EBNA2 Drives Formation of New Chromosome Binding Sites and Target Genes for B-Cell Master Regulatory Transcription Factors RBP-jκ and EBF1

Abstract

Epstein-Barr Virus (EBV) transforms resting B-lymphocytes into proliferating lymphoblasts to establish latent infections that can give rise to malignancies. We show here that EBV-encoded transcriptional regulator EBNA2 drives the cooperative and combinatorial genome-wide binding of two master regulators of B-cell fate, namely EBF1 and RBP-jκ. Previous studies suggest that these B-cell factors are statically bound to target gene promoters. In contrast, we found that EBNA2 induces the formation of new binding for both RBP-jκ and EBF1, many of which are in close physical proximity in the cellular and viral genome. These newly induced binding sites co-occupied by EBNA2-EBF1-RBP-jκ correlate strongly with transcriptional activation of linked genes that are important for B-lymphoblast function. Conditional expression or repression of EBNA2 leads to a rapid alteration in RBP-jκ and EBF1 binding. Biochemical and shRNA depletion studies provide evidence for cooperative assembly at co-occupied sites. These findings reveal that EBNA2 facilitate combinatorial interactions to induce new patterns of transcription factor occupancy and gene programming necessary to drive B-lymphoblast growth and survival.

Conflict of interest statement

I have read the journal's policy and the authors of this manuscript have the following competing interests: PML is a founder of a company with interest in EBV associated disease. This does not alter our adherence to all PLOS policies on sharing data and materials.

Figures

Fig 1. Differential binding of transcription factors on EBV genomes in type I and type III latency.

(A) ChIP-Seq tracks mapped to EBV genome in LCL (red) or Mutu I (blue) for EBF1 or RBP-jκ as indicated. EBNA2 ChIP-Seq for LCL (black). Y-axis represent raw read counts. (B) ChIP-qPCR for EBF1 or RBP-jκ in LCL (red) or Mutu I (blue) at various EBV genome regulatory regions as indicated and Actin genomic region as negative control. (C) Same as in B, except for isogenic EBV positive BL cells Kem III (red) or Kem I (blue). Asterisk indicates p < 0.05. (D) Western blot of Mutu I, LCL, Kem I, and Kem III probed with antibody to EBF1, RBP-jκ, LMP1, and EBNA2, with cellular loading controls for GAPDH and H3K4me3, as indicated. (E) RT-qPCR for EBV type III gene expression in LCL or Mutu I (left panel) or Kem I and Kem III (right panel).

Fig 2. Differential binding of transcription factors on cellular genome in type I and type III infected B-lymphocytes.

(A) Venn diagram showing EBF1 and RBP-jκ occupancy sites specific to LCL (red) and Mutu I (blue) or common to both cell types. (B) Scatter plot showing the distribution of EBF1 (left) or RBP-jκ (right) occupancies that are specific to either LCL (L, red) or Mutu I (M, green) or common to both cell types (grey). (C) Heatmaps showing co-occupancy of EBF1 and RBP-jκ from either LCL (L) or Mutu I (M) cells, centered on either EBF1 in LCL (left panel) or RBP-jκ in LCL (right panel). Binding sites were clustered based on their common occurrence in MutuI and LCL (a), LCL-specific (b), or Mutu I-specific (c). (D and F) ChIP-qPCR validation for various cellular gene ChIP-Seq peaks for EBF1 (left) or RBP-jκ (right) in either Mutu I (blue) or LCL (red) cells. (E) Same as in (D) except for Kem I and Kem III cell types. (G) Same as in (F) except for Kem I and Kem III cell types. Asterisk indicates p < 0.05.

Fig 3. Colocalization of EBNA2 with RBP-jκ and EBF1 co-occupied sites.

(A) Heatmap of EBNA2 ChIP-Seq peaks from LCL was compared with EBNA2 peaks from Mutu III, EBF1 and RBP-jκ from LCL or Mutu I, EBNA3C from LCL or Mutu III, or EBNALP from LCL. Peaks were sorted based on the EBNA2 levels in LCL and show -/+ 4kb window. (B) The fold changes (LCL versus Mutu I) were plotted for EBF1 (x-axis) and RBP-jκ (y-axis) for all EBF1 and RBP-jκ peaks. The peaks were colored based on the EBNA2 occupancy. The positive correlation (Pearson’s correlation coefficient = 0.8) between the fold changes suggests that EBF1 and RBP-jκ co-localize with each other. The strong occupancy of EBNA2 at the top-right panel of the plot shows that high occupancy EBNA2 correlates with sites of high occupancies for EBF1 and RBP-jκ. (C) Heatmap comparison of histone modification marks (H3K4me3, H3Km4me1, H3K27me3) and transcription factors BATF, JunD, and EBNA2 from LCL centered at EBF1 peaks enriched in LCL (cluster i) or Mutu I (cluster ii). (D) Box plot quantitation of peak number and intensity for ChIP-Seq clustered sets i and ii shown in panel C as indicated. (E and F) Same as in C and D, except centered at cell-specific RBP-jκ peaks.

Fig 4. Characterization of cellular genes highly activated by EBF1-RBP-jκ-EBNA2 co-occupancy in LCL.

(A) Heatmap of mRNA expression for genes with co-localized EBF1, RBP-jκ, and EBNA2 and at least 3-fold over expression in LCL relative to Mutu I. (B) Percentage of genes with co-localized EBF1, RBP-jκ, and EBNA2 sites at least 10kb from the TSS that are over-expressed in LCLs relative to Mutu I. (C) Predicted regulation network for LCL over-expressed genes co-occupied by EBNA2, EBF1, and RBP-jκ.

Fig 5. EBNA2-dependent transcription factor redistribution in chromatin occupancy.

EREB2.5 cells were treated with (+) or without (-) estradiol (E2) for 24 or 48 hrs and then assayed by ChIP for binding to EBF1 (A and B), RBP-jκ (C and D), or EBNA2 (E and F) at cellular (A, C, E) or EBV genome sites (B, D, F). Actin genomic region (cellular) or Qp (EBV) was used as negative binding control for EBF1, RBP-jκ, ορ EBNA2 ChIP. (G) ChIP binding for CTCF, PU.1, or PAX5 in EREB2.5 cells treated (blue) or untreated (red) with E2 for 48 hrs. PPP1R1B or KCTD17 genomic region was negative binding control PU.1 or PAX5, respectively. Asterisk indicates p < 0.05. (H) Western blot showing protein levels for EBF1, RBP-jκ, EBNA2, Actin, and CTCF in EREB2.5 cells at 24 and 48 hrs after E2 withdrawal.

Fig 6. Cooperative binding and colocalization of EBF1, RBP-jκ, and EBNA2 at DNA regulatory elements.

(A) DNA-affinity binding assays with nuclear extracts from EREB2.5 cells with (+) or without (-) estradiol (E2) using DNA regulatory elements from IL7, HES1 (left), or LMP1, Qp (middle), or TOM1 (right). Input (10%) is indicated, and bound proteins were assayed by Western blot for EBF1, RBP-jκ, EBNA2, and specificity controls for PARP1 and Actin. (B) ChIP-reChIP assays in LCLs with EBF1 as first ChIP, followed by a reChIP with either no antibody control or antibody to EBF1, RBP-jκ or EBNA2. ReChIP DNA was quantified by qPCR at LMP1, IL7, and BDH2 (an EBF1 only site) binding sites. (C) Same as in B, except first ChIP was with RBP-jκ followed by a ReChIP with no antibody control, or EBF1, RBP-jκ, or EBNA2 antibody and assayed at LMP1, IL7, or LZTFL1 (an RBP-jκ only site) binding sites. Asterisk indicates p < 0.05.

Fig 7. EBF1 and RBP-jκ function at EBNA2 co-occupied sites.

(A) Western blots of LCL cells transduced with either control shRNA (shCtrl) or shEBF1 (left) or shCtrl or shRBP-jκ (right) lentivirus and probed for EBF1, RBP-jκ LMP1, EBNA2, EBNA3C, or Actin. (B) Cell viability determined by Resazurin assay for LCL cells 7 days after shRNA lentivirus transduction. (C) RT-qPCR (ΔΔCT) analysis of EBV or cellular gene transcription (as indicated) in LCLs transduced with either shEBF1 (red), shRBP-jκ (green), or shCtrl lentivirus (black). (D-M) ChIP-qPCR in LCLs with antibodies to either EBF1 (panels D, E, J, K), RBP-jκ (panels F, G, H, I), or EBNA2 (panels L and M). LCLs were transduced with either shCtrl (black), shEBF1 (red), or shRBP-jκ (green). Viral (panels D, F, H, J, L), or cellular (E, G, I, K, M) sites were analyzed as indicated. Asterisk indicates p < 0.05.

Fig 8. Model of EBNA2-induced transcription factor redistribution in chromosome occupancy.

EBNA2 is shown to facilitate the formation of new binding sites that are co-occupied by RBP-jκ and EBF1 relative to cells lacking EBNA2.