In this issue of Blood, Nguyen et al report that models of germinal center (GC)-derived B-cell lymphoma (BCL) display synthetic vulnerability to combined inhibition of EZH2 and DOT1L.1 The story underlying this investigation began roughly 15 years ago, with the finding that in both follicular lymphoma (FL) and diffuse large BCL of GC phenotype (GCB-DLBCL), mutations in chromatin-modifying genes are especially recurrent. This involvement of epigenetic regulators is perhaps not surprising, given the extensive epigenetic and gene expression changes involved in the normal GC reaction, associated with clonal selection, proliferation, and differentiation into postmitotic memory B cells or plasma cells.2 Furthermore, the fact that most of these mutations involve loss of function is compatible with lymphomagenesis, because it interferes with normal exit from GC differentiation stages, thereby promoting proliferation and acquisition of additional mutations through aberrant somatic hypermutation.
However, loss of function does not suggest obvious or direct therapeutic solutions. EZH2 was the notable exception among recurrently mutated genes of GC-derived BCL, in that its mutations (most commonly affecting a single amino acid residue, Y646, in its SET domain) involve gain of function. EZH2 was already known to function in many cell types as the catalytic subunit of the polycomb repressive complex-2, which trimethylates histone 3 at lysine 27 (H3K27me3), a repressive chromatin mark that overlaps with activating H3K4me3 at promoters of transiently repressed “bivalent” genes. It was subsequently shown that in normal and neoplastic GC B cells, EZH2 marks many additional genes as bivalent3 and cooperates with BCL6, a transcriptional repressor normally expressed in B cells at only the GC stage, in a noncanonical PRC1-BCOR-CBX8 complex.4 These extensive molecular effects of EZH2 are underscored by profound phenotypic effects in mice: GCs do not form in the absence of EZH2, and knockin of the mutation equivalent to Y646 produces GC hyperplasia and accelerates BCL development,3 associated with H3K27me3 redistribution.5
It was therefore plausible that GC-derived BCL would respond to targeted treatment with specific inhibitors of EZH2’s methyltransferase activity, and one of these (tazemetostat) is now approved in the United States for treatment of FL with certain conditions, both with and without EZH2 mutation. However, the efficacy of tazemetostat against GCB-DLBCL has been only variable, even in patients with EZH2 mutations. The explanation for this is not obvious, but it is increasingly recognized that when cancers are treated with targeted inhibitors to which they are expected to respond, compensatory or “bypass” mechanisms of resistance can appear rapidly, even before the first clinical assessment of efficacy. Consistent with this, in vitro studies showed that DLBCL cell lines that had acquired resistance to EZH2 inhibitors had not lost their EZH2 activity, or the ability of the inhibitor to block it, but had increased the activity of compensatory signaling pathways.6,7
By an unbiased approach, Nguyen et al sought to find other epigenetic regulators whose abrogation might create synthetic vulnerability with EZH2 inhibition. Using a BCL cell line described as “insensitive” to EZH2 inhibition, and a library of guide RNAs targeting 1387 epigenome regulatory genes, they performed CRISPR knockout screening to identify genes whose knockout was growth inhibitory, exclusively in the presence of tazemetostat. Among 12 such genes also meeting criteria for expression and nonessentiality was DOT1L, a histone methyltransferase catalyzing H3K79 mono-, di-, and trimethylation, considered to be an activating mark. DOT1L was investigated with pinometostat, a selective small-molecule inhibitor that has been used in clinical trials. Synergy and slower growth were observed for the combination of tazemetostat and pinometostat across a panel of GCB-DLBCL lines, more so (but not exclusively) in those with EZH2 mutations. The combination also produced more pronounced changes in gene expression and H3K27 trimethylation, implicated by pathway analysis to be associated with reduced proliferation and increased plasma cell differentiation. In addition, dual inhibition reduced H3K79 methylation and was more effective against patient-derived cell lines in vitro.
The findings of Nguyen et al are very similar to those of an independent study recently published in Blood,8 which started from the hypothesis that EZH2 inhibition would synergize with inhibition of DOT1L, similarly required for GC formation.9 In both studies, combining tazemetostat with an oral DOT1L inhibitor10 increased efficacy against cell line xenografts, even without EZH2 mutation, and combination treatment increased epigenetic and transcriptional changes leading to plasma-cell differentiation and apoptosis.
These findings raise numerous questions, ranging from the academic to the practical, such as:
Since both EZH2 and DOT1L are required for GC formation, and gain-of-function DOT1L mutations are not uncommon in lung cancer, why do mutations in BCL not involve DOT1L? EZH2 and DOT1L are both H3K methyltransferases but also are substantially different, in more than just their chromatin marks; DOT1L is the only lysine methyltransferase that does not contain a SET domain, facilitating the development of selective DOT1L inhibitors.
Can combined inhibition of EZH2 and DOT1L be safely tolerated? Both studies found minor toxicity to mice in limited evaluations, but potential toxicity to humans remains to be determined. As compared with some other molecular targets of drug development for BCL, notably ones involved in signaling from the B-cell receptor (BTK, SYK, PIK3CD), EZH2 and DOT1L are more widely expressed in nonlymphoid tissues, with consequent potential for “on-target, off-tumor” toxicity.
Are there other combinations that may enhance the efficacy of EZH2 inhibition against BCL? Yes, various studies have implicated combinations targeting BCL2, IKZF1, or B-cell receptor signaling.
Finally, what are the effects of EZH2-DOT1L dual inhibition on the immune system, particularly antitumor responses? Nguyen et al found increased expression of major histocompatibility complex I (MHC-I), MHC-II, and interferon response genes, likely to increase recognition of tumor antigens by T cells. Very recently, a study using syngeneic models of GC-derived BCL found that EZH2 inhibition alone reduced immunosuppressive regulatory T cells and enhanced the ability of chimeric antigen receptor T cells to kill tumor cells.11
In summary, combined inhibition of EZH2 and DOT1L warrants further investigation for potential clinical application.
Conflict-of-interest disclosure: R.E.D. declares no competing financial interests.
