aza-treated cells were measured relative to the levels in DMSO-treated control cells. (E) Aza-treated EBV(−) and EBV(+) BL cells from 2 independent experiments were processed for DNA extraction and analyzed by pyrosequencing for the methylation level at CpG15695155 (CG15) and CpG21423404 (CG21) as well as for the average methylation at 17 positions within CpG island 127 (CG127). The average difference in the percentage of methylation between Aza- and DMSO-treated cells is indicated in the histogram (*, P < 0.05; ns, not significant). (F) Three EBV(+) and 3 EBV(−) BL cell lines were cultured and analyzed by immunoblotting for KDM2B expression levels. The histogram shows the average KDM2B expression levels normalized to the β-actin signal, measured in 4 independent experiments by Image Lab software (Bio-Rad) in EBV(+) versus EBV(−) BL cells (*, P < 0.05). (G) KDM2B levels in EBV(+) and EBV(−) BL samples were analyzed by immunohistochemistry. The same samples were analyzed for EBER expression by ISH for EBER (EBER-ISH) as described in Materials and Methods. The images shown are representative of the KDM2B staining obtained in 11 EBV(+) and 11 EBV(−) BL specimens.

TABLE 1.

Description of BL-derived cell lines used in the present study^a

BL case identifier	Diagnosis	EBV infection	Cytogenetic information	Clinical data	Ethnic origin
BL103	BL	EBV(−)	t(8;14)		Caucasian
BL70	BL	EBV(−)	t(8;14)	Dec	Caucasian
BL56	BL	EBV(−)	t(8;14)	Dec
BL58	BL	EBV(−)	t(8;14)	Dec	Caucasian
BL53	BL	EBV(−)	t(8;14)	Dec	Caucasian
BL102	BL	EBV(−)	t(8;22)	Dec	Caucasian
BL104	BL	EBV(−)	t(8;22)		Caucasian
BL2	BL	EBV(−)	t(8;22)	Dec	Caucasian
BL41	BL	EBV(−)	t(8;14)		Caucasian
BL110	BL	EBV(+)	t(8;14)		Caucasian
BL135	BL	EBV(+)	t(8;14)		African
BL65	BL	EBV(+)	t(8;14)		African
BL116	BL	EBV(+)	t(8;14)		African
BL60	BL	EBV(+)	t(8;22)	Dec	African
BL79	BL	EBV(+)	t(8;14)		African
BL112	BL	EBV(+)	t(8;14)		Caucasian
I100		EBV(+)
I373		EBV(+)
I176B		EBV(+)

^aThe cells were obtained from the IARC Biobank. Abbreviations: BL, Burkitt lymphoma; EBV, Epstein-Barr virus; Dec, deceased.

TABLE 2.

BL case main features^a

BL subtype	Age (yr)	Sex	Site of biopsy	Detection of the following:
				EBER	MYC by FISH B.A.	MYC-IGH by FISH	MYC-IGK or IGL by FISH
sBL	40	F	Inguinal lymph node	−	+	n.p.	n.p.
sBL	18	M	Ileum	−	+	n.p.	n.p.
sBL	38	F	Lymph node	−	+	n.p.	n.p.
sBL	40	F	Bone marrow	−	+	n.p.	n.p.
sBL	20	M	Lymph node	−	−	−	n.p.
sBL	15	M	Lymph node	−	+	n.p.	n.p.
sBL	25	F	Ileum	−	+	n.p.	n.p.
sBL	45	M	Stomach	−	+	+	n.p.
sBL	16	M	Abdomen mass	−	+	n.p.	n.p.
sBL	20	M	Lymph node	−	−	−	n.p.
sBL	14	M	Lymph node	+	+	n.p.	n.p.
eBL	9	F	Abdomen mass	+	+	n.p.	n.p.
eBL	5	F	Soft tissues	+	+	n.p.	n.p.
eBL	6	F	Maxilla	+	+	n.p.	n.p.
eBL	10	F	Ileum	+	+	n.p.	n.p.
eBL	3	F	Lymph node	+	+	n.p.	n.p.
eBL	4	F	Lymph node	+	+	+	n.p.
eBL	10	M	Maxilla	+	+	−	+ (IGL), − (IGK)
eBL	7	M	Abdomen	−	+	n.p.	n.p.
eBL	12	F	Maxilla	−	+	n.p.	n.p.
eBL	3	M	Stomach	−	+	n.p.	n.p.
eBL	5	M	Ileum	−	+	n.p.	n.p.

^aAbbreviations: F, female; M, male; n.p., not performed; EBER, Epstein-Barr virus-encoded small RNA; B.A., break apart; eBL, endemic Burkitt lymphoma; sBL, sporadic Burkitt lymphoma, FISH, fluorescence in situ hybridization.

EBV infection of B cells downregulates KDM2B expression.

To assess the ability of EBV to deregulate KDM2B expression, primary B cells isolated from 3 independent donors were infected with EBV. Cells were collected at different times after infection and analyzed for the expression level of KDM2B mRNA by reverse transcription-quantitative PCR (RT-qPCR). It appeared that soon after infection, the expression level of KDM2B was drastically reduced (Fig. 2A). This downregulation was also seen at the protein level (Fig. 2B). The reduced expression of KDM2B was maintained during the immortalization process and stayed low in lymphoblastoid cell lines (LCL). This result suggests that EBV infection plays a role in the regulation of KDM2B mRNA expression. To exclude the possibility that this result could be due to the activation of the primary B cells independently of EBV infection, as a consequence of the engagement of the membrane B-cell receptors by the virus, we stimulated primary B cells from 2 independent healthy donors with CD40 ligand (CD40L) and interleukin 4 (IL-4) with or without EBV. Analysis of the EBNA1 expression by RT-qPCR showed that the infection worked efficiently (Fig. 2C). Both treatment with CD40L and IL-4 and infection with EBV similarly activated the B cells, as shown by the strong induction of expression of CCL22 (Fig. 2D), a cytokine that is known to be produced during primary B-cell activation and EBV infection (11). Importantly, significant downregulation of KDM2B mRNA expression was observed only when the cells were infected with EBV (Fig. 2E). Moreover, infecting the primary B cells with UV-inactivated EBV led to reduced downregulation of KDM2B compared with that in the cells infected with the untreated virus (Fig. 2F), further indicating that this event is independent of B-cell activation and requires active expression of the viral genes. To assess whether downregulation of KDM2B depends on differences between the proliferation status of the cells, we measured KDM2B expression in Akata2000 cells, an EBV(+) BL-derived cell line, and in Akata31 cells, which were derived by expansion of a clone of Akata2000 cells that had lost the virus. Analysis of the viral DNA in both cell lines confirmed that Akata31 cells had few or no EBV genomes compared with Akata2000 cells (Fig. 2G). Interestingly, Akata2000 cells displayed very low levels of KDM2B mRNA compared with Akata31 cells, indicating that KDM2B expression varies according to the amount of virus (Fig. 2G) and independently of the cell type. Similarly, reactivation of EBV by exposing Raji cells to 12-O-tetradecanoylphorbol-13-acetate (TPA) and sodium butyrate (NaB) treatment led to reduced expression of KDM2B (Fig. 2H). Finally, at 48 h after EBV infection, KDM2B mRNA and protein expression levels were reduced in RPMI-8226 (RPMI) cells (Fig. 2I), confirming the results obtained in EBV-infected primary B cells and further proving that downregulation of KDM2B expression is a direct effect of the virus and not a side effect of the activation of the cells or of the immortalization process.

FIG 2.

EBV-dependent silencing of KDM2B expression in vitro. (A) Primary B cells from 3 independent donors were infected with EBV and collected to make dry pellets at the indicated time points. Some of the infected cells were left in culture for 4 weeks (4w) to generate LCL. Cell pellets were processed and analyzed for the expression level of KDM2B mRNA by RT-qPCR. The difference between the levels of KDM2B in primary B cells (time point 0) and in EBV-infected cells was significant (*, P < 0.05; **, P < 0.01). The levels of KDM2B were measured relative to its levels in LCL (which was used as the calibrator, for which the value was 1). (B) KDM2B protein expression levels in primary B cells mock infected (MI) or infected with EBV for 48 h were analyzed by Western blotting (left). The KDM2B protein signal was normalized to the levels of GAPDH. The histogram (right) shows the average from 3 independent experiments. (C to E) Primary B cells from 2 independent healthy donors were activated by treating them for 24 h with CD40L and IL-4, as described in Materials and Methods, and/or infected with EBV for 48 h. Cells were then processed and analyzed for the expression levels of EBNA1 (C), CCL22 (D), and KDM2B (E) mRNA by RT-qPCR. The results shown in the histogram are the average from 2 independent experiments (*, P < 0.05; ns, not significant). EBNA1 mRNA levels were measured relative to its levels in EBV-infected untreated cells. CCL22 and KDM2B mRNA levels were measured relative to their levels in mock-infected untreated cells. (F) Primary B cells were infected with 2 aliquots of the same EBV batch, one of which was UV inactivated before infection. At 48 h after infection, cells were collected and EBNA1 and KDM2B expression levels were assessed by RT-PCR and qPCR, respectively. The levels of KDM2B mRNA were measured relative to its levels in mock-infected cells. (G) Akata2000 and Akata31 cells were collected, processed for DNA/RNA extraction, and analyzed for the presence of the EBV genome by PCR (left) as well as for KDM2B expression levels by RT-qPCR (lower right) or RT-PCR (upper right). Viral DNA (left) indicates the relative amount of the BFRF3 gene present in Akata2000 versus Akata31 cells that was analyzed by real-time PCR and normalized to the amount of GAPDH in the extracted DNA. The levels of KDM2B mRNA were measured relative to its levels in mock-infected cells (*, P < 0.05). (H) Raji cells untreated or treated with TPA (50 ng/ml)-NaB (3 mM) for 48 h were analyzed for expression levels of BZLF1 and KDM2B mRNA by RT-qPCR (*, P < 0.05). (I) RPMI cells untreated or infected with EBV were collected, processed for RNA and protein extraction, and analyzed for the levels of the KDM2B protein (Western blotting, top) and mRNA (RT-qPCR, bottom). The levels of KDM2B mRNA were measured relative to its levels in mock-infected cells.

To determine whether the reduced expression of KDM2B in EBV-immortalized cells could be due to an increase in DNA methylation, LCL were treated with the demethylating agent Aza and analyzed by RT-qPCR for the KDM2B expression level. Aza-treated LCL showed increased expression of KDM2B mRNA compared with their untreated counterparts, indicating that downregulation of KDM2B in EBV-infected cells is mediated by DNA methylation (Fig. 3A). Because incorporation of Aza into DNA impedes its methylation by DNA methyltransferases (DNMTs) (12), we hypothesized that EBV could contribute to KDM2B silencing by increasing the recruitment of DNMT1 to the KDM2B gene. Indeed, depletion of DNMT1 in LCL by transfection of DNMT1 small interfering RNA (siRNA) led to an increased KDM2B mRNA level (Fig. 3B). Moreover, chromatin immunoprecipitation (ChIP) experiments showed increased recruitment of DNMT1 to the CpG site CpG21423404 of the KDM2B gene in EBV-infected RPMI cells (Fig. 3C).

FIG 3.

Methylation-dependent silencing of KDM2B expression. (A) LCL were cultured for 96 h in the presence of DMSO (nontreated [NT]) or 5-aza-2′-deoxycytidine (Aza). Cells were then collected, and the KDM2B expression levels were analyzed by RT-qPCR. The histogram shows the average KDM2B mRNA levels measured in 3 independent experiments (*, P < 0.05). KDM2B mRNA levels in Aza-treated LCL were measured relative to its levels in DMSO-treated cells. (B) LCL were transfected with stabilized siRNA targeting DNMT1 (siDNMT1) in two independent experiments. At 4 days after transfection, cells were collected and analyzed for the protein levels of DNMT1 (top) and the mRNA levels of KDM2B (bottom). KDM2B mRNA levels in DNMT1-targeted siRNA-treated cells were measured relative to its levels in cells treated with scrambled (scr) siRNA. (C) RPMI cells infected with EBV or mock infected were fixed and processed for ChIP with a DNMT1 antibody or an IgG antibody as a negative control. The eluted DNA was analyzed by qPCR with primers flanking CpG15695155, CpG21423404, and CpG island 127 (primer sequences are described in Table 3) (*, P < 0.05; ns, not significant).

Taken together, these results indicate that infection of B cells with EBV leads to reduced expression of KDM2B, mediated by the recruitment of DNMT1 to its gene promoter.

The oncogenic viral protein LMP1 induces the silencing of KDM2B.

Next, we assessed whether LMP1, the main EBV oncoprotein, plays a role in the deregulation of KDM2B expression. To do this, we generated RPMI cells stably expressing LMP1 (RPMI-LMP1 cells). As a negative control, the cells were transduced with the empty retroviral vector (pLNSX) (Fig. 4A). As revealed by RT-qPCR analysis, the expression level of KDM2B was reduced in the RPMI cells expressing LMP1 (Fig. 4A). Thus, LMP1 appears to play a role in the EBV-mediated downregulation of KDM2B expression. In addition, treating RPMI-LMP1 cells with Aza led to a rescue of KDM2B mRNA levels (Fig. 4B). In contrast, no change in KDM2B mRNA expression was observed in the control cells after treatment with Aza (Fig. 4B).

FIG 4.

LMP1 mediates downregulation of KDM2B. (A) RPMI cells were stably transduced with pLXSN or with pLXSN-LMP1 (LMP1) in four independent transduction experiments. Cells were then collected, and the expression levels of LMP1 (left) and KDM2B (right) were analyzed by RT-qPCR (****, P < 0.0001). (B) RPMI-pLXSN or RPMI-LMP1 cells were cultured in the presence of Aza (+) or DMSO (−) for 48 h, and the KDM2B mRNA expression level was analyzed by RT-qPCR. The histogram shows the average from 2 independent experiments (*, P < 0.05; ns, not significant). (C and D) Louckes cells were stably transduced with pLXSN, pLXSN-LMP1 (LMP1), or pLXSN-LMP1 mutants (3AAA, 378, and 3AAA/378). Cells were collected and processed for RNA and protein analysis. mRNA expression and protein levels of LMP1 were detected by RT-PCR (C, top) and immunoblotting (C, bottom). KDM2B mRNA and protein levels were also shown by RT-qPCR and immunoblotting, respectively (D, left and right). In panel D, the difference between KDM2B mRNA and protein levels in Louckes cells with pLXSN and in Louckes cells stably expressing wild-type LMP1 was significant (*, P < 0.05; **, P < 0.01). The levels of KDM2B mRNA in cells expressing wild-type (WT) LMP1 and LMP1 mutants were measured relative to its level in cells expressing pLXSN. (E) RPMI-LMP1 cells were treated for 2 h with BAY11-7082 (Bay11) (10 μM) or with JNK inhibitor II (JNKII) (10 μM) in three independent experiments. KDM2B mRNA levels were analyzed by RT-qPCR (*, P < 0.05). (F) RPMI pLXSN and RPMI-LMP1 cells, the latter of which were untreated or treated with BAY11-7082 (10 μM) for 2 h, were fixed to perform ChIP with DNMT1 or IgG antibodies. The eluted DNA was analyzed by qPCR with primers designed to surround CpG21423404. The histogram shows the average percentage of recruitment of DNMT1 in 4 independent experiments (*, P < 0.05).

LMP1 activates both the NF-κB and Jun N-terminal protein kinase (JNK) pathways through its CTAR1 and CTAR2 domains, respectively. Therefore, to gain insights into the mechanism by which LMP1 deregulates KDM2B expression, we generated RPMI and Louckes cells expressing LMP1 mutants harboring mutations in CTAR1 (3AAA mutant), CTAR2 (378 mutant), or both (3AAA/378 double mutant) and therefore having a hampered ability to activate the NF-κB pathway, the JNK pathway, or both (13). After selection, cells were analyzed for LMP1 expression by RT-qPCR and immunoblotting. All the cells generated expressed similar levels of the LMP1 transcript and protein (Fig. 4C), with the exception of the LMP1 3AAA mutant, which appeared to be present at lower protein levels than the wild-type LMP1 and the other LMP1 mutated molecules. This is probably due to a reduced protein stability of the LMP1 3AAA mutant caused by the mutation in the CTAR1 in the context of a wild-type CTAR2 domain; however, the 3AAA/378 double mutant and the 378 mutant showed mRNA and protein levels similar to those of wild-type LMP1. We then compared the ability of the different LMP1 mutants to deregulate KDM2B expression. Whereas wild-type LMP1 efficiently downregulated KDM2B mRNA and protein expression (Fig. 4D), both the 3AAA mutant and the double mutant were unable to downregulate KDM2B mRNA expression, and the 378 mutant partially maintained the ability to inhibit KDM2B expression (Fig. 4D), indicating that the CTAR1 domain and, in part, the CTAR2 domain play a role in the LMP1-mediated regulation process of KDM2B expression. The same results were observed in RPMI cells stably expressing LMP1 and its CTAR mutants (data not shown). To further evaluate the impact of the NF-κB and JNK pathways on LMP1-mediated KDM2B deregulation, we treated RPMI-LMP1 cells with specific chemical inhibitors: BAY11-7082 and JNK inhibitor II, respectively. Whereas the JNK inhibitor had no effect, treating RPMI-LMP1 cells with the NF-κB inhibitor BAY11-7082 led to a rescue of KDM2B mRNA expression (Fig. 4E).

Previous studies showed that LMP1 activates and induces the recruitment of DNMT proteins to the promoters of cancer-related genes (14–16). In line with these findings, our ChIP experiments showed an increase in the recruitment of DNMT1 to the KDM2B gene at the CpG21423404 position in LMP1-expressing cells compared with control cells (Fig. 4F). Treating RPMI-LMP1 cells with BAY11-7082 significantly reduced the amount of DNMT1 recruited to the KDM2B gene. Taken together, these data indicate that EBV induces silencing of KDM2B expression mainly via the ability of its main transforming protein, LMP1, to activate the NF-κB signaling pathway.

LMP1 is not the only EBV protein able to induce downregulation of KDM2B expression.

LMP1 is only rarely expressed in EBV(+) BL samples. Therefore, our observation that KDM2B is downregulated and silenced in EBV(+) BL cell lines and specimens led us to investigate whether EBV genes other than LMP1 could play a role in this event. Infection of primary B cells with a recombinant EBV lacking the LMP1 gene (EBVΔLMP1) still led to decreased expression of KDM2B mRNA (Fig. 5A). Downregulation of KDM2B occurred at 12 h after infection of primary B cells with EBV; as expected at this early time point, we could not yet detect LMP1 expression, and only the EBNAs were efficiently expressed (Fig. 5B). Moreover, when comparing KDM2B mRNA expression levels in different EBV(+) BL-derived cell lines, in an LCL, and in primary B cells, we observed a reduced level of KDM2B transcript in all the BL cells, independently of whether they were in latency phase I or III (Fig. 5C). As expected, although BL cells in phase I expressed EBNA1 but had little or no LMP1 expression, BL cells in phase III and the LCL efficiently expressed both genes (Fig. 5C). Taken together, these experiments indicate that KDM2B deregulation can also occur in the absence of LMP1 expression; therefore, other EBV proteins could be involved in this event in BL cells. Transient transfection of Louckes and RPMI cells with different constructs expressing a panel of EBV genes (Fig. 5D and E) indicated that, in addition to LMP1, different latent viral proteins could downregulate KDM2B mRNA. To rule out the possibility that this result was the consequence of the activation of the NF-κB pathway as a side effect of the transfection of the cells, we checked the levels of IκBα phosphorylation. As expected, based on its known function (17), among the analyzed EBV proteins, LMP1 was the most efficient in inducing NF-κB activation (Fig. 5F), indicating that the other viral proteins may use other mechanisms to downregulate KDM2B mRNA expression. MicroRNA-146-5p (miRNA-146-5p), which is known to target KDM2B expression, has been reported to deregulate KDM2B mRNA levels during human papillomavirus (HPV) infection (18). A previous study showed that EBNA2 induces the expression of miRNA-146-5p (19). Therefore, we tested whether miRNA-146-5p could contribute to KDM2B downregulation during EBV infection. To do this, we treated primary B cells with a miRNA-146-p5-specific inhibitor before EBV infection. As shown in Fig. 5G, this treatment partially rescued KDM2B expression in the miRNA-146-p5 inhibitor-treated cells. These results show that EBV may be able to reduce intracellular levels of KDM2B by a redundancy of mechanisms, further indicating that this event could be important for the virus.

FIG 5.

EBV also uses LMP1-independent mechanisms to downregulate KDM2B. (A) B cells from two donors were infected with EBV or EBVΔLMP1 or mock infected (MI) and collected at 48 h after infection. Reverse-transcribed RNA samples were analyzed by qPCR for the KDM2B expression level. (B) B cells from two donors were infected with EBV and collected at 12, 24, and 48 h after infection. Cells were processed for RNA extraction and analyzed by qPCR for the expression levels of EBNA1, EBNA2, EBER, LMP1, and KDM2B transcripts. The expression levels of viral gene transcripts were measured relative to their levels in B cells collected 12 h (EBNA2) or 48 h (EBNA1, EBER, LMP1) postinfection. The levels of KDM2B mRNA were measured relative to its levels in mock-infected cells. (C) Different EBV(+) cell lines in latency phase I (BL110, I100, MUTU) or in latency phase III (LCL, Raji, B95) and primary B cells (BC) were collected, processed for RNA extraction and reverse transcription, and analyzed by qPCR for LMP1 (top), EBNA1 (middle), and KDM2B (bottom) mRNA levels. (D and E) RPMI (D) and Louckes (E) cells were transiently transfected with different constructs carrying individual EBV genes (TC) in three independent experiments. At 48 h after transfection, cells were processed for RNA and protein extraction and analyzed for the KDM2B expression level by RT-qPCR analysis. The levels of KDM2B mRNA were measured relative to its levels in mock-infected cells. Western blotting (LMP1, EBNA1, EBNA3A/3B, and EBNALP) or RT-PCR analysis (LMP2A, EBNA2, and EBNA3C) was performed to measure the expression of the different viral latent proteins. (F) RPMI cells were transfected with different EBV gene-carrying constructs. At 44 h after transfection, cells were exposed to the proteasome inhibitor MG132 for 4 h and then collected, processed for total protein extraction, and analyzed for the indicated proteins by immunoblotting. The histogram shows the average phosphorylated IκBα (P-IκBα) signal normalized to the levels of total IκBα. (G) B cells from different donors were untreated or treated with a miRNA-146a-5p inhibitor for 24 h before infection with EBV. At 48 h after infection, cells were collected and processed for RNA extraction and reverse transcription. cDNA samples were analyzed by qPCR for LMP1 and KDM2B expression levels. The levels of KDM2B mRNA were measured relative to its levels in mock-infected cells. The histograms show the average expression levels measured in 2 independent experiments (*, P < 0.05). T0, time zero.

KDM2B regulates expression of viral genes in EBV-infected B cells.

To evaluate the biological relevance of EBV-mediated KDM2B downregulation, we hypothesized that the epigenetic enzyme could regulate EBV transcription, similar to what was observed for other histone modifiers and chromatin-interacting proteins (e.g., EZH2, CTCF, KMT5B) (20–22). To assess whether ectopic expression of KDM2B in B cells could have an impact on EBV infection, we transfected increasing concentrations of a KDM2B construct into Louckes cells, an EBV(−) BL-derived cell line. One day after transfection, cells were infected with EBV-green fluorescent protein (GFP) and monitored for the infection efficiency 24 h later. Figure 6A shows efficient overexpression of ectopic KDM2B at both the protein and mRNA levels at 24 h after transfection. The infection efficiency of Louckes cells transfected with the KDM2B expression vector or the control empty vector (pCDNA) was indistinguishable, as revealed by TaqMan PCR, showing that the increased KDM2B expression level did not alter the viral genome copy number per cell (Fig. 6B). In contrast, the GFP mean fluorescence intensity (MFI) decreased in the presence of enhanced KDM2B expression (Fig. 6C), indicating that KDM2B could affect viral gene expression. Indeed, RT-qPCR analysis of the expression levels of different EBV transcripts (LMP1, EBNA1, and BZLF1) at 24 h after infection showed a significant and dose-dependent reduction in their mRNA levels in the presence of an increasing amount of KDM2B (Fig. 6D). Taken together, these data suggest that KDM2B plays a role in regulating EBV gene expression.

FIG 6.

KDM2B deregulation alters EBV gene expression. (A) Louckes cells were transiently transfected with increasing concentrations of the KDM2B expression vector in three independent experiments, collected at 24 h after transfection, and processed for protein and RNA extraction to assess KDM2B levels by immunoblotting (top) or qPCR (bottom). (B) Cells were then infected with EBV, and at 24 h after infection they were collected and processed for FACS analysis and for RNA/DNA extraction. DNA samples were used to measure EBV genome copy number by TaqMan PCR (ns, not significant). (C) Live cells were analyzed by FACS to measure the GFP mean fluorescence intensity (MFI). (D) cDNA samples were analyzed for the expression level of EBV early and late genes by qPCR. The levels of EBV genes transcripts were normalized to the mRNA levels of the housekeeping gene β-globin and calculated relative to their levels in cells transfected with the empty vector (KDM2B, 0 μg). The values shown in the histogram are the average from 3 independent experiments (*, P < 0.05). (E to G) Louckes cells were transfected with KDM2B siRNA and scrambled siRNA as a control in three independent experiments. (E) At 24 h after transfection, half of the cells were collected and analyzed for the expression levels of KDM2B protein (top) and mRNA (bottom), and half of the cells were infected with EBV. At 24 h after infection, cells were collected and processed for RNA/DNA extraction. The number of EBV genome copies per cell was determined by TaqMan PCR on the DNA template (MI, mock infected; Louckes cell DNA was also included as a negative control) (F), and qPCR of the cDNA samples enabled the assessment of the mRNA expression levels of different viral genes, calculated as explained in the legend to panel D (G). siSCR, scrambled siRNA; siKDM2B, siRNA targeting KDM2B mRNA.

The results presented above, obtained by overexpressing KDM2B, could be due to a generalized chromatin demethylation as a consequence of high intracellular levels of the demethylase. Therefore, next we depleted endogenous KDM2B in Louckes cells by transfecting an siRNA targeting KDM2B mRNA. RT-qPCR and Western blot analysis showed that KDM2B was efficiently downregulated 24 h after transfection with the specific siRNA (Fig. 6E). Cells were then infected with EBV-GFP and monitored for the infection efficiency, as described above. At 24 h after infection, neither the percentage of GFP-positive cells nor the genome copy number had changed significantly between the control cells (cells transfected with scrambled siRNA [siSCR]) and the cells transfected with the siRNA directed against KDM2B (Fig. 6F and data not shown), indicating that the loss of KDM2B does not affect the efficiency of infection. However, efficient depletion of KDM2B by siRNA led to a significant increase in EBV transcripts 24 h after infection (Fig. 6G). This result indicates that a reduced intracellular level of KDM2B promotes the expression of the viral genes, further confirming the ability of KDM2B to repress EBV gene expression at early stages of infection.

KDM2B inhibits viral gene expression in latently infected EBV-immortalized B cells.

To test whether the activity of KDM2B in regulating EBV gene expression is required for the maintenance of virus latency, similar to what has been reported for other epigenetic enzymes (20), we aimed to overexpress KDM2B ectopically and examine its impact on viral gene expression in EBV-immortalized B cells. LCL displayed detectable KDM2B mRNA and protein levels, although they were lower than those in primary B cells (Fig. 2A and B and data not shown). Therefore, KDM2B was overexpressed in LCL by ectopic expression of a KDM2B construct. At 48 h after transfection, LCL were collected and processed for total protein and DNA/RNA extraction. Western blotting and RT-qPCR showed substantial increases of KDM2B at the protein and mRNA levels, respectively (Fig. 7A). LCL overexpressing KDM2B carried a similar number of EBV genome copies compared with the same cells transfected with the control pCDNA vector (Fig. 7B), as revealed by TaqMan PCR analysis. We then analyzed the mRNA expression level of different EBV transcripts by RT-qPCR. Ectopic expression of KDM2B led to a significant reduction in the mRNA level of all the analyzed EBV genes (Fig. 7C). In contrast, depletion of KDM2B from LCL by transfecting KDM2B siRNA (Fig. 7D) led to a significant increase in the expression of the viral genes compared with the expression level of the same genes in LCL transfected with scrambled siRNA (Fig. 7F). Similar to what we observed in EBV-infected Louckes cells, removal of KDM2B did not significantly affect the EBV genome copy number (Fig. 7E). Taken together, these data indicate that KDM2B controls viral gene expression in latently infected cells.

FIG 7.

KDM2B regulates EBV gene expression in latently infected cells. (A to C) LCL were transfected with 1.5 μg of pCDNA3-KDM2B or with pCDNA as a control in three independent experiments. Cells were collected at 24 h after transfection and processed for RNA/DNA and total protein extraction. (A) KDM2B mRNA and protein levels were shown by qPCR (bottom) and immunoblotting (top; the lines corresponding to conditions with 0.5 μg and 1 μg of KDM2B, originally present in the Western blot, are not shown because they were excluded from all the analyses). (B) DNA samples were analyzed by TaqMan PCR to assess the number of EBV genome copies per cell (ns, not significant). (C) The mRNA expression levels of EBV latent and early genes were assessed by qPCR (*, P < 0.05). The mRNA levels of viral genes in LCL overexpressing KDM2B were measured relative to their levels in pCDNA-transfected control cells and normalized on the mRNA levels of β-globin. (D to F) (D) LCL transfected in three independent experiments with KDM2B siRNA (siKDM2B) or scrambled siRNA (siSCR) as a control were collected at 48 h after transfection and analyzed for the levels of KDM2B protein (top) and transcript (bottom). (E) DNA samples were analyzed by TaqMan PCR to assess the number of EBV genome copies per cell. (F) The mRNA expression levels of EBV latent and early genes were assessed by qPCR (*, P < 0.05). Viral transcript levels in KDM2B siRNA-treated cells were measured relative to their levels in siSCR-treated cells and normalized to the mRNA levels of β-globin.

Next, we determined whether KDM2B could directly bind to EBV gene promoters to regulate their expression. To do this, purified primary B cells infected or mock infected with EBV for 48 h, as well as the corresponding LCL, were formaldehyde fixed and processed for KDM2B ChIP (Fig. 8A). ChIP analysis showed that KDM2B can be recruited to the EBV genome and, more precisely, to the Qp promoter at 48 h after infection. We also found KDM2B recruited to the Cp promoter and to a region of the BZLF1 promoter proximal to the start site (Zp 0) in EBV-immortalized B cells. However, we did not detect KDM2B recruitment when we analyzed a region of the BZLF1 promoter 600 bp upstream of the start site (Zp +600) in the same cells (Fig. 8A, top), indicating that KDM2B is specifically recruited to a specific region of the viral genome. As expected, we did not detect KDM2B recruitment to the promoter of the housekeeping gene GAPDH (glyceraldehyde-3-phosphate dehydrogenase); however, we observed efficient KDM2B binding to a sequence located downstream of the transcription start site of the ribosomal DNA (rDNA) repeated units 8 (H8) (Fig. 8A, bottom), which was previously reported to be targeted by the epigenetic enzyme (23). These data indicate that KDM2B is directly recruited to specific EBV promoter regions. To further assess the ability of KDM2B to be recruited to the EBV genome, B cells were untreated or EBV infected, collected, and processed for immunofluorescent in situ hybridization (immuno-FISH) experiments. EBV DNA molecules were detected by FISH with a labeled probe directed against the BamHI W EBV genomic repeated region; KDM2B was concomitantly detected by immunofluorescence by using an anti-KDM2B antibody. In EBV-infected B cells, KDM2B patches partially overlapped or were found in close proximity to the viral DNA (on average, 40% of the green dots colocalized with the red dots) (Fig. 8B), suggesting that the epigenetic enzyme is recruited to or is close to the viral genome at early stages of infection. Taken together, our data demonstrate a role of KDM2B in controlling EBV gene expression.

FIG 8.

KDM2B is recruited to the EBV genome. (A) Primary B cells from three donors were infected with EBV for 48 h or left immortalized until they generated LCL. Mock-infected (MI) B cells, B cells infected with EBV for 48 h, and LCL were formaldehyde fixed and processed for ChIP using a KDM2B antibody and an IgG antibody. The eluted DNA was analyzed by qPCR with primers designed on different EBV gene promoters. As negative controls, the GAPDH promoter was also amplified, and the recruitment of KDM2B to its known cellular gene target, H8, was also assessed. Qprom, Cprom, and Zprom, Qp, Cp, and Zp EBV promoters, respectively. (B) Louckes cells untreated or infected with EBV (the B95-8 strain) were fixed on glass slides and processed for immuno-FISH. EBV DNA was detected by FISH, and KDM2B was detected by immunofluorescence. Overlapping EBV and KDM2B signals are shown in the merge fields as yellow dots. The histogram shows the percentage of merged dots, estimated by counting the number of EBV signals that colocalized with the KDM2B patches in at least 3 different fields from 3 independent stainings. DAPI, 4′,6-diamidino-2-phenylindole.

Deregulation of KDM2B mRNA level in B cells alters their expression profile.

KDM2B has been shown to play a role in cell differentiation, cell growth, and the proliferation of hematopoietic cells (9). We therefore assessed whether altered KDM2B expression could also have an impact on the B-cell phenotype in our experimental models. Louckes cells and LCL overexpressing KDM2B did not show an altered proliferation ability, nor did they show an altered cell cycle or apoptotic profiles (Fig. 9A to D). Therefore, to gain further insights into the impact of KDM2B deregulation on B cells, we performed an analysis with an RNA expression chip array in Louckes cells expressing increasing levels of KDM2B (Fig. 9E). Cellular genes whose expression was significantly altered by the enhanced levels of KDM2B were identified by bioinformatics analysis (Fig. 9E). In line with the known function of KDM2B as a demethylase of H3K4me3, the set of genes altered by KDM2B was enriched in H3K4me1 (Fig. 9F). Among the genes that were significantly deregulated in the presence of altered KDM2B expression, some play a role in immunity (Fig. 9G). Our analysis revealed that KDM2B overexpression was associated with deregulated expression of genes involved in the tumor necrosis factor receptor 2 (TNFR2) pathway (Fig. 9G). Interestingly, this pathway is important for the transition of B cells from germinal center B cells to resting memory B cells (24). Moreover, the TNFR2 pathway mediates specific tumor necrosis factor (TNF) effects and is an important mediator of the cell antiviral response (9). Deregulated expression of genes in this pathway was confirmed and validated by RT-qPCR (Fig. 9H) and by Western blot analysis for two of them (MLKL and PRDX2), which were, respectively, induced and downregulated during KDM2B overexpression in Louckes cells (Fig. 9I). Taken together, these data indicate that KDM2B plays a role in B-cell differentiation and in regulation of the TNF pathway, which is often altered during the lymphomagenic process.

FIG 9.

KDM2B deregulation alters cellular gene transcription. (A to D) Louckes cells (A and B) and LCL (C and D) were transfected with increasing concentrations of the KDM2B expression vector. (A and C) At 24 and 48 h after transfection, viable cells were counted. (B and D) Transfected cells were ethanol fixed and processed for cell cycle analysis by flow cytometry. (E) Louckes cells, generated as described in the legend to Fig. 5A, were collected and processed for RNA extraction and RNA expression profiling, as described in Materials and Methods. The differential expression analysis was conducted using BRB-ArrayTools software. (F and G) Genes differentially regulated in Louckes cells overexpressing KDM2B compared with their expression in Louckes cells transfected with an empty pCDNA vector were analyzed for their enrichment in specific pathways by using the Enrichr web tool. Enrichment results for the Epigenomics Roadmap HM ChIP sequencing and BioCarta 2016 databases (top and bottom, respectively) are shown. hESC, human embryonic stem cells; fMLP, N-formyl-methionyl-leucyl-phenylalanine; mhc, major histocompatibility complex; TCR, T-cell receptor; MEF2D, myocyte enhancer factor 2D. (H and I) Differential expression of a set of genes found to be deregulated in the expression profile array was validated by qPCR (H) and immunoblotting (I). The levels of the different transcripts were measured relative to their levels in Louckes cells transfected with 0.5 μg of the KDM2B construct.

DISCUSSION

In this study, we show that a regulatory region within the gene encoding the KDM2B protein is more methylated in EBV(+) BL than in EBV(−) BL, confirming data from our previous study, which aimed to characterize the whole epigenetic profile of a set of BL-derived cell lines of eBL or sBL origins. This region is found to be methylated in certain cancer-derived cell lines (ENCODE DNA methylation tracks, CpG methylation by Methyl 450K bead arrays from ENCODE/HAIB, UCSC Genome Browser). Furthermore, KDM2B levels appear to be deregulated in cancers of hematological origin (9). We therefore assessed whether increased methylation of the KDM2B gene in eBL is mediated by EBV to deregulate intracellular levels of the histone modifier and promote EBV-mediated lymphomagenesis. Indeed, KDMs have been shown to have altered expression in cancer (25). For instance, a KDM2B paralogue, KDM2A, behaves as a tumor suppressor in hematopoietic stem cells, in which it antagonizes mixed-lineage leukemia-associated leukemogenesis by erasing H3K36me2 markers (26). Similarly, altered levels of KDM2B could modify the chromatin structure and the pattern of expression of B cells and favor their transformation.

eBL specimens and derived cell lines showed low KDM2B protein expression, underscoring a potentially important role of KDM2B downregulation in the lymphomagenic process in vivo. Treating eBL cell lines with a chemical that blocks DNA methylation led to a rescue of the KDM2B mRNA level, indicating that DNA methylation contributes to the silencing of KDM2B expression in these cells. However, the same treatment left the level of the KDM2B transcript in EBV(−) sBL-derived cell lines unchanged. These events are similar to what we have previously described for another gene, ID3 (4). ID3, which plays a key role in lymphomagenesis and which is often found mutated in the sBL variant (27, 28), was found to be silenced by hypermethylation in eBL-derived cell lines (4). EBV infection therefore plays a direct role in eBL pathogenesis by altering the cellular epigenome and deregulating the expression of genes with a key role in lymphomagenesis. Indeed, in vitro infection of primary B cells with EBV led to rapid downregulation of KDM2B expression. Low KDM2B mRNA levels at early stages of EBV infection of primary B cells were also observed in a data set from RNA expression profiling performed in an independent study (E. Manet, unpublished data). Our data also show that downregulation of KDM2B by EBV is not due to a specific B-cell activation status but is, rather, a direct result of infection by an actively transcribing virus. We also show that downregulation of KDM2B expression in EBV-infected cells is mediated by DNMT1 recruitment to its gene and by its DNA methylation. It has already been reported that EBV can alter DNMT1 activity through its viral proteins LMP1 and LMP2A (29). In particular, Tsai and colleagues showed that LMP1 activates DNMT1 activity (15) and that this event requires activation of the JNK/AP1 pathway.

Here, we show that cells stably expressing LMP1 display lower levels of KDM2B. The ability of LMP1 to downregulate KDM2B depends mostly on its NF-κB activity, as shown by using an LMP1 molecule mutated on its NF-κB-activating domain and by blocking the NF-κB pathway via a specific chemical inhibitor. A LMP1 mutant lacking the JNK activation pathway also downregulates KDM2B, but to a lesser extent than wild-type LMP1. This indicates a partial contribution of that pathway to LMP1-mediated KDM2B downregulation. However, exposing the cells to a specific JNK inhibitor had no effect on the KDM2B mRNA expression level. Moreover, our ChIP experiments showed that expression of LMP1 in the EBV(−) RPMI cells is able to trigger DNMT1 recruitment to the KDM2B gene. The finding that the recruitment of DNMT1 to the KDM2B gene can be hampered by treating LMP1-RPMI cells with the IκBα kinase inhibitor further confirms its dependence on the ability of LMP1 to induce the NF-κB pathway. Recently published data showed that HPV16 E6/E7 transforming proteins inhibit the expression of miRNA-146-5p, known to target the KDM2B transcript, which results in an increase in the KDM2B expression level in HPV16-infected cells. In contrast to E6/E7, the EBV transforming protein LMP1 induces miRNA-146-5p via its ability to activate NF-κB (18). EBNA2 also induces miRNA-146-5p (19). This event could contribute to the reduction in the KDM2B mRNA level in EBV-infected cells. Our data show that blocking miRNA-146-5p in primary B cells before infection leads to a partial rescue of KDM2B levels compared with those in untreated cells. It is therefore possible that the virus uses alternative mechanisms to target KDM2B: (i) increasing its DNMT1-mediated gene methylation associated with LMP1 protein expression and (ii) controlling KDM2B mRNA levels by a specific miRNA associated with the expression of LMP1 and/or EBNA2. Further studies are needed to assess the contributions of deregulated miRNA-146-5p levels to the events described here. Our data also indicate that proteins other than LMP1 and EBNA2 appear to be able to mediate downregulation of KDM2B. EBV may need to tightly control KDM2B levels and activity to regulate expression of its own genes as well as that of cellular genes. Indeed, a recent study from Gillman and colleagues showed that EBNA3C interacts with KDM2B by its TFGC motif (HD motif) (30). Mutation of the latter motif impairs the ability of EBNA3C to bind to KDM2B and repress its target genes (30).

Previous studies by us and others showed that EBV can modulate the level and the activity of different epigenetic enzymes (20, 21, 31), which in turn play a role in regulating viral gene expression. One example is EZH2, whose intracellular expression level is induced in B cells by EBV infection in an LMP1-dependent manner (32). EZH2, in turn, is recruited to the viral genome, where it participates in the establishment and maintenance of EBV latency by methylating H3K27 in proximity to the BZLF1 and BRFL1 promoters (22). Our observation that EBV infection alters KDM2B expression prompted us to assess whether the epigenetic enzyme could regulate EBV infection and/or the EBV life cycle. Altering KDM2B expression in B cells before EBV infection or in latently EBV-infected B cells led to the deregulated expression of all analyzed viral genes. This was consistent with the recruitment of KDM2B to their respective promoters, as observed by ChIP experiments. In line with its ability to demethylate the active chromatin marker and repress transcription, KDM2B depletion in EBV-infected B cells led to increased viral gene expression, whereas its forced ectopic expression had the opposite effect and caused a strong reduction in the levels of viral transcripts. Taken together, our data indicate that KDM2B plays a role in controlling EBV gene expression, for instance, during the establishment of latency. Recruitment of KDM2B to the viral episome during the first step of infection could be necessary for the repression of viral gene transcription, especially at the end of the pre-latent phase, when the EBV lytic genes are silenced to allow the virus to persist in resting peripheral B cells (32). In contrast, KDM2B could work as a host restriction factor; therefore, its downregulation during the early stages of EBV infection would allow efficient viral gene expression and replication during the pre-latent phase. Future studies will be needed to investigate the exact role of EBV-mediated KDM2B deregulation in EBV life cycle control.

It is known that in order to escape the immune surveillance of the host and establish a chronic infection, EBV has evolved different mechanisms to maintain B cells in a status of long-lived circulating memory B cells and prevent them from differentiating into antibody-secreting plasma cells. A recent study showed that EBNA3A and EBNA3C block the terminal differentiation of memory B cells to plasma cells by epigenetically repressing the gene encoding BLIMP-1, a master regulator in B-cell differentiation (33). Notably, a previous study showed that KDM2B plays a key role in the differentiation of hematological stem cells (9). Therefore, EBV-mediated deregulation of KDM2B expression could contribute to the mechanism that prevents the cells from undergoing terminal differentiation. This hypothesis is supported by our data from an RNA profile analysis conducted on cells transfected with increasing levels of a KDM2B-expressing construct. Cells harboring high KDM2B levels showed a deregulated expression of genes enriched in the TNFR2 pathway, which is known to play a role in the differentiation from B cells to plasma cells (24). TNFR2 also regulates the interferon pathway, an important mediator of the antiviral response. Ablation of KDM2B in hematopoietic cells has previously been shown to downregulate the interferon response (9). The downregulation of the KDM2B expression level upon EBV infection could therefore contribute to the viral escape from immune system surveillance.

Taken together, our data show a novel interplay between EBV infection and the host epigenome. EBV alters KDM2B expression via an epigenetic mechanism involving LMP1 and other latent viral proteins. The histone modifier, in turn, plays a role in regulating the expression of viral and host cell genes. These data, in addition to the observed deregulated KDM2B levels in BL-derived cell lines, indicate that altered expression of this epigenetic enzyme could contribute to B-cell transformation. Future studies aimed at investigating the functional importance of KDM2B gene methylation and downregulated expression during the EBV-mediated lymphomagenic process are warranted.

MATERIALS AND METHODS

Case selection, immunophenotype, and FISH.

We studied 22 morphologically and immunophenotypically typical BL cases (11 sporadic and 11 endemic). All cases were diagnosed according to the updated World Health Organization (WHO) classification of tumors of hematopoietic and lymphoid tissues (34). The cases were retrieved from the archives of Siena University Hospital (Siena, Italy; n = 2) and Nairobi University (Nairobi, Kenya; n = 20). Before enrolling the cases in this study, they were reevaluated by an expert hematopathologist (L.L.), and the diagnosis was confirmed by morphology on histological slides stained with hematoxylin and eosin (H&E) or Giemsa and by immunophenotyping. The main clinical features of our samples are summarized in Table 2. The study was conducted at the University of Siena in Italy according to the principles of the Helsinki declaration after approval of the local review board. All the procedures were carried out automatically on representative paraffin sections from each case by Bench Mark Ultra (Ventana, Monza, Italy) using extended antigen retrieval and with diaminobenzidine (DAB) as the chromogen.

Cell culture and treatment.

Peripheral B cells were purified from blood samples using the RosetteSep human enrichment kit (catalog number 15064; Stem Cell Technologies). LCL were generated in this study by infection of primary B cells from different donors. The myeloma-derived RPMI-8226 cells (http://web.expasy.org/cellosaurus/CVCL_0014) and the BL-derived cell lines, including the BL EBV(−) Louckes cell line (http://web.expasy.org/cellosaurus/CVCL_8259), were obtained from the International Agency for Research on Cancer (IARC) Biobank. The Akata2000 and Akata31 cell lines (35) were also used in the present study. Primary and immortalized B cells were cultured in RPMI 1640 medium (Gibco, Invitrogen Life Technologies, Cergy-Pontoise, France) supplemented with 10% fetal bovine serum, 100 U/ml penicillin G, 100 mg/ml streptomycin, 2 mM l-glutamine, and 1 mM sodium pyruvate (PAA; Pasching, Austria) or in advanced RPMI 1640 medium (catalog number 12633012; Life Technologies). EBV (the B95-8 strain) particles produced by culturing HEK293_EBVgfp cells were used to infect B cells. EBV infection of B cells was performed either using a recombinant EBV-GFP genome or using the EBV mutant strain EBVΔLMP1 lacking the entire LMP1. The percentage of GFP-positive cells was assessed by fluorescence-activated cell sorting (FACS; FACSCanto system; Becton, Dickinson) and spanned from 10% to 15% at 24 to 48 h postinfection in Louckes and RPMI cells and 60% to 80% when measured at 48 h postinfection in primary B cells. Analysis of the cell cycle and apoptosis (sub-G₁ phase) was performed by ethanol fixing the cells and staining their DNA with propidium iodide at a final concentration of 5 μg/ml. Subsequently, cells were analyzed by FACS. Inactivation of EBV was performed with UV light (6 × 10 mJ). EBV reactivation in BL cells was done by treatment of the cells with TPA (50 ng/ml) in association with NaB (3 mM). Anti-miRNA treatment was done by the addition of Hsa-miR-146a-5p miRCURY LNA miRNA power inhibitor (250 nM; Qiagen) directly to the cell culture medium 24 h before infection with EBV. To induce their activation, primary B cells were seeded at a density of 0.5 × 10⁶ cells in 6-well dishes and treated with 100 ng/ml recombinant human CD40 ligand (hCD40L; catalog number 6245-CL; R&D Systems) and 20 ng/ml of recombinant human IL-4 (R&D Systems) for 48 h. Cells were then collected and processed for further analysis. To block DNA methylation, cells were treated with 5-aza-2′-deoxycytidine (≥97%; catalog number A3656; Sigma-Aldrich) at a final concentration of 10 μM for 48 or 96 h. To inhibit the different pathways, cells were treated with the IκBα kinase inhibitor BAY11-7082 (Calbiochem) at a final concentration of 10 μM or with the JNK inhibitor II SP600125 (catalog number 420119; VWR International), used at a final concentration of 10 μM. Cells were preincubated with the different inhibitors for 1.5 h and 2 h.

qPCR.

Total RNA was extracted using an AllPrep DNA/RNA minikit (Qiagen). RNA reverse transcription to cDNA was carried out by the use of RevertAid H Minus Moloney murine leukemia virus reverse transcriptase (Thermo Fisher Scientific), according to the manufacturer’s protocol. Quantitative PCR (qPCR) was performed using a MesaGreen qPCR MasterMix Plus for SYBR assay (Eurogentec). For each primer set, the qPCR was performed in duplicate and the mRNA levels obtained were normalized to the average mRNA levels of three housekeeping genes (β-globin, β-actin, and GAPDH) measured in the same samples or to the mRNA level of β₂-microglobulin only. For each PCR, a sample in which the DNA template was replaced with PCR-grade water was included as a negative control. To measure the EBV genome copy number per cell, total DNA was extracted using an AllPrep DNA/RNA minikit (Qiagen) and measured by use of a NanoDrop spectrophotometer. Similar amounts of DNA were used as a template for TaqMan PCR, performed according to the protocol described by Accardi et al. (36). The PCR primer sequences are indicated in Table 3. All the primers used for the first time in the present study were assessed for their efficiency (90% to 110%).

TABLE 3.

Primers used for qPCR and ChIP-qPCR

Primer use and primer	Sequence
	Forward	Reverse
qPCR
LMP1	CCAGTCCAGTCACTCATAACG	CCTACATAAGCCTCTCACACT
EBNA1	GGTCGTGGACGTGGAGAAAA	GGTGGAGACCCGGATGATG
DNMT1	GAG GAA GCT GCT AAG GAC TAG TTC	ACT CCA CAA TTT GAT CAC TAA ATC
β2 microglobulin	CTCACGTCATCCAGCAGAGA	CGGCAGGCATACTCATCTTT
Beta globin	GCATCTGACTCCTGAGGAGA	AGCACACACACCAGCACATT
CCL22	ACTGCACTCCTGGTTGTCCT	CGGCACAGATCTCCTTATCCC
PRDX2	GTG TCC TTC GCC AGA TCA CT	ACG TTG GGC TTA ATC GTG TC
Actin	CTG GGA GTG GGT GGA GGC	TCA ACT GGT CTC AAG TCA GTG
GAPDH	GCCAAAAGGGTCATCATC	TGCCAGTGAGCTTCCCGTTC
KDM2B	CCC AGC ATC TGA AGG AGA AG	GTT GGA GGA ATC AGC CAA AA
LAT2	ACTCCTCTCTCTCCTGCAGA	CGAGGATAGTAGGGGCAAGG
GNA15	AGAATCGCTTGAACCCAGGA	ATTTCGAACTCCTGGCCTCA
TNFRSF13B	AACTCGGGAAGGTACCAAGG	GAAGACTTGGCCGGACTTTG
TNFRSF18	CTCTTGAAACCCGAGCATGG	ACTCGGAACAGCACTCCTC
LTA	ACTCCTCTCTCTCCTGCAGA	AGGAAGAGACGTTCAGGTGG
MLKL	AGGTCTAGGCCACACTTGTC	TGCAGGTCATGGGCTTCTAA
BZLF1	AATGCCGGGCCAAGTTTAAGCA	TTGGGCACATCTGCTTCAACAGGA
BDLF1	CGCAGACATGCTCGATGTA	TAGTGGTGCCCCAGGTATG
EBER	CCCTAGTGGTTTCGGACACA	ACTTGCAAATGCTCTAGGCG
BDRF1	CGGAGTGGCTCAGTCTAAGG	AGGTGGGCTGACACAGAC
ChIP-qPCR
CpG island 127	TGACCTCTGCAGCTTCCTCT	GATGATCTGCCGCCAACTT
Z prom (0)	TAGCCTCGAGGCCATGCATATTTCAACT	GCCAAGCTTCAAGGTGCAATGTTTAGTG
Z prom (+600)	AGGTATGTTCCTGCCAAAGC	GTTCATGGACAGGTCCTGTG
H8 rDNA	AGTCGGGTTGCTTGGGAATGC	CCCTTACGGTACTTGTTGACT
BZLF1 prom	GGAGAAGCACCTCAACCTG	CTCCTTACCGATTCTGGCTG
EBNA Cp	AGT TGG TGT AAA CAC GCC GT	TCCACCTCTAAGGTCCCACG
Globin prom	AGGACAGGTACGGCTGTCATC	TTTATGCCCAGCCCTGGCTC
GAPDH prom	CGTGCCCAGTTGAACCAGG	AGGAGGAGCAGAGAGCGAAG
EBNA Qp	GGCTCACGAAGCGAGAC	GTCGTCACCCAATTTCTGTC
KDM2B cg21423404	ACCTGACACACCTCAACTCC	TTGTGGTTTGGGAGAAGGGT
KDM2B cg15695155	CTTGCCCCTTCCCACTAGAG	CCCTCTTCCCCAAACCATG

miRNA-146-5p analysis.

To analyze cellular miRNA, total RNA extracted by the AllPrep DNA/RNA minikit (Qiagen) was reverse transcribed using an miRCURY LNA miRNA PCR system (miRCURY LNA RT kit; catalog number 339340) according to the manufacturer’s protocol. The cDNA was then analyzed for the levels of miRNA-146-5p using a specific miRNA qPCR primer (catalog number 339306; Hsa-miR-146a-5p miRCURY LNA miRNA PCR assay) and a PCR kit (catalog number 339345; miRCURY LNA SYBR green PCR kits) on a Bio-Rad qPCR machine (CFX96 Touch real-time PCR).

KDM2B overexpression and gene expression silencing.

The KDM2B coding region was cloned into a pCDNA3 vector in frame with a hemagglutinin (HA) tag at the N terminus. LCL (1 × 10⁷) and Louckes cells (5 × 10⁶) were transfected with increasing concentrations of HA-KDM2B pCDNA3 (0.5, 1.0, and 1.5 μg) or with the pCDNA3 vector as a control by electroporation using a Neon transfection system (10-μl tips; pulse voltage, 1,350 V; pulse width, 30 ms; pulse number, 1). At 24 h after transfection, the cells were collected and processed for RNA/DNA extraction. Gene silencing of KDM2B was performed using KDM2B (human) unique 27-mer siRNA duplexes (catalog number HSS150072; Thermo Fisher Scientific). LCL (1 × 10⁷) and Louckes cells (5 × 10⁶) were transfected with the siRNA (final concentration, 250 nM) by electroporation using the Neon transfection system (10-μl tips; pulse voltage, 1,350 V; pulse width, 30 ms; pulse number, 1). At 48 h after transfection, the cells were collected and processed for RNA/DNA extraction. The levels of silencing were evaluated by qPCR using KDM2B-specific primers, indicated in Table 3.

Immunoblotting and antibodies.

Whole-cell lysate extracts were obtained using lysis buffer, as previously described (37). The cell extracts were then fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and processed for immunoblotting using standard techniques. The following antibodies were used for immunoblotting: KDM2B (catalog number 09-864 from Merck Millipore and catalog number ab5199 from Abcam), mouse monoclonal anti-human MLKL (catalog number SC-293201; clone 3B2; CliniSciences), LMP1 (S12 monoclonal antibody), β-actin (clone C4; MP Biomedicals), and GAPDH. Images were produced using a ChemiDoc XRS imaging system (Bio-Rad).

Immuno-FISH.

Fifty thousand cells were resuspended in 5 μl of phosphate-buffered saline (PBS). The cells were gently spread on a microscope glass slide, air dried, and fixed in 4% paraformaldehyde–PBS for 10 min at room temperature. Slides containing fixed cells were washed 3 times in PBS for 5 min, permeabilized with PBS–0.5% Triton X-100 (Sigma-Aldrich) for 15 min, and then washed twice with PBS–0.05% Tween. The slides were then soaked in methanol–0.3% H₂O₂ for 30 min and incubated for 1 h with antibody diluent (catalog number S3022; Dako) and then for 30 min with Image-iT FX signal enhancer. The slides were incubated overnight at 4°C with an anti-KDM2B antibody (catalog number ab5199; Abcam) diluted to a concentration of 1 μg/ml, followed by incubation with a secondary antibody (anti-goat immunoglobulin; 5 μl/ml; Elite kit; Vector). To amplify the signal, the slides were incubated for 30 min at 37°C with ABC kit reagents according to the manufacturer’s protocol. EBV DNA staining by FISH was performed as previously described (36), using a biotinylated probe to the EBV DNA genomic region BWRF1 (A300P.0100 DS-Dish-Probes). The stained cells were visualized with a fluorescence microscope with an incubator (Nikon Eclipse).

Immunohistochemistry and ISH for EBER.

Immunohistochemistry analysis for KDM2B (dilution, 1:200; catalog number ab5199; Abcam) was performed by an automated staining system (Ventana BenchMark Ultra; Roche Diagnostics, Monza, Italy) on formalin-fixed, paraffin-embedded 4-μm-thick sections. An UltraView universal detection kit (Ventana) using a horseradish peroxidase multimer and DAB (as the chromogen) was used. ISH for EBER was carried out in each sample on 4-μm-thick sections, as previously reported (38). A control slide, prepared from a paraffin-embedded tissue block containing a metastatic nasopharyngeal carcinoma in a lymph node, was used as a positive control.

Chromatin immunoprecipitation.

ChIP was performed with Diagenode Shearing ChIP and OneDay ChIP kits according to the manufacturer’s protocols. The following antibodies were used: KDM2B (catalog number ab5199; Abcam), DNMT1 (catalog number MAB0079; Abnova), and IgG (Diagenode). The eluted DNA was used as a template for qPCR with primers designed on the KDM2B gene. The primers used for quantitative ChIP are listed in Table 3. The value of binding obtained for each antibody was calibrated on the input sample and normalized to the values for IgG.

Bisulfite modification and pyrosequencing.

Samples for pyrosequencing were processed as previously described (4, 39). The primers are indicated in Table 4.

TABLE 4.

Primers used for pyrosequencing

Pyrosequencing primers	Sequence
	Forward	Reversea	Sequencing primer
KDM2B cg15695155	GGAGTGGGGTAGAGTTGAA	CCTACATACTACTAAACCCCC	AGGTTTGGT
			GAGTTTTAGGTGG
			GGATGGGTAGTT
			AGGGAAGGAATG
			AGTGGAGATAATG
CpG.127	AAATACAACAACCCTCCTACC	AAATACAACAACCCTCCTACC	GGGTGGTTGGGATAG
			TTTGGTTGGTTTGTT
			TTTTTTTAAGTATAT
			TTAAGTTTTTTTTA
KDM2B cg21423404	GATAAGTATAGGGAGGTTTGTGA	CTATAAAACCATTTCCAACCC	GTAGGTGGTGATT
			GGTTATTAGAGT
	GGGTTGGAAATGGTTTTATAG	CCTCCCTAATAACTAAAACTACA	GTTGTTTATATG
			TTAATATAATGGT
KDM2B cg15695155	GAGTTTTAGGTGGTAYGGATG	GAYGGATAGGGAGGAGTTAGT	GGATGGGTAGTT
KDM2B cg00031896	GTAAAGGAGGAAATTAGGATTA	TATGTTTAAAGGAGGTTGTATG	TGGTTTGGTTAT
KDM2B cg21423404	AGGAGGAGTTTAGAGGTTATAGT	AGTTATTGTAGGGGTAGATTTTAG	GTAGGTGGTGATT
KDM2B cg12251659	GGAGGGAGTTYGGGAGGTAT	TTGGAGGGTYGAGTTGTAGG	AGTGYGTTTTTGTA

^aThe reverse primer was labeled with biotin at the 5′ end.

Whole-genome expression analysis.

Differential expression analysis was performed using human HT-12 expression BeadChips (Illumina) as previously described (4, 36). Probes with P values of <0.01, a false-discovery rate (FDR) of <0.05, and a fold change in expression of at least 1.5 were considered differentially expressed.

Statistical analysis.

Statistical significance was determined by Student's t test. The P value of each experiment is indicated in the corresponding figure legend. Error bars in the graphs represent the standard deviation.