Summary: Histone H3 lysine 36 dimethylation (H3K36me2), a modification associated with transcriptional activation, is required for mixed-lineage leukemia–dependent transcription and leukemic transformation. In this issue of Cancer Discovery, Zhu and colleagues map the network of readers, writers, and erasers of H3K36me2 and uncover the ASH1L histone methyltransferase as a novel target for therapeutic intervention. Cancer Discov; 6(7); 700–2. ©2016 AACR.
See related article by Zhu and colleagues, p. 770.
Epigenetic mechanisms are essential for establishment of cellular identity. In cancer, perturbed epigenetic mechanisms subvert cellular identity to reactivate processes required for earlier development, including proliferation and migration, in differentiated, quiescent cells. Modification of histones provides a means to transmit epigenetic information from one cell to its progeny. Methylated lysine residues of histone tails are “written” by histone methyltransferases and “read” by proteins that interpret this histone code into specific gene expression states. The Cancer Genome Atlas has revealed that genes of the histone lysine N-methyltransferase 2 (KMT2) family are among the most frequently mutated in human cancer (1). Its founding member, the histone H3 lysine 4 (H3K4) methyltransferase gene mixed-lineage leukemia (MLL; also known as KMT2A), was originally identified at recurrent breakpoints in an aggressive subtype of infant leukemia characterized by 11q23 translocations. Subsequently, MLL translocations have been found in many other leukemias, all of which share a dire prognosis. These translocations lead to chimeric oncogenes in which the methyltransferase domain of MLL is lost and novel functionalities are gained depending on the >40 different fusion partners. A hallmark of MLL-rearranged leukemia is high expression of the developmental MEIS1 and HOXA genes, which are sufficient to induce leukemic transformation of hematopoietic progenitors (2). Wild-type MLL normally maintains the expression of these genes in hematopoiesis. Thus, chimeric MLL oncoproteins hijack a pathway utilized in normal development to drive leukemia. MLL fusion proteins are extremely potent in oncogenesis, as 11q23 translocations may be the sole detectable genetic lesion in many instances. Interestingly, normal hematopoietic stem cells (HSC), which are endowed with intrinsic self-renewal capacity, a critical feature of leukemia, are highly resistant to MLL fusion–induced transformation (3). Understanding how normal HSCs constrain self-renewal within narrow limits may reveal novel pathways amenable for therapeutic manipulation.
Despite their aggressive nature, MLL-rearranged leukemias are surprisingly vulnerable to loss of specific gene products under experimental conditions. For example, MLL-rearranged leukemias are highly dependent on the H3K79 methyltransferase DOT1L, a member of large multiprotein complexes containing MLL fusions. Hypermethylation of H3K79 at loci targeted by MLL fusions is a distinguishing feature of MLL-rearranged leukemias and reflects an aberrant translocation-associated epigenetic program (4). Genetic loss or small-molecule inhibition of DOT1L impairs proliferation of transformed cells. The DOT1L inhibitor EPZ-5676 has been used in ongoing phase I clinical trials, which have demonstrated activity in a subset of patients with MLL-rearranged leukemias (5). Although these results are encouraging, clinical experience teaches that targeted therapies involving a single pathway invariably fail due to evolution of drug resistance as cancer cells evolve.
In this issue of Cancer Discovery, Zhu and colleagues uncover another epigenetic Achilles' heel of MLL-rearranged leukemia (6). In normal hematopoiesis, MLL cooperates with another epigenetic writer, the H3K36 methyltransferase ASH1L, to maintain self-renewal in quiescent HSCs (7). MLL and ASH1L share multiple targets, including the MEIS1 and HOXA genes (8). Zhu and colleagues demonstrate that MLL fusion–induced leukemia depends on dimethylation of H3K36 at MLL target loci by ASH1L to promote binding of the epigenetic reader lens epithelium–derived growth factor (LEDGF; also known as PSIP1) and MLL, which also serves as an epigenetic reader in this context (Fig. 1, top). LEDGF has been reported to serve as a critical cofactor of MLL fusion proteins (9) and to recognize H3K36 methylation (10).
The authors use chromatin immunoprecipitation, coupled with high-throughput sequencing (ChIP-seq), to reveal that LEDGF, MLL, and histone H3 lysine 36 dimethylation (H3K36me2) co-occupy transcription start sites of MLL target genes. Knockdown of LEDGF reduces MLL binding at target genes and leads paradoxically to increased occupancy of MLL fusion proteins, suggesting competition for sites vacated by wild-type MLL. However, in the absence of LEDGF, MLL fusion proteins are unable to recruit higher-order protein complexes, such as AEP, which contain transcription elongation factors. Decreased recruitment of the acetyltransferase MOF (males absent on the first; also known as KAT8), which complexes with wild-type MLL and not MLL fusions, is evidenced by lower levels of acetylated chromatin at histone H4 lysine 16 (H4K16ac), which is necessary for transcriptional elongation. These findings are consistent with the observation that wild-type MLL is required for fusion-induced leukemogenesis (11). The authors also show that the binding of LEDGF and MLL is mutually dependent. This feature opens an attractive approach for targeted therapy (9), particularly because small molecules directed to the interaction pocket are already under development as anti-HIV drugs.
Zhu and colleagues also demonstrate that ASH1L colocalizes with H3K36me2, MLL, and LEDGF around the transcription start sites of MLL target genes whose expression highly distinguishes MLL-rearranged human leukemias (Fig. 1, middle). Knockdown of ASH1L reduces H3K36me2 at MLL target promoters, resulting in reduced LEDGF binding and transcript levels. As a consequence, ASH1L knockdown compromises growth of hematopoietic stem and progenitor cells transformed by MLL oncogenes in vitro and in transplantation assays, but does not affect non–MLL-transformed cells. Given the subtle effects of ASH1L loss on HSCs (7), this strategy is attractive for targeted therapy. Unexpectedly, although ASH1L loss reduces occupancy of wild-type MLL at target promoters, MLL fusion proteins again occupy these vacant target loci.
In contrast to previous reports (10), the authors show that LEDGF specifically binds dimethylated H3K36, whose distribution peaks at the transcription start site, in contrast to H3K36me3, which increases through the gene body to the transcription termination site. MLL/LEDGF binding is increased upon knockdown of SETD2, the only known histone H3K36me3 methyltransferase. A previous report showed that ASH1L carrying an inactivating SET domain mutation was more efficient in activating HOX gene expression than wild-type ASH1L, questioning whether K36 methylation was indeed a prerequisite for HOX gene expression (8). However, forcing demethylation of H3K36me2 by overexpression of the histone-lysine demethylase KDM2A promotes dissociation of MLL and LEDGF from chromatin and reduces expression of MLL target genes (Fig. 1, bottom). Ectopic expression of KDM2A induces differentiation of MLL fusion–transformed cells and antagonizes leukemogenesis, emphasizing the requirement for H3K36me2.
MLL fusion proteins trigger multiple and complex epigenetic pathways that reestablish developmental programs in otherwise differentiated cells. The study by Zhu and colleagues sheds light onto the role of H3K36 dimethylation in transcription. The dependence of MLL leukemia on H3K36me2 and the dynamic interplay between its writer ASH1L, its readers LEDGF and MLL, and its eraser KDM2A implicates ASH1L as a promising target for novel molecular therapy. However, the observed competition of MLL fusion proteins for target sites vacated by wild-type MLL upon ASH1L loss raises the question of how MLL fusions are recruited to their targets, and how wild-type MLL ignores them in the absence of ASH1L and LEDGF. It is to be feared that the residual binding of MLL fusion proteins at these key targets may provide a means for rapid acquisition of resistance to ASH1L or LEDGF inhibition.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The authors acknowledge support from the German Cancer Aid to S.T. Balbach. S.H. Orkin is an Investigator of the Howard Hughes Medical Institute.
- ©2016 American Association for Cancer Research.