EZH2 gain-of-function events distinctly impact germinal center B cells depending on timing in the B lineage

Abstract The two most common types of Non-Hodgkin Lymphoma (NHL) are Diffuse Large B Cell Lymphoma (DLBCL) and Follicular Lymphoma (FL). Up to 30% of germinal center B cell-derived (GC)-DLBCL and FL cases exhibit gain-of-function events in the histone 3 lysine 27 (H3K27) methyltransferase EZH2, including a heterozygous hotspot mutation (EZH2Y641F/+) or overexpression of wild-type protein (EZH2OE). We previously found that the EZH2Y641F/+ mutation not only causes a net gain of H3K27 trimethylation (H3K27me3), a repressive histone mark, but also results in spreading of this mark past canonical boundaries with accompanying changes in gene expression (Souroullas et al., Nature Medicine2016). However, the precise oncogenic mechanism of EZH2Y641F/+ is incompletely understood, including when the mutation occurs in the B lineage to prime GC B cells for transformation. Additionally, the phenotypic and transcriptional impacts of EZH2OE in GC B cells are unknown. By directly comparing EZH2Y641F/+ and EZH2OE at early and late stages in the B lineage, we can delineate their shared and distinct effects and determine the temporal context(s) in which they contribute to B cell lymphoma. We previously crossed conditional alleles for Ezh2Y641F/+ and Ezh2OE with CD19Cre/+, which activates the alleles in pro-B cells, and with Cγ1Cre/+, which activates the alleles in GC B cells. To determine how EZH2Y641F/+ and EZH2OE impact GC cell phenotypes, we evaluated GC population sizes, apoptosis, and proliferation via flow cytometry. Neither early nor late activation of EZH2OE yields a phenotype for GC cells. Compared to CD19Cre/+ control, CD19Cre/+;Ezh2Y641F/+ causes a significant decrease in GC cells due to an increase in apoptosis. In contrast, and in agreement with literature (Béguelin et al., Cancer Cell 2013), Cγ1Cre/+;Ezh2Y641F/+ causes a significant increase in GC cells compared to Cγ1Cre/+ control due to an increase in cell proliferation. Therefore, early and late activation of EZH2Y641F/+ in the B lineage oppositely impact GC cell phenotypes, and only late activation causes phenotypes that support GC transformation to NHL. To interrogate the transcriptional impacts of EZH2Y641F/+ and EZH2OE in GC B cells, we performed RNA sequencing in both dark zone and light zone GC cells. Both Cγ1Cre/+;Ezh2OE and Cγ1Cre/+;Ezh2Y641F/+ exhibit higher EZH2 activity compared to Cγ1Cre/+ control, i.e. downregulation of PRC2 target signatures. In the CD19Cre/+ cohort, these signatures are downregulated in CD19Cre/+;Ezh2OE mice but not CD19Cre/+;Ezh2Y641F/+ mice. Thus, only late activation of EZH2Y641F/+ causes significant repression of canonical PRC2 target signatures, which indicates that the effect of H3K27me3 redistribution may be more pronounced when the mutation is activated prior to the GC stage. The timing of the mutation also determines the regulation of other pathways. For example, extracellular matrix and cell adhesion signatures are upregulated in CD19Cre/+;Ezh2Y641F/+ mice compared to CD19Cre/+ control, while these pathways are downregulated in Cγ1Cre/+;Ezh2Y641F/+ mice compared to Cγ1Cre/+ control. EZH2OE causes minimal differential expression of individual genes compared to Cre-only controls but does significantly impact cumulative gene signatures, including upregulation of cell cycling and DNA damage repair signatures. Altogether, the evidence supports that EZH2Y641F/+ causes distinct transcriptional changes in GC B cells depending on timing, while the impacts of EZH2OE are more subtle.In summary, we identified that the timing of the EZH2Y641F/+ mutation in the B lineage significantly determines its phenotypic and transcriptional impact in GC B cells. We further demonstrate that while EZH2OE phenotypically resembles EZH2WT in GC B cells, it causes subtle transcriptional changes that may enhance the typical functions of EZH2. While EZH2Y641F/+ and EZH2OE both result in increased EZH2 activity, they likely promote transformation of B cells through different mechanisms. Ongoing work involves assessing the cooperation of EZH2Y641F/+ and EZH2OE with other oncogenes (e.g. BCL2) and determining their H3K27me3 patterns in GC B cells. Ultimately, by closely investigating how EZH2Y641F/+ and EZH2OE impact GC B cells, our work may illuminate new strategies to stratify and treat GC-DLBCL and FL patients with EZH2 alterations.