An ATM/Chk2-mediated DNA damage-responsive signaling pathway suppresses Epstein-Barr virus transformation of primary human B cells

Abstract

Epstein-Barr virus (EBV), an oncogenic herpesvirus that causes human malignancies, infects and immortalizes primary human B cells in vitro into indefinitely proliferating lymphoblastoid cell lines, which represent a model for EBV-induced tumorigenesis. The immortalization efficiency is very low, suggesting that an innate tumor suppressor mechanism is operative. We identify the DNA damage response (DDR) as a major component of the underlying tumor suppressor mechanism. EBV-induced DDR activation was not due to lytic viral replication, nor did the DDR marks colocalize with latent episomes. Rather, a transient period of EBV-induced hyperproliferation correlated with DDR activation. Inhibition of the DDR kinases ATM and Chk2 markedly increased transformation efficiency of primary B cells. Further, the viral latent oncoprotein EBNA3C was required to attenuate the EBV-induced DDR. We propose that heightened oncogenic activity in early cell divisions activates a growth-suppressive DDR that is attenuated by viral latency products to induce cell immortalization.

Figure 1. EBV induced a DNA damage response in primary B cells

(A) Indirect immunofluorescence (IF) images of EBNA-LP (green) and γ-H2AX (red) in uninfected B cells, B cells 4 and 7 days after infection with EBV B95-8 (MOI ~5), the recently derived LCL EF3D, and uninfected γ-irradiated B cells (0.2, 1, and 5 Gy, 1h). DNA is stained with DAPI. These images are representative of infections in five different normal donors. (B) Ser1981 phosphorylated ATM (pATM, red) in uninfected B cells and B cells 7 days after EBV B95-8 infection. EBNA-LP or other EBV latent antigen staining was not possible in these samples due to antibody source, however we know that ~50% of these infected cells are EBNA-LP-positive. (C) EBNA-LP (green) and 53BP1 (red) in uninfected B cells and B cells 7 days after infection. (D) EBNA-LP (green) and γ-H2AX (red) in B cells 4 days after infection with UV-inactivated EBV B95-8 (UV EBV) or the non-transforming EBV strain P3HR1 (P3).

Figure 2. EBV induced a period of hyper-proliferative early after infection that was associated with activation of the DNA damage response

(A) Histograms show CD19+ B cell division as measured by CFSE staining at different days after EBV infection. Mock, mock infected cells. (B) The mean division number based on precursor cohort analysis for EBV-infected B cells is plotted at different times post infection. Vertical dashed lines estimate the hyper-proliferation period. Data are presented from 5 normal donors. (C) IF of γ-H2AX (red) and EBNA-LP (green) of uninfected cells, infected cells that have yet to divide (PD0), infected cells after 1 or 2 divisions (PD1-2), or 7 or more divisions (PD7+) and LCLs. DNA is stained with DAPI. (D) The percentage of EBNA-LP positive cells with γ-H2AX signal >5X over background is graphed from uninfected B cells, sorted PDs, and LCLs. Uninfected B cells following 0.2, 1, and 5 Gy (1hr) γ-irradiation are also shown as a positive control. These data are representative of similar experiments from three independent normal donors. (E) Immunoblot of p-Chk2 Thr68 and Chk2 in sorted cells as in (D) including an LCL following 5 Gy γ-irradiation (1hr). (F) The percentage of EBNA-LP positive cells containing 4 or more 53BP1 foci per cell in sorted populations as in (D) are shown along with uninfected irradiated B cell controls. PD3-4 contained significantly more 53BP1 foci per cell than uninfected B cells (p<0.0001), PD0 (p<0.0001), PD7 (p<0.01), and LCL (p<0.0001).

Figure 3. Transcriptional changes correlate with an EBV-induced early period of hyper-proliferation and DNA damage response followed by attenuation upon LCL outgrowth

(A) Heatmap of average expression data across four normal donors for the gene ontology (GO) category “Cell Proliferation” in uninfected resting B cells (B cell), EBV-infected early proliferating B cells (Prolif), and monoclonal LCLs (LCL). The genes presented were derived from GATHER analysis of all genes with significant expression changes (2-way ANOVA, p<0.01) where the expression level increased from B cell to Prolif at least 1.5-fold and decreased from Prolif to LCL at least 1.2-fold (left). Heatmap of individual samples of top 20 “Cell Proliferation” genes (right). (B) Heatmap of “Response to DNA Damage Stimulus” GO genes across individual samples. (C) Gene Set Enrichment Analysis (GSEA) of known DNA damage induced ATM and p53-depending genes in the context of B-Prolif-LCL expression data. The reference list of ATM/p53 target genes was derived from Clusters 2 and 3 of (Elkon et al., 2005) and compared with a pre-ranked list (by fold) of global average gene expression changes from B cell to Prolif (left) and Prolif to LCL (right). Statistical scores are inset into the top right of analysis images (NES: Normalized enrichment score and FWER: Familywise error rate).

Figure 4. Growth suppression and DNA damage enrichment in early cell divisions

(A) A flowchart shows the separation of arrested and proliferating EBV-infected CD19+ PBMC used for IF and FACS. PBMC were first infected with EBV and labeled with the red fluorescent dye PKH26. (B) Then, 8 days post infection, proliferating CD19+ B cells were sorted for PD1-4 and PD6+ based on PKH26 intensity, labeled with CFSE, and cultured for two days. (C) Sorted cells were then analyzed in a FACS-based growth assay where cells in the CFSE^low population were considered proliferating and cells in the CFSE^hi population were considered arrested. Forward scatter (FSC) low reflects dying cells. Results are representative of three normal donors. (D) PKH26^low (PD1-4) cells were subsequently labeled with CFSE as above, sorted after 48h in culture into CFSE^hi (arrested) and CFSE^low (proliferating) populations, and analyzed by IF for γ-H2AX (red).

Figure 5. Inhibition of ATM and Chk2 kinases increased EBV transformation efficiency and proliferation of B cells during a critical period 4-8 days post infection

(A) Quantification of EBV-induced B cell outgrowth following PBMC infection in the presence of 0.1% DMSO (black), 2 μM ATMi (green), or 5 μM ATMi (red). The percentages of wells positive for LCLs at five weeks post infection are plotted relative to the transforming units (TU) of B95-8 virus per well. Results shown are the average of experiments with at least four independent normal donors. Error bars represent standard error of the mean (SEM). (B) Similar experiments were performed using DMSO (black), 2 μM Chk2i (green), or 5 μM Chk2i (red). (C) CFSE-stained PBMC were infected with EBV in the presence of increasing amounts of ATMi or Chk2i (DMSO, 1 μM, 2 μM, 5 μM, and 10 μM). The percentage of CD19⁺/CFSE^low cells of total PBMCs at 14 days post infection are plotted. The data shown are the average values from two different donors +/-SEM. These data are representative of more than five independent experiments. (D) Dot plots show CFSE-and CD19-stained PBMC that were treated with DMSO, 5 μM ATMi, 5 μM Chk2i, or infected with EBV for six days. (E) This table summarizes when ATM and Chk2 suppressed EBV-mediated proliferation at different times following infection. CFSE-stained PBMC were infected with EBV at day 0. ATMi or Chk2i (5 μM) was added at different times after infection (top) or at day 0 and washed out at different times after infection (bottom). EBV-mediated B cell proliferation was detected by FACS at day 14 post infection using CD19-PE and CFSE as in (C). A more than two-fold increase in treated cells versus DMSO is represented by a green plus; a less than two-fold increase is represented by a yellow plus; and no increase is represented by a red dash. The lines indicate the period of incubation and are colored with the proliferation phenotype after ATMi and Chk2i treatment. Average values from two independent donors are shown. (F) EBV-induced outgrowth following PBMC infection was measured as in (A) in the presence of 5 μM Chk2i (blue), 5 μM ATMi (red), or DMSO (black) added at day 0, day 4 or day 12 after EBV infection. Results shown are the average of four independent normal donors +/-SEM. (G) Efficiency of EBV outgrowth from (F) was calculated and the average ratio of inhibitor-treated to DMSO-treated infections +/-SEM for four normal donors is plotted.

Figure 6. EBV latency and consequential host gene expression changes from initial B cell proliferation through LCL outgrowth

(A) Expression of Wp (filled triangles) and Cp (open circles) derived mRNAs in EBV-infected cells sorted by population doubling (PD) and monoclonal LCLs. Relative mRNA abundance normalized to a beta-actin control is plotted versus PD. These data are representative of two normal donors and consistent with published time course experiments (Woisetschlaeger et al., 1989; Woisetschlaeger et al., 1990). (B) The ratio of Wp to Cp mRNA expression levels from (A) is plotted versus division and through LCL outgrowth. (C) Protein expression of EBNA-LP, EBNA2, EBNA3A, and EBNA3C are shown from sorted infected PDs and a polyclonal LCL from the same donor. (D) Proteins detect by Western blotting from three independent normal donors similar to those in panel (C) were quantified. The average ratio of total EBNA-LP protein (i.e., all isoforms) relative to total EBNA3A or EBNA3C +/-SEM is plotted versus PD through LCL. (E) Average CD23 surface expression as mean fluorescence intensity (MFI) is plotted versus PD +/-SEM for two donors. (F, left) The expression level of c-Myc mRNA is plotted versus sorted PD. (F, right) The activity of the c-Myc target gene expression signature (Bild et al., 2006) is plotted from the average expression of targets in microarray samples from four independent donors of resting B cells (B), early proliferating B cells (Prolif), and monoclonal LCLs. Error bars represent SEM.

Figure 7. EBNA3C attenuates the EBV-induced DNA damage response

(A) Representative IF images are shown of γ-H2AX staining (red) from WT, EBNA3A KO (Δ3A), and EBNA3C KO (Δ3C) infected and sorted PD1-4 B cells. DAPI DNA stained (blue) and DAPI/γ-H2AX merged images are also shown. (B) Representative IF images are shown of 53BP1 staining (red) from WT, Δ3A, and Δ3C infected and sorted PD1-4 B cells. (C) Quantification of IF data from (A) is plotted as percentage γ-H2AX positive cells. Average values are plotted for infected cells, uninfected B cells, and 5 Gy γ-irradiated B cells. (D) Model for EBV-induced DDR/hyper proliferative period and attenuation during LCL outgrowth. Early in infection EBNA2 and EBNA-LP associate with cellular transcription factors (TF) to potently up-regulate expression of growth control genes and B cell activation markers, including c-Myc and CD23, activating the host DNA damage response (left). Later in infection, the activity of the EBNA3 proteins, in particular EBNA3C, down-regulate EBNA2 function, as LMP1 and LMP2 are up-regulated and may cooperate in the constitutive, but attenuated expression of host growth control genes and enhanced cell survival (right).