Combined Epigenetic Therapy to Induce Latency II/III Antigen Expression in Latency I EBV⁺ Lymphoma

Epstein-Barr virus (EBV) infection contributes to the development of a significant subset of human lymphomas. In EBV⁺ lymphomas, the virus exists in one of three latent states: latency I, in which only EBNA1 viral protein is expressed. Conversely, latency II and III tumors express immunogenic viral proteins such as EBV latent membrane protein 1 or 2 (LMP1/2) and EBV nuclear antigen 2 or 3 (EBNA2/3) that can be recognized by cytotoxic T cells (CTLs). As such, cellular therapy with EBV-CTLs is effective in latency II/III tumors but not in latency I tumors, such as EBV⁺ Burkitt lymphoma (BL).

Previously, we reported that EBV latency restriction is regulated through DNA methyltransferase 1 (DNMT1) (Guo et al, Nat Microbiol 2020), and treatment with decitabine (DCB) induces latency II/III antigen expression in latency I BL, rendering BL tumors sensitive to EBV-specific CTLs (Dalton et al, Blood 2020). We also found that a selective DNMT1 inhibitor (DNMT1i) converts BL to latency II/III, highlighting the key role of DNMT1 in latency regulation. One limitation of DNMT1 inhibition as a therapeutic approach in latency I EBV⁺ lymphoma is that the induction of latency II/III is only observed in a subset of cells. Here, we investigated the mechanisms of resistance to latency conversion to develop rational combinations to maximize the conversion of latency I tumors to latency II/III thereby sensitizing them to EBV-directed immunotherapy.

To determine whether epigenetic mechanisms are responsible for resistance to latency conversion, we performed ATAC-seq and ChIP-qPCR on BL cell line cells sensitive or resistant to DCB-induced latency conversion. We observed minor reduction in chromatin accessibility at the promoter region of EBNA3A (logFC=-0.06, p=4.98E-21) and LMP2A (logFC=-0.13, p=0.002), but no significant differences in other EBV promoter regions, suggesting that chromatin compaction plays a minor role in regulating resistance to DCB-induced latency conversion. Next we performed ChIP-qPCR to evaluate activating (H3K27ac, H3K4me³, H3K4me and H3K36me²) and repressive (H2K27me³ and H3K9me³) marks at Cp (encodes for EBNA2) and LMP1p. We observed an increase in H3K27me³ on Cp and LMP1p in resistant cells. H3K27me³ was increased 2-fold on Cp (adj-p=0.0009) and 2.45-fold on LMP1p (adj-p=0.0086) in DNMT1i resistant versus sensitive cells.

Given our observation of increased H3K27me³ resistant cells, we next evaluated the effect of the EZH2 inhibitor tazemetostat (TAZ), which targets H3K27me³, alone and in combination with DCB or DNMT1i. TAZ treatment alone did not induce expression of latency II/III. However, when combined with DCB or DNMT1i, TAZ increases the percentage of EBNA2⁺ LMP1⁺ cells compared to DCB or DNMT1i alone (DCB vs DCB+TAZ: 32.3% vs 44.5% adj-p=0.0006; DNMT1i vs DNMT1i+TAZ: 22.2% vs 35.6% adj-p=0.0002). To examine this in vivo, we treated mice with DCB and/or TAZ and found that the combination of DCB and TAZ significantly increases the proportion of LMP1⁺ cells (DCB vs DCB+TAZ: 4.58% vs 6.62% adj-p≤0.0001).

Our data suggests that repressive histone mark H3K27me³ contributes to resistance to DNMT1 inhibition-induced latency switch and that the addition of TAZ to DCB or DNMT1i can overcome resistance to latency switch, resulting in a more immunogenic tumor. These findings will guide the development of therapies to improve the treatment of otherwise immune refractory EBV⁺ latency I lymphoma.

Zappasodi:IFLI: Consultancy; BMS: Research Funding; iTEOS Therapeutics: Other: Scientific Advisory Board; Astrazeneca: Research Funding. Roth:Roche: Consultancy; Merck: Consultancy.