Abstract Introduction. Enhancer of Zeste Homolog 2 (EZH2), which trimethylates Histone 3 at lysine 27 (H3K27me3), is frequently affected by somatic gain-of-function (GOF) mutations in follicular lymphoma (FL) and germinal center B-cell–like diffuse large B-cell lymphoma (GCB-DLBCL). Tazemetostat, a highly specific EZH2 inhibitor (EZH2i), is FDA-approved for relapsed or refractory FL with EZH2 mutation but has shown only limited efficacy in DLBCL. It has been reported that EZH2i synergizes with inhibitors of B-cell receptor (BCR) signaling in DLBCL. Since GCB-DLBCL cells depend on tonic (antigen-independent) BCR signaling, which exclusively triggers the PI3K/AKT pathway, we investigated how EZH2i and EZH2 GOF affect tonic BCR signaling. Methods. We used a spectrum of 10 GCB-DLBCL lines, some modified to allow comparison of unmutated and GOF-mutated EZH2 (Y646S or Y646F). We used CRISPR-Cas9 based technology for knock in (KI) and knock out (KO) and Sleeping Beauty transposon vectors for overexpression. A genetically encoded Förster resonance energy transfer-based reporter was used to measure AKT activity. Key findings were evaluated using primary normal B-lymphocytes and a mouse model having EZH2 GOF mutation. Results. Sensitivity of GCB-DLBCL cell lines to the EZH2i tazemetostat was highly variable, but correlated with the degree of their dependence on BCR signaling. EZH2i treatment produced dose-related increases in phosphorylation of multiple BCR signaling mediators. The most EZH2i resistant cell lines have the highest dependence on BCR signaling, and the most prominent BCR signaling activation following EZH2i. Moreover, BCR KO sensitized the highly BCR dependent SUDHL-6 line to EZH2i, but not the BCR-independent line DB. KI of a Y188F mutation into the BCR co-receptor protein CD79A, known to block tonic BCR signaling, sensitized SUDHL-6 cells highly to EZH2i; this showed that EZH2i resistance was specifically mediated by tonic BCR signaling. As expected, H3K27me3 was increased by expression of EZH2 GOF mutations, and reduced or eliminated by EZH2i. Importantly, expression of EZH2 GOF mutations led to robust increases in AKT activity and coincided with increased H3K27me3. As the likely cause for this increase in AKT activity, the abundance of PTEN protein was decreased by induction of EZH2 Y646S, associated with a significant increase of PTEN promoter methylation. Additionally, using PTEN KO and overexpression models, we detected negative feedback between AKT and EZH2 activity. AKT activation decreased EZH2 expression and the level of H3K27me3, both mediated at least partially by mTOR. Overexpression of Y646S EZH2 increased H3K27me3 and decreased PTEN protein also in normal human B cells. Using a mouse model with Ezh2 Y641F mutant allele for conditional expression in CD19+ B cells, we confirmed our observations in vivo. CD19+ B cells of mutant mice displayed increased pAKT levels associated with decreased PTEN. In vitro, we detected broad synergism of EZH2i with FDA-approved or clinically tested inhibitors of BCR signal-mediating kinases SYK, BTK, and PI3Kδ. Mechanistically, EZH2i sensitized cells to AKT inhibition (direct or mediated by upstream BCR signal inhibition). Therefore, the mechanism of detected synergy is at least partially driven by the capability of EZH2i to increase dependence of cells on AKT activity and consequent increased sensitivity to its inhibition. Conclusion. Reciprocal negative relationships affecting EZH2i resistance were found between EZH2, PTEN, and AKT, comprising a regulatory network with contributions from BCR signaling. GOF mutations increased the quantitative effect of EZH2 but were not essential to the network. Our findings provide further rationale for combining EZH2 inhibition with inhibitors of other targets, such as BCR signaling mediators, in treating lymphoma.First two authors contributed equally. Supported by GACR (20-01969Y), MHCR (DRO - VFN00064165), National Institute for Cancer Research (EXCELES - LX22NPO5102), and MEYSCR (Cooperatio, SVV 260637).
Goverdhan et al. (Mon,) studied this question.