A Novel Epigenetic Mechanism in the Pathogenesis of Myelodysplastic Syndrome (MDS)

Abstract 2783

Refractory cytopenia with multilineage dysplasia (RCMD) is a subtype of MDS characterized by cytopenia, multilineage dysplasia, but no increase in blasts. These features suggest dysregulation in the lineage commitment of progenitors early in hematopoietic cell development. PU.1, an ETS transcription factor, is considered a master regulator of hematopoiesis. In humans three functionally distinct PU.1 isoforms are present including PU.1, SPIB and SPIC. While the essential role of PU.1 in normal myelopoiesis has been well-documented, its role as well as the role of the PU.1 isoforms in MDS pathogenesis is largely unknown. Recent studies have shown that PU.1 and other transcription factors require a suitable chromatin microenvironment to function properly. EZH2, a histone methyltransferase-containing component in the polycomb complex, can add three methyl groups to histone 3 lysine 27 (H3K27me3) and cause chromatin condensation and gene silencing. Although EZH2 and H3K27me3 are critical in modulating the self-renewal capacity and the differentiation property of hematopoietic stem cells (HSCs), the role of EZH2 and H3K27me3 in MDS has not yet been clarified.

We used morphology/cytogenetics to guide genome-wide analysis of epigenetic alterations in RCMD. Ten initial diagnostic RCMD bone marrow specimens were selected for epigenetic profiling experiments in this study. Ten age/sex matched normal bone marrows were used as controls. All RCMD cases had similar morphology with trilineage dysplasia, blasts <5% and normal cytogenetics. Chromatin immunoprecipitation-coupled genome-wide promoter array analyses (ChIP-on-chips) with anti-H3K27me3 antibody were performed on the pools of RCMD and normal marrow cells to profile the changes in H3K27me3. Genome-wide ChIP-on-chips with antibodies against H3K27me3, PU.1 and EZH2 were also performed on selected individual RCMD cases. Computational motif analysis was used to identify statistically significant enrichment of DNA motifs. Quantitative RT-PCRs were used to measure mRNA expression. Flow cytometric and immunohistochemical studies were performed to measure expression of PU.1 and other proteins.

Our epigenetic profiling of H3K27me3 showed that there are about 4600 annotated gene promoters that have increased levels (> 1.5 fold) of H3K27me3 in RCMD compared to the normal. Further computational motif analysis of the DNA sequences in the hyper-H3K27me3 regions in RCMD showed that there is a statistically significant enrichment of the PU.1 binding DNA motif (PU-box). Further, an inverse relationship exists between the hyper-H3K27me3 and PU.1 binding as well as the mRNA expressions of the PU.1 down-stream key myeloid genes at the genome-wide scale. A significant increase in H3K27me3 is seen at the PU.1 gene locus in RCMD, while the three PU.1 isoform encoding genes including SPI1, SPIB, and SPIC show different patterns of H3K27me3 at their promoters, which are inversely correlated with the mRNA expression of these genes. Using an erythroid/myeloid cell line derived from a patient with high-grade MDS, we demonstrated that H3K27me3 inhibitors can inhibit cell proliferation and promote cell differentiation. In contrast, these effects were not found with all-trans retinoic acid (ATRA), a well-known histone deacetylase inhibitor (HDACI). We also demonstrated that H3K27me3 inhibitors can effectively inhibit cell growth and promote differentiation as does ATRA in promyelocytes. Flow cytometric studies showed the H3K27me3 inhibitors-induced granulocytic differentiation is correlated with increased expressions of CD18 and PU.1 genes, which in RCMD were shown to have hyper-H3K27me3 at the promoters. Our data revealed a novel epigenetic mechanism involving hyper-H3K27me3 in the PU.1 pathway in the pathogenesis of MDS and suggests possible new biomarkers to support clinical diagnosis, epigenetic therapy monitoring, and potential new therapy development.