Human cord blood B cells infected with an EBNA2-defective EBV mutant (ΔEBNA2 EBV) grow indefinitely on a CD40L/IL21 expressing feeder cell line

To construct an EBV mutant virus that cannot express EBNA2 but does not remove sequences of other EBV proteins such as EBNA-LP, we used CRISPR/Cas9 technology to create an out-of-frame mutation in the EBNA2 gene (starting at amino acid 25) in the AG876 EBV strain (Fig 1A) as described in the Methods. EBV-infected HeLa cell clones containing only the mutant (and not wild-type) EBNA2 gene were transfected with the EBV immediate-early BZLF1 and BRLF1 genes to induce lytic viral reactivation, and the infectious virions produced were used to infect human cord blood cells cultured on an irradiated feeder cell layer (YK6-CD40lg-IL21) expressing CD40 ligand and soluble IL21 as previously described [38]. Uninfected or ΔEBNA2 EBV-infected cells were passaged and expanded as needed on the feeder layer. Similar to previous studies using primary human tonsil germinal center (GC) B cells on the same feeder layer [38], the uninfected cord blood B cells initially proliferated on the feeder layer but subsequently died within 1 month. In contrast, cord blood B cells derived from two different donors (“Donor 1” and “Donor 2”) resulted in long term EBV-infected B cell lines that can be maintained indefinitely on the feeder layer (but not off the feeder layer) when infected with the ΔEBNA2 EBV mutant, although the production of such lines is much less efficient compared to cells infected with the wild-type EBV. Both Donor 1 and Donor 2 ΔEBNA2 EBV-infected cord blood lines express Kappa, but not Lambda, light chain, suggesting that they are clonal or oligoclonal.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Construction of an EBNA2-deleted EBV mutant (ΔEBNA2 EBV).

A. CRISPR/Cas9 gene editing was used to insert a 10 bp deletion in the EBNA2 gene of the AG876 EBV strain. The nucleotide sequences deleted, and the resultant translated EBNA2 protein, are shown. B. Protein extracts of human cord blood B cells stably infected with the ΔEBNA2 EBV mutant and maintained on a CD40L/IL21 expressing feeder layer were examined by immunoblot to assess expression of the EBV latency proteins EBNA2, EBNA1, LMP1, and EBNA3A and the GFP protein as indicated. EBV positive and EBV negative Akata Burkitt lymphoma cell lines and an LCL infected with Akata EBV (containing the GFP gene inserted into the viral BXLF1 gene locus) serve as negative and positive controls for expression of various EBV proteins and GFP.

https://doi.org/10.1371/journal.ppat.1012132.g001

To confirm that ΔEBNA2 EBV-infected cord blood (CB) B cell lines lack EBNA2 expression, immunoblot analysis was performed in CB cells infected with ΔEBNA2 EBV. As shown in Fig 1B, ΔEBNA2 EBV-infected cells (grown on the feeder layer) express EBNA1 and LMP1, but not EBNA2 or EBNA3A, indicating the cells have “type II”-like viral latency (EBNA1+/EBNA2-neg, LMP1-pos). The expression of LMP1 in ΔEBNA2 EBV-infected CB B cells grown on the CD40L/IL21 expressing feeder layer is consistent with the known ability of IL21 to induce LMP1 expression in the absence of EBNA2 [34,35]. Together, these in vitro results indicate that infection with ΔEBNA2 EBV is sufficient to confer long-term proliferation potential to primary cord blood B cells on the IL21/CD40L expressing feeder layer even in the absence of any cooperating oncogenes.

ΔEBNA2 EBV cooperates with Myc over-expression to induce lymphomas in NSG mice

To determine if Myc over-expression in human CB B cells cooperates with ΔEBNA2 EBV infection to form B-cell lymphomas in NSG mice in vivo, and/or alters the phenotype of ΔEBNA2 EBV-infected lymphomas, ΔEBNA2 EBV-infected CB B cells were co-infected with a Myc-expressing retrovirus as described in the Methods. ΔEBNA2 EBV-infected “Donor 1” CB cells were co-infected with a retrovirus vector expressing the wild-type mouse Myc gene, and ΔEBNA2 EBV-infected “Donor 2” CB cells were co-infected with a retrovirus vector expressing a human Myc gene containing a T58A mutation previously shown to inhibit Myc-induced BIM expression [12]. Similar results were obtained using both Myc vectors. CB cells co-infected with ΔEBNA2 EBV and the Myc vectors outgrew the cells infected with ΔEBNA2 EBV alone on the feeder layer system even in the absence of any drug selection. Similar to previous results using primary human tonsil GC B cells [38], CB B cells infected with the Myc-expressing vectors alone did not result in long term lines, presumably due to Myc-induced apoptosis. To examine levels of Myc expression, immunoblot analysis was performed on CB cell extracts harvested from cells infected with ΔEBNA2 EBV alone or co-infected with ΔEBNA2 EBV plus a Myc retrovirus vector. As shown in Fig 2A, CB cells co-infected with ΔEBNA2 EBV and a Myc vector express Myc at similar levels found in human BL cell lines, and at a level much higher than that which occurs in wild-type EBV-transformed lymphoblastoid cell lines (LCLs) with type III latency. Expression of the EBV latency protein LMP1 was not altered by Myc co-expression when CB cells were grown in vitro on the feeder layer.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Cord blood B cells co-infected with ΔEBNA2 EBV and a Myc vector cause more lymphomas in NSG mice compared to B cells infected with ΔEBNA2 EBV alone.

A. Human cord blood B cells stably infected with the ΔEBNA2 EBV mutant were co-infected with or without a retrovirus expressing mouse wild-type Myc protein as indicated and maintained on a CD40L/IL21 expressing feeder layer without antibiotic selection. One month after retrovirus infection, protein extracts were harvested and immunoblot analysis was performed to assess expression levels of Myc, LMP1, and tubulin as indicated. EBV-negative and EBV-positive BL cells (Akata) serve as a positive control for elevated Myc expression. B and C. NSG mice were injected subcutaneously with 10 million cord blood B cells (suspended in Matrigel) infected with ΔEBNA2 EBV alone, or co-infected with ΔEBNA2 EBV and a Myc expressing vector. Donor 1 cells (used in B) were co-infected with a wild-type mouse Myc expressing retrovirus, while Donor 2 cells (used in C) were co-infected with a retrovirus expressing a mutant human Myc (T58A) protein. The number of animals sacrificed with tumors at various time points after injection of cells is shown. The p-values were calculated using the Kruskal-Wallis method and program M-Stat 7.0. D. Tumors were paraffin-fixed and H&E staining performed to examine tumor morphology and IHC staining performed to examine expression of the germinal center B cell marker, CD10.

https://doi.org/10.1371/journal.ppat.1012132.g002

We next injected NSG mice subcutaneously in the flank with 10 million CB cells (suspended in Matrigel) infected with ΔEBNA2 EBV alone, or the combination of ΔEBNA2 EBV and a Myc expressing vector. Similar results were obtained using either of the two Myc vectors. As shown in Fig 2B, 15/15 mice injected with CB “Donor 1” cells containing both ΔEBNA2 EBV and wild-type mouse Myc protein developed tumors by day 50, whereas only 14/22 mice injected with “Donor 1” CB cells infected with ΔEBNA2 EBV alone developed tumors by day 120, and the latter tumors occurred at later time points. Similarly, 10/10 animals injected with CB “Donor 2” cells containing both ΔEBNA2 EBV and the human Myc T58A mutant protein developed tumors by day 50, whereas 0/10 mice injected with “Donor 2” CB cells infected with ΔEBNA2 EBV alone developed lymphomas by day 120 (Fig 2C). Tumors obtained from Donor 1 were also paraffin fixed and examined by microscopy. Lymphomas infected with ΔEBNA2 EBV alone resemble diffuse large B cell lymphomas with large polymorphous B cells, including some multinucleated cells, while tumors infected with both ΔEBNA2 EBV and Myc are composed of homogenous, medium-sized B cells with distinct cytoplasmic borders and numerous mitotic figures, closely mimicking human BL (Fig 2D). Thus, over-expression of Myc clearly cooperates with ΔEBNA2 EBV to allow human cord blood B cells to form lymphomas in NSG mice, although there is some donor variability in the ability of cord blood cells infected with ΔEBNA2 EBV alone to form tumors.

ΔEBNA2 EBV plus Myc lymphomas resemble human BLs, turn off LMP1 expression and have low BIM expression and wild-type p53

As the Donor 1 CB cells produced lymphomas in cells infected with both ΔEBNA2 EBV alone (“ΔEBNA2”), or the combination of ΔEBNA2 EBV and Myc retrovirus (“ΔEBNA2 + Myc”), we compared the phenotypes of the tumors derived from this CB donor in the presence and absence of Myc over-expression. Immunoblot analyses of tumor proteins revealed that lymphomas infected with ΔEBNA2 + Myc have high level Myc expression (similar to that in the Akata human BL cell line) and turn off expression of the EBV latent LMP1 protein in most tumors (Fig 3A), as well as expression of the EBV latent LMP2A protein (S1A Fig). Thus, as occurs in human BL tumors, the ΔEBNA2 + Myc lymphomas have a more stringent form of EBV latency (EBNA1-pos, LMP1-neg, EBNA2-neg) in comparison to the lymphomas infected with ΔEBNA2 alone, which have high level LMP1 expression. The ΔEBNA2 + Myc lymphomas also express the GC marker CD10 (MME), while ΔEBNA2 lymphomas have higher expression of CD30 (a B cell activation marker), BLIMP1 (PRDM1, a marker for plasma cell differentiation) and BZLF1 (a lytic EBV protein) in comparison to ΔEBNA2 + Myc lymphomas (Figs 2D and 3A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Cord blood B cells co-infected with ΔEBNA2 EBV and a Myc retrovirus cause Burkitt-like EBV-positive tumors that lose LMP1 expression.

A. Protein extracts were harvested from tumors derived with Donor 1 cells infected with ΔEBNA2 EBV alone, or co-infected with ΔEBNA2 EBV and a Myc-expressing vector, and immunoblot analysis was performed to examine expression of various different EBV latency proteins, GFP and various cellular proteins as indicated. EBV-negative Akata BL cells serve as a positive control for Myc expression (and other cellular proteins such as CD10 expressed in BL tumors) and a negative control for EBV protein expression, while an AG876 wild-type EBV-infected LCL serves as a positive control for viral proteins expressed in type III latency. B. Lymphomas infected with ΔEBNA2 EBV and Myc or ΔEBNA2 EBV alone were examined for expression of the EBV EBNA3A latency protein by immunoblot. EBV-neg and EBV+ Akata BL cells serve as a negative control. C. Lymphomas infected with ΔEBNA2 EBV and Myc or ΔEBNA2 EBV alone were examined for expression of Myc, the EBV BHRF1 protein or the cellular BIM protein (the predominant isoform BIM_EL or extra-large) by immunoblot. EBV-neg Akata BL cells serve as a negative control for BHRF1 expression and a BL5 strain EBV infected LCL serves as a positive control for BHRF1 expression. Tumors derived from Donor 1 versus Donor 2 are indicated. D. p53 immunoblot was performed to compare p53 expression in extracts isolated from stable cell lines derived from Donor 2 ΔEBNA2 + Myc tumors (N1, N3, N4, R3, R4), the Akata (p53-deleted) BL line, the Mutu I (p53 mutant) BL line, the P3HR1 (p53 mutant) BL line, or an EBV-transformed LCL (BL5) line as indicated.

https://doi.org/10.1371/journal.ppat.1012132.g003

In contrast to the type II-like latency observed when the ΔEBNA2 + Myc cord blood cell lines were grown in vitro on the CD40L/IL21 expressing feeder layer (Fig 1B), we found that the ΔEBNA2 + Myc lymphomas, in addition to losing LMP1 expression, also express high levels of the EBNA3A (Fig 3B) and EBNA-LP (S1B Fig) proteins, similar to Wp-restricted human BLs, although unlike Wp-restricted human BLs they do not express the EBV BCL2 homologue, BHRF1 (Fig 3C). Although commercially available antibodies do not recognize the form of EBNA3C protein expressed in type 2 EBV strains such as AG876, since EBNA3A and EBNA3C are essentially always expressed together in EBV infected cells, it is highly likely that the tumors also express EBNA3C. Because EBNA3A and EBNA3C collaboratively inhibit expression of the Myc-induced pro-apoptotic protein, BIM, in vitro [15], we compared BIM expression in ΔEBNA2 + Myc versus ΔEBNA2 alone lymphomas, as well as in an EBV-negative BL cell line (Akata), which has spontaneously lost the EBV genome. In contrast to EBV+ Akata cells, EBV-negative Akata cells are unable to form tumors in NSG mice and are more susceptible to apoptosis-inducing stimuli [39]. The level of BIM in the ΔEBNA2 + Myc Donor 1-derived lymphomas (which have wild-type mouse Myc) is much lower than that expressed in the EBV-negative Akata BL line, although similarly high levels of Myc are expressed (Fig 3C). RNA-seq analysis of Donor 1 tumors infected with ΔEBNA2 + Myc versus ΔEBNA2 alone also showed that BCL2L11 gene expression is not increased in the Myc-expressing tumors (S1 Table). These results suggest that one or more EBV genes/RNAs expressed in ΔEBNA2 + Myc lymphomas prevents Myc-induced BIM expression.

As p53 is commonly mutated in human BLs (particularly EBV negative BLs), we also examined the level of p53 expression in ΔEBNA2 + Myc lymphomas and determined the p53 (TP53) sequence (from our RNA-seq data) in three different ΔEBNA2 + Myc Donor 1 derived tumors as described in the Methods. The level of p53 protein expression in stable cell lines derived from the ΔEBNA2 + Myc lymphomas is similar to that expressed in EBV-transformed LCLs (Fig 3D), while the p53-deleted Akata BL line as expected lacks p53 expression, and the Mutu I and P3HR1 BL cell lines (both of which have mutant p53) have massively increased p53 expression. Analysis of the p53 transcript sequence obtained by RNA-seq confirmed that all three Myc-expressing tumors examined have wild-type p53 (S2 Table). Together, these results suggest that the relatively few EBV genes/RNAs expressed in the Myc-induced tumors are sufficient to inhibit Myc-induced BIM expression, and to substitute for the p53 mutations that often occur in human BLs.

ΔEBNA2 + Myc lymphomas have enhanced expression of BL-associated cellular genes

To further compare cellular gene expression in lymphomas infected with ΔEBNA2 alone versus ΔEBNA2 + Myc, RNA was isolated from tumors and bulk RNA-seq analysis was performed (using tumors derived from Donor 1). As shown in Figs 4A and S2 and S1 Table, tumors infected with ΔEBNA2 + Myc have highly altered cellular gene expression in comparison to tumors infected with ΔEBNA2 alone. Analysis of the EBV gene expression pattern revealed that the ΔEBNA2 + Myc tumors have a lower level of lytic EBV gene expression compared to the tumors infected with ΔEBNA2 alone (S2C Fig), consistent with a previous study showing that Myc decreases lytic EBV gene expression in Burkitt lymphomas [40]. MIXCR analysis of the RNA-seq results (performed as described in the Methods) revealed that each of the Donor 1-derived tumors (whether infected with ΔEBNA2 only or both ΔEBNA2 + Myc) is a monoclonal B cell tumor with low IGHM expression and a rearranged Kappa light chain with a 2D-30/J3 clonotype. Thus, the ΔEBNA2-infected cord blood Donor 1 cell line used to make the tumors (with or without Myc) is likely derived from a single cell. The relatively low level of IGHM expression in Donor 1 tumors was confirmed at the protein level, although Donor 2 derived tumors express high levels of IGHM (S2D Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. RNA-seq analysis of tumors infected with ΔEBNA2 EBV alone versus both ΔEBNA2 EBV and Myc vector reveals Burkitt lymphoma-like gene expression in the Myc positive tumors.

RNA was isolated from Donor 1 derived tumors infected with ΔEBNA2 EBV alone or both ΔEBNA2 EBV and Myc vector and RNA-seq analysis was performed. A. A heatmap comparing cellular gene expression in tumors with and without Myc vector co-infection is shown. B. GSEA analysis was performed to compare gene expression in tumors with and without Myc. Selected gene sets that have up-regulated expression in the Myc-expressing tumors are shown. C. Selected gene sets that have down-regulated expression in the Myc-expressing tumors are shown.

https://doi.org/10.1371/journal.ppat.1012132.g004

GSEA analysis of RNA-seq results confirmed that Myc-expressing tumors have upregulated expression of genes in the “Hummel_Burkitt_Lymphomas_Up” gene set (Fig 4B), which distinguishes genes expressed in human Burkitt lymphomas versus human GC type DLBCLs [24]. Another top gene signature activated in Myc-expressing lymphomas is the “Kegg_Ribosome” signature (Fig 4B), consistent with Myc’s well known ability to increase expression of many ribosome components [41]. Somewhat unexpectedly, Myc-expressing tumors also have higher expression of genes in the “Haddad_B_Lymphocyte_Progenitor” signature (Fig 4B) [42], including a number of genes highly expressed in immature B cells. Gene sets down-regulated in the Myc-expressing lymphomas include “Hallmark_TNFA_signaling_via_NF-κB” [43] and “Dirmeier_LMP1_Response_Early” [44] (Fig 4C), reflecting Myc’s ability to turn down NF-κB signaling [45] as well as LMP1 expression. The “Heller_Silenced_By_Methylation_Up” gene set (which includes cellular genes in multiple myeloma that are silenced by DNA methylation and turned on by demethylating agents) [46] is also down-regulated in Myc-expressing lymphomas (Fig 4C), suggesting that Myc may inhibit expression of some viral and/or cellular genes by inducing DNA methylation. In addition, the “Azare_Neoplastic_Transformation_By_STAT3_Up” [47] gene set was down-regulated in Myc-expressing lymphomas, consistent with the previously described down-regulation of STAT3 expression in human BLs [24].

ΔEBNA2 + Myc lymphomas have enhanced expression of BL core survival proteins and GC B cell proteins

CRISPR/Cas9 screens and other studies have identified cellular genes required for proliferation and survival of human BL cells in vitro, including some like CCND3 (cyclin D3) that are also required for proliferation of normal human GC B cells [48], but are not required for growth and survival of EBV-transformed lymphoblastoid cell lines [49]). Immunoblot analysis of tumor extracts confirmed that cellular proteins that are highly expressed in GC B cells, including cyclin D3, GCSAM, TCL1, Myb, TCF3 and BACH2 are expressed at higher levels in the ΔEBNA2 + Myc lymphomas in comparison to the ΔEBNA2 alone lymphomas (Fig 5A–5C). The transcription factor TCF3, which is required for tonic B cell receptor signaling, is an essential survival factor for human BLs [50], while transcription factors Myb and BACH2 regulate expression of many different GC B cell genes and are required for formation of the GC [51,52]. BACH2 also turns off BLIMP1 (PRDM1) expression, thereby blocking plasma cell differentiation [52]. In contrast, cyclin D2, which is expressed in EBV-transformed LCLs but not in EBV+ BLs [49], is turned down in the ΔEBNA2 + Myc lymphomas (Fig 5A). Interestingly, the one lymphoma infected with ΔEBNA2 + Myc that still expresses LMP1 (tumor #8 in Fig 5A) expressed cyclin D2 but not cyclin D3, suggesting that loss of LMP1 expression contributes to the cyclin D2 to cyclin D3 switch. Nevertheless, the ΔEBNA2 + Myc lymphoma cells do not express the master regulator of GC differentiation, BCL6 (S3 Fig) and thus are not true GC B cells. Together, these results reveal that high level Myc expression in naïve B cells activates expression of many cellular genes normally associated with the GC B cell differentiation state even in the absence of BCL6 expression, particularly those genes expressed in the GC “dark zone”.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. ΔEBNA2 EBV plus Myc lymphomas have enhanced expression of GC B cell proteins.

Protein extracts were harvested from tumors infected with ΔEBNA2 EBV alone, or co-infected with ΔEBNA2 EBV and Myc vector, and immunoblot analysis was performed to examine expression of various cellular proteins known to be highly expressed in GC B cells and/or BL tumors as indicated. A. Extracts from Donor 1 derived tumors with or without Myc expression were examined for expression of Myc, the EBV LMP1 protein, cyclin D3, cyclin D2, GCSAM or TCL1 as indicated. EBV negative Akata BL cells serve as a positive control for Myc and GC B cell expressed proteins. The same actin immunoblot will be seen in Figs 5D, 6A and 7A, while the LMP1 and Myc blots will be repeated in Figs 5D and 6A. B. Extracts harvested from Myc expressing tumors derived from Donor 1 or Donor 2 as indicated or Donor 1 derived tumors without Myc were examined by immunoblot for expression of Myc, Myb and BACH2 as indicated. Control extracts include the EBV-negative Akata human BL cell line and a BL5 strain EBV infected LCL. C. Extracts harvested from Myc expressing tumors derived from Donor 1 or Donor 2 as indicated or Donor 1 derived tumors without Myc were examined by immunoblot for expression of TCF3 as indicated. Control extracts include the EBV-negative Akata human BL cell line and a BL5 strain EBV infected LCL. D. Extracts from Donor 1 derived tumors with or without Myc expression were examined for expression of Myc, the EBV LMP1 protein, DNMT3B protein, DNMT1 protein and UHRF1 protein as indicated. The same extracts used in Fig 5A were used for this blot and the Myc, LMP1, and actin blots from Fig 5A are reproduced here to show levels of expression in each condition.

https://doi.org/10.1371/journal.ppat.1012132.g005

Myc-expressing lymphomas have enhanced expression of the DNMT3B and UHRF1 proteins

Although human BLs are globally hypo-methylated, specific cellular genes, including genes activated by NF-κB and STAT signaling, are hyper-methylated in human BL tumors [53]. The RNA-seq results (S1 Table) revealed that the de novo DNA methyltransferase gene DNMT3B and the UHRF1 gene (required for function of the maintenance DNA methyltransferase, DNMT1) are expressed at higher levels in the Myc-expressing tumors, and this result was confirmed at the protein level using immunoblot analysis (Fig 5D). Expression of DNMT1 is also generally increased in the Myc-expressing tumors (Fig 5D) at the protein level. UHRF1 is expressed at extremely high levels in normal GC-type B cells and is required for their proliferation [54], while DNMT3B and DNMT1 are highly expressed in human BLs in a Myc-dependent manner [55]. Furthermore, DNA methylation-mediated repression of LMP1 expression in EBV-superinfected Burkitt lymphoma cell lines in vitro is initiated by the DNMT3B de novo methyltransferase and maintained by the DNMT1/UHRF1 maintenance methylation complex [33]. These results suggest that Myc-induced expression of DNMT3B and UHRF1 expression could play a role in inhibiting LMP1 expression in the Myc-expressing lymphomas.

Myc-expressing lymphomas have decreased expression of NF-κB2, phosphorylated STAT3, and Src kinase

RNA-seq analysis also showed that the Myc-expressing lymphomas, like human BLs, have decreased expression of the NF-κB2 and STAT3 transcripts, as well as decreased Src kinase transcripts (S1 Table). Except for the one Myc-expressing tumor that also expresses high level LMP1, immunoblot analysis of tumor extracts confirmed that expression of both the full-length and cleaved (activated) forms of the NF-κB2 (p100/52) is much lower in the tumors expressing Myc (Fig 6A). The expression level of total Src protein, and tyrosine 705-phosphorylated STAT3 protein, is also decreased in the Myc-expressing lymphomas (Fig 6B) although the level of total STAT3 protein is not consistently different. Src directly phosphorylates STAT3 on tyrosine 705, allowing STAT3 to enter the nucleus and transcriptionally activate its target genes [56]. The human Akata BL cell line was also found to have much reduced Src expression and phosphorylated STAT3 in comparison to an EBV-infected LCL (Fig 6B). These results indicate that Myc-expressing lymphomas in this new model system have low levels of NF-κB2 and STAT3 activity, as is the case for human BLs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Lymphomas infected with ΔEBNA2 EBV plus Myc have decreased NF-κB2 signaling and decreased STAT3 activity compared to lymphomas with ΔEBNA2 EBV alone.

A. Protein extracts were harvested from tumors derived with Donor 1 cells infected with ΔEBNA2 EBV alone, or co-infected with ΔEBNA2 EBV and a Myc expressing vector, and immunoblot analysis was performed to examine expression of Myc, LMP1, and p100 or its cleaved (activated) product, p52. The same extracts used in Fig 5A were used for this blot and the Myc, LMP1 and actin blots from Fig 5A are reproduced here to show levels of expression in each condition. B. Immunoblot analysis was performed on extracts isolated from Donor 1 tumors with or without Myc and expression of total STAT3, activated STAT3 (phosphorylated on tyrosine 705) or total Src was examined by immunoblot as indicated.

https://doi.org/10.1371/journal.ppat.1012132.g006

Myc-expressing lymphomas express proteins associated with the pro-B cell and pre-B cell differentiation states

To further investigate the RNA-seq finding that genes in the “Haddad_B_Lymphocyte_Progenitor” gene set [42] are more highly expressed in the Myc-positive lymphomas (Fig 4B), immunoblot analysis was performed to examine protein expression of genes in this gene set that are upregulated in the RNA-seq results (S1 Table). In comparison to lymphomas infected with ΔEBNA2 alone, the ΔEBNA2 + Myc tumors have much higher expression of IGLL1 (CD179B, surrogate light chain), DNTT (terminal deoxynucleotidyltransferase, TDT), RAG1 and IL7R (Fig 7A and 7B). The expression of these proteins is normally highest in the pro-B and/or pre-B cell differentiation states, although RAG1, RAG2 and IL7R are also reported to be turned on in GC B cells undergoing light chain receptor revision [57,58]. Furthermore, we found that stable cell lines derived from the ΔEBNA2 + Myc lymphomas (grown off the feeder layer) also often express one or more pre-B cell and/or pro-B cell associated proteins (S4 Fig) and respond to IL7 cytokine treatment by inducing STAT5 tyrosine phosphorylation (Fig 7C), indicating that the IL7R is functional. IL7 has been reported to inhibit TDT expression [59] and IL7 treatment decreases TDT expression in two out of three stable tumor-derived cell lines (S4 Fig). Importantly, we found that some human BL cell lines also express the IL7R and respond to IL7 cytokine (Fig 7B and 7C), and express the surrogate light chain CD179B (Fig 8). Thus, GC-derived human BLs express certain genes that are normally associated with an immature B cell phenotype, but their expression of immunoglobulin light chains and germinal center associated proteins confirms their mature B cell differentiation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. ΔEBNA2 EBV plus Myc lymphomas express cellular proteins associated with an immature B cell differentiation state.

A. Protein extracts were harvested from tumors derived with Donor 1 cells infected with ΔEBNA2 EBV alone, or co-infected with ΔEBNA2 EBV and a Myc expressing vector, and immunoblot analysis was performed to examine expression of cellular proteins highly expressed in pro-B cells and/or pre-B cells, including CD179B (IGLL1), DNTT (TDT) and RAG1 as indicated. The extracts used are the same as those in Fig 5A with the same actin blot. B. Protein extracts harvested from Donor 1 derived tumors with or without Myc expression, or from an EBV-negative Akata BL cell line or an EBV strain BL5-infected LCL line, were examined by immunoblot analysis for expression of IL7R. C. Stable cell lines derived from Donor 2 ΔEBNA2 EBV plus Myc tumors (right panel), or the EBV-positive Akata Burkitt cell line (left panel) were treated with or without IL7 for 30 minutes and the level of Y694-phosphorylated STAT5 was examined by immunoblot.

https://doi.org/10.1371/journal.ppat.1012132.g007

Myc and Myb collaboratively regulate cellular gene expression in Burkitt and Burkitt-like tumors

As the changes in cellular gene expression identified by RNA-seq in tumors with and without Myc could potentially be due to the effects of Myc, and/or loss of LMP1 expression, we performed experiments in the P493-6 human B cell line to identify genes regulated by Myc even in the absence of LMP1 expression. The P493-6 B cell line, which is often referred to as having a “Burkitt-like” phenotype and has been widely used to identify cellular genes turned on and off by Myc in B cells, is infected with the EBNA2-deleted P3HR1 EBV strain and contains a doxycycline-regulated Myc vector but does not express LMP1 or EBNA2 unless treated with estrogen (which turns on EBNA2 expression) [22]. As shown in Fig 8A (upper left panel), we confirmed that P493-6 cells express the EBV EBNA1 protein but not the LMP1 or EBNA2 proteins, allowing us to examine the effect of Myc expression on various cellular proteins without the confounding effect of altered LMP1 expression. As expected, Myc expression was dramatically decreased in this cell line after three days of doxycycline treatment. Furthermore, loss of Myc expression was accompanied not only by increased expression of the STAT3, Src and cleaved NF-κB2 proteins, but also by decreased expression of the Myb, DNMT3B and UHRF1 proteins, confirming that Myc regulates expression of each of these proteins in B cells independent of LMP1. Of note, the P493-6 cell line expresses much lower levels of some BL-associated (and GC B cell expressed) cellular proteins in comparison to the N1, N3 and N4 stable cell lines derived from the ΔEBNA2 + Myc lymphomas (Fig 8B), including CD10, GCSAM, BACH2, and CD179B (IGLL1). In comparison to the P493-6 line, the tumor-derived lines express similar levels of EBNA1 and EBNA3A but express much less BHRF1 (S5 Fig). Thus, the tumor derived cell lines derived from ΔEBNA2 + Myc lymphomas (except for the LMP1-expressing R3 line) more closely resemble human EBV+ BLs in comparison to the P493-6 cell line.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Both Myc and Myb are involved in altering expression of cellular genes in the ΔEBNA2 EBV plus Myc lymphomas.

A. P493-6 cells were treated with or without doxycycline for 3 days, and then protein extracts were harvested to examine expression of various cellular proteins and EBV proteins as indicated. Mutu I is an EBV+ human BL cell line with type 1 latency. The blots shown on the left and top right used the same set of extracts and have the same actin blot. The numbers below each immunoblot quantify the results using Image Studio Lite software to normalize the levels of STAT3, Src, Myb, IL7R, UHRF1, DNMT3B, p100, and p52 expression to actin expression. Results are presented as the ratio of STAT3, Src, Myb, IL7R, UHRF1, DNMT3B, p100, and p52 expression in doxycycline-treated cells relative to untreated P493-6 cells, after being normalized to actin. Values of untreated cells are set as 1. The same extracts are used in S5 and S7 Figs and the actin blot is reproduced there. B. Immunoblot analysis was performed to compare expression of various different GC B cell proteins as indicated in stable tumor-derived cell lines (N1, N3, N4) versus the P493-6 cell line (with Myc expression turned on). The Mutu I BL line and the BL5 LCL line serve as positive and negative controls, respectively, for GC protein expression. C. EBV positive Akata and Mutu I BL cell lines were infected with lentiviral vectors expressing a non-targeting (“NT”) control shRNA or Myb-targeting shRNAs (left panel) or with a Myb-expressing or control lentivirus vector (right panel). Three days later, cells were treated with puromycin for 2 days prior to harvesting protein extracts for immunoblot analysis. The effect of Myb knockdown or Myb over-expression in the BL cell lines on the levels of various cellular proteins regulated by Myc expression is shown. The numbers below each immunoblot quantify the results using Image Studio Lite software to normalize the levels of Myb, Myc, CD179B, Cyclin D3, UHRF1, Rag1, and DNMT3B expression to actin expression. Results are presented as the ratio of Myb, Myc, CD179B, Cyclin D3, UHRF1, Rag1, and DNMT3B expression in Myb knockdown (or Myb over-expression) conditions relative to control conditions after being normalized to actin. Values of control conditions are set as 1. D. EBV positive Akata and Mutu I BL cell lines were infected with lentiviral vectors expressing a non-targeting (“NT”) control shRNA or Myb-targeting shRNAs and immunoblots performed to examine expression of Myb, STAT3, and Src as indicated. Akata BL line and the BL5 LCL line serve as positive and negative controls for protein expression. The same actin immunoblot will be seen in Fig 9F.

https://doi.org/10.1371/journal.ppat.1012132.g008

As the transcription factor Myb is expressed at very high levels in the ΔEBNA2 + Myc lymphomas (Fig 5B), and Myb not only serves as a master regulator of GC B cell identity and survival [51,60–62] but is also expressed at very high levels in immature B cells [63–65], we next examined whether Myc activation of Myb is required for regulation of a subset of cellular genes in human BLs. As shown in Fig 8C and 8D, we found that knock-down of Myb expression in the EBV+ Akata and Mutu I BL cell lines inhibits expression of CD179B (IGLL1), cyclin D3, UHRF1, and DNMT3B, although it did not affect expression of IL7R, CD10, TCF3, BACH2, STAT3 and Src. Furthermore, over-expression of Myb in the human EBV+ Akata BL cell line is sufficient to induce expression of CD179B, Rag1 and cyclin D3, although it did not turn on expression of TDT (Fig 8C). Together, these results indicate that Myc expression in human BL cell lines induces activation of some genes indirectly via upregulation of the Myb transcription factor.

Myc inhibits LMP1 expression by decreasing STAT3 activity

Immunohistochemical analysis of the (rare) ΔEBNA2 + Myc lymphomas found to express both high level Myc and LMP1 on immunoblot analysis revealed that tumors are composed of geographically distinct foci containing either high level expression of Myc and Myc-induced target genes (including CD10, TDT, TCL1, and CD179B) along with very low level LMP1 expression, or high level LMP1 expression and LMP1-induced Kappa light chain expression, but low-level expression of Myc and Myc-induced target genes (S6 Fig). The higher-level of Kappa expression in the LMP1-high regions of tumors is consistent with the high level BLIMP1 expression in these tumors (Fig 3A). The finding that high level LMP1 expression and high level Myc expression do not occur in the same cells suggests that high level Myc expression inhibits LMP1 expression and/or that high level LMP1 expression is not tolerated in the tumor cells driven by high level Myc expression.

As LMP1 was reported to autoactivate its own expression in EBV-infected HeLa cells by increasing STAT3 expression [66], the ability of Myc to block LMP1 expression in BL-like lymphomas could potentially be due to Myc inhibition of STAT3 and/or Src gene expression (Fig 6B and S1 Table). Consistent with this possibility, we found that the one ΔEBNA2 + Myc tumor-derived cell line (“R3”) that constitutively expresses LMP1 when grown off the CD40L/IL21 expressing feeder cell layer also has a much higher level of total STAT3, phosphorylated STAT3, and Src in comparison to the other cell lines that do not express LMP1 (Fig 9A). LMP1 expression in the R3 cell line was also associated with a much lower level of Myc expression compared to the other cell lines (including the N3 cell line which was derived from the same tumor) (Fig 9A) as well as a lower level of Myc-induced proteins including Myb, BACH2, TCF3, and CD10 and a much higher level of p100/52 (NF-κB2) (S7 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Activated STAT3 and Src regulate LMP1 expression in BL cell lines.

A. Protein extracts were isolated from stable cell lines derived from Donor 2 ΔEBNA2 EBV plus Myc tumors (grown off the feeder layer) and immunoblot analysis was performed to examine expression of Myc, LMP1, total STAT3, phosphorylated STAT3, and total Src. The EBV negative Akata BL line and a BL5 EBV infected LCL serve as controls. B. Mutu I BL cells were transfected with a vector control, a STAT3 expression vector alone, a Src (constitutively active) expression vector alone, or both STAT3 and Src expression vectors. Two days later protein extracts were harvested and immunoblot analysis performed to examine expression of total STAT3, phosphorylated STAT3, Src, and LMP1 as indicated. C. Mutu I BL cells were infected with lentiviral vectors expressing a non-targeting control shRNA (NT) or STAT3 targeting shRNAs (left panel) or NF-κB2 (p100/52) targeted shRNAs (right panel). Three days later cells were treated with puromycin for 2 days, and then grown on or off an IL40/IL21 expressing feeder cell layer for another 2 days before harvesting protein extracts for immunoblot analysis. The levels of STAT3, LMP1, EBNA2, and p100/52 expression are shown for each condition as indicated. The numbers below each immunoblot in the left panel quantify the results using Image Studio Lite software to normalize the levels of STAT3 and LMP1 expression to actin expression. Results are presented as the ratio of STAT3 and LMP1 expression in STAT3 knockdown conditions relative to control (NT) shRNA, after being normalized to actin. Values of control conditions are set as 1. D. EBV+ Akata BL cells were infected with a lentivirus vector expressing LMP1 or a control lentivirus and two days later cells were treated with puromycin for 3 days. Protein extracts were harvested from puromycin resistant cells and immunoblot analysis performed to examine the levels of LMP1, total STAT3, phosphorylated STAT3, Src, and Myc as indicated. E. Cell lines derived from ΔEBNA2 EBV plus Myc lymphomas or Mutu I BL cells were treated with or without 5’azacitidine for 3 days and then protein extracts harvested for immunoblot to examine expression of LMP1, BZLF1, EBNA2, total STAT3, phosphorylated STAT3 or Src as indicated. F. Extracts isolated from Akata BL cells or Mutu I cells treated with non-targeting shRNA or Myb targeting shRNA (the same extracts used in Fig 8D) were examined by immunoblot analysis for expression of Myb, LMP1, EBNA2 and BZLF1 as indicated. EBV-neg Akata BL cells and a BL5 EBV-infected LCL serve as controls.

https://doi.org/10.1371/journal.ppat.1012132.g009

To confirm that the combination of STAT3 and constitutively active Src is sufficient to induce LMP1 expression in human BL cells with type I latency, Mutu I BL cells were transfected with either a vector control plasmid, a STAT3 vector alone, a constitutively active Src vector alone, or the combination of STAT3 and constitutively active Src vectors (Fig 9B). Although the expression of transfected STAT3 alone was barely detectable due to the low transfection efficiency of Mutu 1 cells, the combination of STAT3 and Src together reproducibly induced LMP1 expression. In addition, we found that culturing Mutu I BL cells for two days on the CD40L/IL21 expressing feeder cell layer is sufficient to reactivate LMP1 expression (in the absence of EBNA2), and that knock-down of STAT3 expression (but not NF-κB2 expression) in Mutu I cells decreases this effect (Fig 9C).

Importantly, we also showed that treatment of the LMP1-expressing ΔEBNA2 + Myc tumor-derived “B1” cell line (derived from Donor 1) with the Src kinase inhibitor, dasatinib, not only strongly inhibits the level of constitutive STAT3 tyrosine 705 phosphorylation but also decreases LMP1 expression (S8 Fig). These results confirm that decreased Src and STAT3 expression in ΔEBNA2 + Myc tumors is an important mechanism by which Myc turns off LMP1 expression. Interestingly, in addition to decreased Src and STAT3 expression, ΔEBNA2 + Myc tumors also have altered expression levels of several other genes predicted to regulate STAT3 activity in vivo [67], including decreased expression of the IL21 receptor and IL21 cytokine, decreased expression of the IL10RA and IL10 cytokine, decreased expression of the IL6 cytokine and IL6R, decreased expression of the JAK1, JAK2 and JAK3 genes, and increased expression of the PTN13 and PTPRE phosphatase genes (S3 Table). We confirmed that loss of Myc expression in the P493-6 cell line increases IL10RA, IL21R, JAK1 and JAK2 protein expression (S9 Fig). Thus, Myc may also decrease STAT3 activity in human BL cells in vivo by decreasing their responsiveness to cytokines, including IL6, IL21 and IL10, and increasing expression of phosphatases that reverse STAT3 phosphorylation [67].

To determine if LMP1 expression at high levels is sufficient to activate STAT3 and Src expression in human BL cells, Akata Burkitt cells were infected with an LMP1 expressing retrovirus or a control vector and immunoblot analysis was performed (Fig 9D). LMP1 expression at high levels induces both total STAT3 and total Src expression, as well as the amount of phosphorylated STAT3, but does not alter the level of Myc expression. Thus, the relative levels of Myc versus LMP1 in B cell lymphomas likely determine whether the activated forms of STAT3 and Src are expressed. Together, these experiments confirm that STAT3 activation is sufficient to induce LMP1 expression in the absence of concomitant EBNA2 expression and strongly suggest that Myc inhibits LMP1 expression in tumors, at least partially, by decreasing expression of STAT3 and Src.

Myc-induced DNA methylation also contributes to loss of LMP1 expression in human BLs

To examine whether Myc repression of LMP1 expression might also be mediated via enhanced DNA methylation of the LMP1 promoters, cell lines derived from the ΔEBNA2 + Myc lymphomas were treated with or without the demethylating agent, 5-azacitadine, and the levels of LMP1 expression were examined by immunoblot analysis (Fig 9E). LMP1 expression was increased by 5-azacitidine treatment, suggesting that enhanced DNA methylation of the proximal (ED-L1) and/or distal (TR) LMP1 promoters in the ΔEBNA2 + Myc lymphoma derived cell lines, as well as in authentic human BLs, helps to repress LMP1 expression. Furthermore, we found that knockdown of Myb expression in both the Akata BL cell line and the Mutu I BL cell line (in which DNMT3B and DNMT1/UHRF1 mediated methylation of the LMP1 promoters has been shown to inhibit LMP1 expression [33]), not only decreased UHRF1 and DNMT3B expression (Fig 8C) but also turned on LMP1 expression (Fig 9F). This result suggests that high level Myb expression in established human BLs contributes to loss of LMP1 expression by increasing DNA methylation of the LMP1 promoters via DNMT3B and UHRF1 activation. Interestingly, Myb knock-down also activated lytic BZLF1 expression in BL cells (Fig 9F), suggesting that Myb also plays a role in repressing lytic viral reactivation. Nevertheless, bisulfite sequence analysis of the proximal ED-L1 LMP1 promoter in the ΔEBNA2 + Myc lymphomas in NSG mice did not reveal CpG DNA methylation, although we could not examine the methylation state of the upstream TR LMP1 promoter due to its high GC content (S10 Fig). Together, these results suggest that in addition to preventing Src expression and STAT3 activation, Myc may further enforce LMP1 silencing in established human BLs by promoting DNA methylation of the LMP1 promoters.

Ethics statement

All animal work experiments were approved by the University of Wisconsin-Madison Institutional Animal Care and Use Committee (IACUC) and conducted in accordance with the NIH Guide for the care and use of laboratory animals (protocol numbers M005197 and M005214).

Plasmids, retroviral vectors and lentiviral vectors

All plasmid DNA was prepared using the Qiagen High-speed Midi-prep kit according to the manufacturer’s instructions. Lentiviral vectors containing shRNA targeting Myb (RHS3979-201768911, RHS3979-201768912, RHS3979-201768914, RHS3979-201768915, RHS3979-201786168, RHS3979-201799542) and STAT3 (RHS3979-201751956, RHS3979-201751957, RHS3979-201751958, RHS3979-201751959, RHS3979-201751960) on a pLKO backbone were obtained from Horizon Discovery (Lafayette, CO, USA). The following plasmids were purchased from Addgene, (Watertown, MA): pMSCV-Myc-GFP was a gift from Scott Lowe (Addgene plasmid #18770; http://n2t.net/addgene:18770; RRID:Addgene 18770), pMSCVpuro-Flag-cMyc T58A was a gift from Juan Belmonte (Addgene plasmid #20076; http://n2t.net/addgene:20076; RRID:Addgene_20076), psPAX2 was a gift from Didier Trono (Addgene plasmid #12260; http://n2t.net/addgene:12260; RRID:Addgene: 12260), phCMV-GALV-MTR was a gift from Daniel Hodson (Addgene plasmid #163612; http://n2t.net/addgene:163612; RRID:Addgene_163612), TFORF2072 (Myb expression vector) was a gift from Feng Zhang (Addgene plasmid #141905; http://n2t.net/addgene:141905; RRID:Addgene_141905), Src-Y530F was a gift from Cheng-han Yu (Addgene plasmid #124659; http://n2t.net/addgene:124659; RRID:Addgene_124659), and CA-Stat3-t2A-mCherry was a gift from Jennifer Mitchell (Addgene plasmid #127540; http://n2t.net/addgene:127540; RRID:Addgene_127540). The plasmid pSG5 was purchased from Stratagene. pSG5-R and pSG5-Z (kind gifts from Diane Hayward of John Hopkins University) contain the BZLF1 (Z) and BRLF1 (R) immediate-early genes driven by the SV40 promoter as previously described [76]. The pCDH-MSCV-MCS-EF1-GFP-puro lentivirus vector is from System Biosciences (Palo Alto, CA). A LMP1-expressing lentiviral vector was constructed by inserting the B95.8 EBV strain LMP1 gene into the pCDH-MSCV-MCS-EF1a-GFP-Puro vector (CD713-LMP1). The pcDNA3 vector is from Invitrogen, (Waltham, MA). pMD-Gag-Pol was the kind gift of Richard Mulligan (Harvard University) via Robert Kalejta (UW-Madison) [77]. The CRISPR/CAS 9 vector pCEP has been described previously [78].

Cell lines and cell culture

The EBV-positive Akata Burkitt lymphoma cell line was originally derived by the Takada lab [79]. The EBV Akata-BX1 Burkitt lymphoma cell line was a gift from Lindsay Hutt-Fletcher and was obtained by super-infecting an EBV-negative clone of Akata Burkitt lymphoma cells with recombinant Akata EBV containing a G418 resistance gene cassette and GFP gene inserted into the EBV BXLF1 gene as previously described) [80]. Type 1 EBV infected Mutu I, originally derived by the Rickinson lab at the University of Birmingham, UK [81], is a Burkitt lymphoma cell line (obtained as a gift from Jeff Sample at Pennsylvania State University). EBV-negative Akata Burkitt lymphoma cells were a kind gift from Kenzo Takada of Hokkaido University, Japan, via Bill Sugden of the University of Wisconsin. The P3HR1 Burkitt lymphoma G668 (Cl 16) cell line was a gift from George Miller at Yale University via Bill Sugden at UW-Madison [82]. All Burkitt lymphoma cell lines were maintained in RPMI media (Gibco) containing 10% fetal bovine serum (FBS), and 1% penicillin-streptomycin (pen/strep, Gibco); Akata-BX1 cells also were treated with 500 μg per ml of G418 (Life Technologies). Lymphoblastoid cell lines (LCLs) were created as previously described [83]; briefly, adult peripheral blood mononuclear cells were isolated by Ficoll gradient and EBV infected LCL lines were obtained by transforming peripheral blood B with EBV. LCLs were grown in RPMI media containing 10% FBS and 1% pen/strep. HeLa (a human cervical carcinoma cell line) was obtained from the American Type Culture Collection (ATCC, Rockville, MD) and grown in DMEM media (Gibco) with 10% FBS and 1% pen/strep. 293 FT cells (ATCC) were maintained in DMEM media (Gibco) with 1mM sodium pyruvate (Gibco), 1X non-essential amino acids (Gibco), 10% FBS, 500 μg per ml of G418, and 1% pen/strep. YK6-CD40lg-IL21 feeder cells were cultured as previously described [38]. Confluent cells were trypsinized and gamma-irradiated to 30 gys, centrifuged and frozen in 10% DMSO/FBS. For use, cells were thawed, diluted with Advanced RPMI (A-RPMI, Gibco) and centrifuged 4 min at 400g. Supernatant was removed and cells resuspended in A-RPMI, 20% FBS, 1% Pen/Strep, and 2mM glutamine. YK6-CD40lg-IL21 were plated at a rate of 10,000 cells/cm² for growth of B-cells in A-RPMI.

Generation of B-cell lines

15x10⁶ commercial cord blood mononuclear cells (Stemcell Technologies #70007.1) were thawed and plated on prepared (irradiated) YK6-CD40lg-IL21 feeder layers and infected the following day with previously prepared ΔEBNA2 virus in a minimal volume of media at an estimated MOI of 3–6 with respect to B-cells. Cells were cultured with media changes as needed, until uninfected cultures died out, with addition of fresh irradiated YK6-CD40lg-IL21 as needed. Once lines were established, they were infected with either the pMSCV-Myc-GFP or pMSCVpuro-Flag-cMyc T58A retroviruses, generated as described below. Of note, cord blood B-cell lines infected with the ΔEBNA2 virus alone are unable to grow off the CD40L/IL21 expressing feeder cell layer and are thus not fully transformed.

Lentivirus and Retrovirus production

For cord blood B-cell infection with retroviral vectors, the retroviral vectors were co-transfected with phCMV-GALV-MTR and pCMV-GAG-Pol vectors into 293FT cells using Lipofectamine 2000 (Thermo Fisher Scientific) as per the manufacturer’s instructions. For Burkitt cell infection with retroviral vectors, the retroviral vectors were co-transfected into 293FT cells with the pMD2.G and pCMV-GAG-Pol vectors. For Burkitt cell infection with lentiviral vectors, the retroviral vectors were co-transfected into 293FT cells with the pMD2.G and pSPAX2 vectors. The media was changed to RPMI, 10% FBS, 1X Pen-Strep the following day and lentivirus/retrovirus was then harvested at 48 hours post-transfection and 0.8 μm filtered.

Retroviral/lentiviral infection of cord blood B cell lines

Once ΔEBNA2 EBV-infected B cell lines were established, they were infected with either the pMSCV-Myc-GFP or pMSCVpuro-Flag-cMyc T58A retroviruses (pseudotyped with the GalV-MTR coat protein in the phCMV-GALV-MTR vector), with polybrene added to the retrovirus suspension at 10 μg/ml. B cells were removed from the feeder layer by shaking, and then 2ml of the released cells were pelleted at 400g for 1 minute, The supernatant was removed and 2ml retrovirus/polybrene was added. The cells/retrovirus were then centrifuged for 90 minutes at 1400g at room temperature, and then the cells and retrovirus were added to a T-25 flask containing irradiated YK6-CD40lg-IL21 cells, and additional A-RPMI to 12.5 ml added. The cells were cultured for 2–3 weeks in the absence of any drug selection to allow outgrowth of infected cells.

Lentiviral infection of Burkitt lymphoma cell lines

Burkitt lymphoma cells were infected with lentiviruses by adding polybrene to the lentivirus containing media at 10 μg/ml and then centrifuging BL cells with the media for 90 minutes at 1400g, room temperature. After infecting cells for 3 days, puromycin (InVivoGen, San Diego, CA) was added to the media at a concentration of 1 μg/ml and cells were harvested for immunoblot analysis 2–3 days later.

Creation of the EBV EBNA2 mutant virus

A GFP/G418R cassette was inserted into the non-essential BXLF1 gene of the AG876 type 2 EBV strain genome in the AG876 BL cell line by homologous recombination using G418 selection, and then stably infected HeLa cell clones containing this virus were derived as previously described [84]. The CRISPR/CAS9 technique was used in the AG876 virus-infected HeLa cells to create a mutation in the EBNA2 gene that puts the reading sequence out-of-frame at EBNA2 residue 25. A pCEP-based vector expressing the blasticidin resistance gene, CAS9 and an insertion site for guide RNAs was created as previously described [78,85]. The EBNA2 gene-directed guide sequence 5’-GTGGGAATTACGGAGTGA-3’was cloned into the pCEP-blasticidin vector, and 100 ng of vector was transfected into AG876 EBV infected Hela cells (in a 35 mm dish) using Lipofectamine 2000 following the manufacturer’s protocol. Twenty-four hours after transfection cells were diluted into a 100 mm dish and treated with blasticidin (Fisher/Invitrogen) at a concentration of 7.5 μg/ml. After one week of blasticidin selection, surviving cells were cultured at low confluence and individual clones were picked and grown in 24 well plates. DNA was isolated from a portion of cells in each clone using the DNAeasy Kit (Qiagen) per the manufacturer’s instructions. The EBNA2 gene sequence surrounding the guide RNA was amplified using 2x Q5 Hot-Start mastermix (New England Biolabs) and the primers 5’-GCGTGTGACGTGGTGTAAAG-3’ and 5’-AGTGTTTTCCCCGACAACGA-3’. The resulting PCR product was purified using QIAquick PCR purification Kit (Qiagen) and sequenced by GeneWiz (South Plainfield NJ). HeLa cell clones containing only the desired EBNA2 mutation and no detectable wild-type AG876 EBNA2 sequence were transfected with expression vectors for the EBV immediate-early proteins, SG5-BZLF1 and SG5-BRLF1, to reactivate lytic EBV infection in 10cm dishes. The following day, the cells were expanded to 15 cm plates and virus production allowed for 3 days. The media was then filtered through a 0.8 μm filter and the virus concentrated by ultracentrifugation in an SW28 rotor at 12,000 RPM, 4°C for 3 hours. The supernatant was decanted and the virus resuspended overnight in RPMI, then aliquoted and stored in liquid nitrogen. The DNA sequence and EBNA2 protein sequence produced by the CRISPR/CAS9 mutagenesis is shown in Fig 1A.

Infection of NSG mice with EBV-positive Cord blood cell lines and tumor analysis

To generate tumors, B-cells infected with the ΔEBNA2 virus or the ΔEBNA2 virus with c-Myc retrovirus were expanded on feeder layers, and then cells were removed from the feeder layer, concentrated and 10x10⁶ cells/mouse were resuspended in 100 μL ice cold PBS and 100 μL Geltrex (Gibco), then injected into the subcutaneous space on the right flank of NSG mice. Mice were monitored weekly to twice weekly for tumor development. When tumors reached 10mm+ size, mice were euthanized with CO₂, and if the tumors were to be cultured, the flank of the mouse sterilized with 70% ethanol, the tumor harvested sterilely, and a portion transferred to A-RPMI, with the balance of the tumor harvested, split and preserved in both 10% formalin (for histology) and frozen on dry ice (for western analysis/RNA-seq) followed by storage at -80°C. Survival data were analyzed by Kruskal-Wallis method using program M-Stat 7.0 (University of Wisconsin).

Isolation of stable tumor-derived cell lines

To culture tumor cells, the tumor was dissociated by trituration through 10- and 5-ml pipets, and the non-dissociated fragments were allowed to settle. The supernatant containing the released cells was transferred to a fresh tube, centrifuged at 400g for 4 minutes at room temperature, the supernatant was removed, and the cells were resuspended in fresh A-RPMI media. Cells isolated from primary tumors were initially cultured for 5 days on the irradiated YK6-CD40lg-IL21 feeder layer and then reintroduced as above into NSG mice. Tumors resulting from these injections were cultured as above, and in some cases were cultured on feeder layers to amplify the cell number before adapting cells to feeder-free growth.

Bulk RNA-seq analysis of Tumors

Bulk RNA-seq libraries were prepared as previously described [86]. Briefly, tumor cell lysates were prepared in TRIzol (Invitrogen) after powdering the tissue in liquid Nitrogen from tissue preserved on dry ice/stored at -80°C. RNA was isolated using the Direct-zol RNA MiniPrep Kit (Zymo Research) and RNA quality was assessed using an Agilent TapeStation. Sequencing libraries were prepared using the TruSeq Stranded mRNA kit (Illumina, San Diego, CA) (polyA enrichment) and sequencing on an Illumina NovaSeq 6000 with 150-bp paired-end reads was performed by the University of Wisconsin Biotechnology Center (Madison, WI). RNA-seq analysis of host transcription was conducted by BioInfoRx (Madison, WI) as previously described [86]. Briefly, fastQC was used to verify raw data quality of the Illumina reads. Reads were aligned to the GRCh38 (hg38) human genome primary assembly using Subjunc aligner from Subread [87] and assigned to genes using Ensembl annotation (v93). Raw counts were normalized using the TMM normalization method [88] using edgeR and the normalized gene counts were transformed to log2 scale using the voom method from the R Limma package [89], then used for differential expression analysis. Immunoglobulin light chain clonotypes were determined from RNA-seq results as previously described [86]. Functional interpretation of the differentially expressed genes was conducted based on GO terms, KEGG pathway and GSEA [90,91] methods. On RNA-seq analysis, one tumor infected with “ΔEBNA2 virus alone” was discovered to have a small amount of contamination with cells from the same donor that had been infected with ΔEBNA2 virus plus a retroviral vector expressing Myc and Bcl2. The RNA-seq results obtained from this tumor were removed from the analysis comparing the RNA-seq results of Myc positive versus Myc negative tumors. However, since the amount of Myc protein contamination in the tumor was extremely low (see Myc expression in Tumor 2 in Fig 3A immunoblot) and only three of the “ΔEBNA2 only” tumors were large enough to obtain sufficient protein for immunoblot analysis, this “ΔEBNA2 only” tumor was included in some immunoblots shown.

EBV gene expression analysis

The RNA-seq results obtained from the ΔEBNA2 virus plus Myc tumor 8 were removed from the analysis since this tumor expressed LMP1 (Fig 5A). Raw RNA-seq reads were aligned to a combined human genome (human release 32, GRCh38.p13, obtained from Gencode) and the AG876 Type 2 genome (GenBank accession number NC_009334) using STAR 2.7.6a [89]. Transcript quantification was computed using RSEM v1.3.1 [92]. This process utilized gencode v38 comprehensive gene annotations and GenBank AG876 annotation (accession number NC_009334), incorporating parameters—estimate-rspd and—strandedness reverse. The heatmap of viral transcripts was constructed by using the TPM (transcripts per million) values estimated by RSEM. Hierarchical clustering of the data was computed and visualized using the Euclidean distance method with the ComplexHeatmap (v2.8.0) R package [93].

p53 sequence analysis

To evaluate variants, RNA-seq read files were processed through nf-core [94] rnavar release 1.0.0 with STAR 2-7-10a [95] and SnpEff 5.1d [96]. Additional variant annotations were added from Clinvar [97] using SnpSift [98]. Variants passing quality and impact filters were considered for further evaluation. No high impact (SnpEff) or Pathogenic (ClinVar) variants were found in TP53.

Drug treatments

5-Aza-2’-deoxycytidine was purchased from Sigma (#A3565), dissolved in DMSO, and used at 5 uM. Doxycycline was purchased from Sigma (#D5207), dissolved in water, and used at 20 ng/mL of media. IL7 was purchased from PeproTech (#200–07), dissolved in 0.1% bovine serum albumin plus water, and used at 5 ng/mL of media. The Src kinase inhibitor dasatinib was purchased from Selleckchem (#S1021), dissolved in DMSO, and used at 1 uM.

Immunoblots

Immunoblots were performed as previously described [99]. Briefly, cell lysates or tumors were combined with SUMO lysis buffer with protease inhibitors (cOmplete, Roche) and sonicated using the Q700 sonicator from QSonica. Quantitation of protein concentration was conducted with a SUMO protein assay kit (Biorad). The lysates were separated using a polyacrylamide gel and then transferred onto a nitrocellulose membrane. The membranes were subsequently blocked with 5% milk consisting of 0.1% Tween 20 and 1X PBS for one hour. Membranes were then incubated with primary antibody overnight. The following day the antibodies were removed and the membrane was washed with wash buffer (1X PBS, 0.1% Tween 20) three times for 5 minutes. The membrane was then incubated with secondary antibody suspended in 5% milk for one hour, before washing with wash buffer three times for 10 minutes before treatment with ECL (Thermo-Fisher) and imaging. Image Studio Lite software was used to quantify levels of proteins of interest relative to loading controls tubulin or actin in certain figures.

Antibodies

The following antibodies were used for immunoblot analyses in this study: anti-R rabbit polyclonal antibody directed against the R peptide (peptide sequence EDPDEETSSQAVKALREMAD, anti-BACH2 (Cell Signaling #80775), anti-BCL6 (Cell Signaling #14895), anti-BHRF1 (Millipore #MAB8188), anti-BIM (Cell Signaling #2933), anti-BLIMP1 (Cell Signaling #9115), anti-BMRF1 (Millipore #MAB8186), anti-BZLF1 (Santa Cruz #sc-53904), anti-CD10 (Abcam #ab227659), anti-CD19 (Abcam #ab134114), anti-CD30 (Invitrogen #MA536219), anti-CD179b (Invitrogen #702901), anti-Cyclin D2 (Cell Signaling #3741), anti-Cyclin D3 (Cell Signaling #2936), anti-DNMT1 (Abcam #ab188453), anti-DNMT3B (Santa Cruz #sc-376043), anti-EBNA1 (Santa Cruz #sc-81581), anti-EBNA2 (Abcam #ab90543), anti-EBNA3A (Exalpha #F115P), anti-GCSAM (Cell Signaling #20289), anti-GFP (Santa Cruz #sc-9996), anti-IL7R (Santa Cruz #sc-514445), anti-IL10Rα (R&D #MAB2742), anti-IL21R (R&D #MAB991), anti-IRF4 (Santa Cruz #sc-56713), anti-JAK1 (Cell Signaling #3332), anti-JAK2 (Cell Signaling #3230), anti-LMP1 (Abcam #ab78113), anti-c-Myb (Cell Signaling #12319), anti-c-Myc (Abcam #ab32072), anti-p100/52 (Cell Signaling #3017), anti-RAG1 (Cell Signaling #3968), anti-SOCS1 (Cell Signaling #55313), anti-SRC (Cell Signaling #2109), anti-phospho-SRC (Tyr416) (Cell Signaling #2101), anti-STAT3 (Cell Signaling #4904), anti-phospho-STAT3 (Y705) (Cell Signaling #9145), anti-phospho-STAT5 (Y694) (Cell Signaling #9359), anti-TCF3 (E2A) (Cell Signaling #4865), anti-TCL1 (Cell Signaling #4042), anti-TDT (Invitrogen #14-9739-82), anti-UHRF1 (Cell Signaling #12387), anti-Tubulin (Sigma #T5168), and anti-Actin (Sigma #A5441). The secondary antibodies used were horseradish peroxide (HRP)-labeled goat anti-mouse antibody (Fisher #G-21040), HRP-labeled donkey anti-goat antibody (Fisher #A16005), HRP-labeled goat anti-rabbit antibody (Fisher #G-21234), HRP-labeled donkey anti-sheep antibody (Fisher #A16047), HRP-labeled goat anti-mouse light chain specific antibody (Millipore #AP200P), and HRP-labeled goat anti-rat light-chain specific antibody (Millipore #AP202P).

Transient transfections

DNA was transfected into Mutu I and Akata BX1 cells by nucleofection using the Amaxa Nucleofector 2b device (Lonza) and program G-016 (with Buffer V) with 1–2 μg DNA per condition.

LMP1 promoter DNA methylation analysis

DNA was isolated from tumor samples in Trizol (Invitrogen) per manufacturer’s instructions and treated with bisulfite using the EZ DNA Methylation-Direct Kit (Zymo Research, Irvine CA). DNA was amplified using 2x Q5U (New England Biolabs) and the following primers: 5’-TTAGGGTAGTGTGTTAGGAGTAAGG-3’ and 5’-CCCTCCACTTTTTTCCAAAAATA-3’. The PCR program was as follows: 98°C for 30 sec, followed by 35 cycles of 98°C for 10 sec, 62°C for 20 sec, and 72°C for 1 min, followed by a 5 minute polishing step at 72°C. PCR products were purified using the Qiaquick kit (Qiagen) and A-tailed for 30 minutes at 72°C using 5U Taq polymerase (Promega), 200 μM dATP (New England Biolabs),1x go Taq Flexi Buffer (Promega), and 400uM MgCl2 (Promega), and then ligated overnight into a pGEM-T easy vector, transformed into DH5α cells (New England Biolabs), and plated on Ampicillin/X-gal plates. Colonies bearing the appropriate insert were identified by EcoRI digest of Qiaprep Miniprep Kit (Qiagen) DNA and submitted for sequencing by GeneWiz (Burlington, MA).