Epigenetic reprogramming sensitizes immunologically silent EBV+ lymphomas to virus-directed immunotherapy

Abstract

Despite advances in T-cell immunotherapy against Epstein-Barr virus (EBV)-infected lymphomas that express the full EBV latency III program, a critical barrier has been that most EBV+ lymphomas express the latency I program, in which the single Epstein-Barr nuclear antigen (EBNA1) is produced. EBNA1 is poorly immunogenic, enabling tumors to evade immune responses. Using a high-throughput screen, we identified decitabine as a potent inducer of immunogenic EBV antigens, including LMP1, EBNA2, and EBNA3C. Induction occurs at low doses and persists after removal of decitabine. Decitabine treatment of latency I EBV+ Burkitt lymphoma (BL) sensitized cells to lysis by EBV-specific cytotoxic T cells (EBV-CTLs). In latency I BL xenografts, decitabine followed by EBV-CTLs results in T-cell homing to tumors and inhibition of tumor growth. Collectively, these results identify key epigenetic factors required for latency restriction and highlight a novel therapeutic approach to sensitize EBV+ lymphomas to immunotherapy.

Conflict of interest statement

Conflict-of-interest disclosure: L.G.-R. is a consultant for Janssen and ADC Therapeutics. D.P. was at Memorial Sloan Kettering Cancer Center while engaged in this project but is currently employed at GlaxoSmithKline. R.J.O. and E.D. received royalties following licensure of the EBV-specific T-cell bank by Atara Biotherapeutics and have subsequently received research support and consultant fees from Atara Biotherapeutics. The remaining authors declare no competing financial interests.

© 2020 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

High-throughput drug screen identifies pharmacologic agents that induce latency III antigen expression. (A) Immunoblot of BL cell lines to characterize latency. BC2, latency I control; LCL 9001, latency III control; Ramos, EBV− control. (B) Heatmap showing the fold change in LMP1 for 2 replicates across 441 compounds. Dendrogram branches on the right illustrate groupings based on unsupervised clustering, highlighting a cluster of 33 compounds inducing a greater than twofold change in both replicates (blue branches). Inset shows the list of 33 compounds grouped based on similarity of pathway targets. (C) Network plot showing the pathway enrichments based on drug targets. Each node denotes a subpathway, with colors delineating pathway groupings (see table). Nodes with multiple colors denote shared pathway groupings. (D) Focused screen of epigenetic modifying agents. qRT-PCR for LMP1 and Cp promoter transcripts in cells treated with drug vs vehicle control for 48 hours. Data are shown as fold change in treated cells compared with vehicle control. Experiments were performed in duplicate. Drug doses were as follows: GSK-126, 5 μM; EPZ-6438, 5 μM; romidepsin, 0.25 nM; HDAC3i, 5 μM; panobinostat, 100 nM; 5-azacytidine, 4 μM; and decitabine, 1 μM. (E) Combination treatment with panobinostat and decitabine. qRT-PCR for LMP1 and Cp transcripts in cells treated with vehicle, panobinostat alone (100 nM), decitabine alone (1 μM), or combination. Experiments were performed in triplicate. Error bars represent standard error of the mean (SEM). DNMTi, DNA methyltransferase inhibitor; EZH2i, EZH2 inhibitor; FC, fold change; HDAC, histone deacetylase; HDACi, histone deacetylase inhibitor; LTR, long terminal repeat.

Figure 2.

Hypomethylating agents induce immunogenic EBV antigens. (A,C) qRT-PCR for LMP1 and Cp promoter in cells treated with drug (decitabine or 5-azacytidine) vs vehicle control for 48 hours at the following doses listed from left to right: vehicle, 10, 25, 50, 100, 250, 500, and 1000 nM. Data are shown as fold change in treated cells compared with vehicle control. Experiments were performed in triplicate. Error bars represent SEM. (B,D) Immunoblot for viral proteins as indicated. BL cells were incubated with drug at the indicated doses for 48 hours. LCL-9001 is a latency III positive control. BC2 is a latency I control. Ramos is an EBV− BL used as a negative control. Lower panel in B represents a longer exposure time for LMP1. (E-F) Immunohistochemistry for EBNA2 and LMP1 in cell blocks generated from Mutu I, Kem I, and Rael cells treated as indicated. Cells were exposed to 5-azacytidine at 4 µM, decitabine at 500 nM, or vehicle control for 48 hours. Experiments were performed in triplicate. Representative images were obtained on an Olympus BX 43 microscope (Camera, Jenoptik ProgResCF; software, ProgRes Mac Capture Pro, 2013. Original magnification ×600 with a 60/0.80 objective lens). (G-F) Image quantification using HALO (Indica labs). Error bars represent SEM. *P < .05, **P < .01, ***P < .001, ****P < .0001. 5-Aza, 5-azacytidine; DCB, decitabine; ns, not significant.

Figure 3.

Decitabine induces expression of viral antigens in BL xenograft models. (A-B) Immunohistochemistry for EBNA2 and LMP1 in tumors obtained from Mutu I, Kem I, or Rael xenograft mice as indicated. Experiments were performed with 6 mice per condition per cell line for each of the following conditions: vehicle treatment, decitabine 0.5 mg/kg intraperitoneally daily, and decitabine 1 mg/kg intraperitoneally daily. Representative images were obtained on an Olympus BX 43 microscope (Camera, Jenoptik ProgResCF; software, ProgRes Mac Capture Pro, 2013). Original magnification ×600 with a 60/0.80 objective lens. (C-D) Image quantification using HALO (Indica labs). Error bars represent SEM. *P < .05, ** P < .01, ***P < .001, ****P < .0001.

Figure 4.

Decitabine induction of viral antigens persists after removal of drug. (A) qRT-PCR for LMP1 and Cp in cells treated with 250 nM decitabine vs vehicle control for 72 hours and then evaluated after removal of drug at the indicated time points. Data are shown as fold change in treated cells compared with vehicle control. Experiments were performed in triplicate. Error bars represent SEM. (B) Quantification of EBNA2+ cells by IHC in Rael xenograft tumors as indicated. Error bars represent SEM. *P < .05, **P < .01. (C) IHC for EBNA2 on Rael xenograft tumors after treatment with decitabine or vehicle at the specified time points. Mice were treated with decitabine or vehicle control and then evaluated immediately after treatment (n = 4), 4 days after discontinuation of drug (n = 4), or at the time of sacrifice due to progressive tumor (n = 8). Microscope, Olympus BX 43 microscope; camera, Jenoptik ProgResCF; software, ProgRes Mac Capture Pro, 2013. Original magnification ×600 with a 60/0.80 objective lens.

Figure 5.

Global EBV DNA hypomethylation is observed after decitabine treatment in latency I EBV+ BL. (A) EBV genome plot. Latent, lytic, noncoding RNA genes and regulatory regions are color coded as indicated. Capture regions for EpiTYPER MassARRAY methylation analysis are indicated with red numbers. Twenty-eight regions were selected across the genome (1-13 CpGs per region) representing primarily EBV gene promoters. Capture region for Methyl-Capture sequencing is also depicted. (B) Heatmap of quantitative DNA methylation levels as analyzed by EpiTYPER MassARRAY in vehicle- and decitabine-treated cells. (C) Heatmap of methylation of CpGs, n = 1022 from methyl-capture sequencing. Washout, cells treated with drug × 48 hours followed by 7 days of incubation in media without drug.

Figure 6.

Localization of differentially methylated CpGs in decitabine-treated BL cell lines and xenografts. Decitabine- and vehicle-treated cells and xenograft tumors were evaluated with Methyl-Capture sequencing as described above. Differentially methylated areas were mapped to the EBV genome using Integrative Genomics Viewer (Broad Institute; https://software.broadinstitute.org/software/igv). DMCs, differentially methylated cytosines.

Figure 7.

Decitabine treatment results in T-cell–mediated lysis in vitro and T-cell trafficking to tumors in vivo. (A-C) Chromium-release assay in the indicated cell lines incubated with EBV-CTLs reactive to EBNA3C, EBNA3A, or LMP1 as labeled. BL cells were treated with decitabine at 50 nM (Rael) or 250 nM (Mutu I) or vehicle control for 72 hours. Controls are as follows: (A) autologous dendritic cells with A0201 HLA loaded with EBNA3C peptide (positive control) and autologous dendritic cells with A0201 HLA alone (negative control), (B) EBV-transformed autologous BLCL (positive control) and autologous dendritic cells (negative control), and (C) EBV-transformed autologous BLCL (positive control) and autologous phytohemagglutinin-activated blasts (negative control). ** P < .01, ***P < .001, ****P < .0001. (D-E) IHC for EBNA2 and CD8 in xenograft tumors as indicated. Microscope, Olympus BX 43 microscope; camera, Jenoptik ProgResCF; software, ProgRes Mac Capture Pro, 2013. Original magnification ×600 with a 60/0.80 objective lens. (F) Bioluminescence in Rael-luciferase xenografts. *P < .05. (G) IHC for CD8 in Mutu I xenografts in the indicated treatment cohorts. Upon engraftment, Mutu I xenograft mice were randomized to treatment with decitabine at 1 mg/kg daily × 3 days or vehicle control followed by EBV-CTLs twice weekly vs control with 4 mice in each cohort. Mice were humanely sacrificed at the time of tumor growth >2000 mm3 or at day 18 to evaluate for T-cell trafficking to tumor.