LSD1 Influences Both Histone and DNA Methylation During Terminal Erythroid Maturation

The molecular mechanisms that drive the maturation of a committed erythroid progenitor to a functional red blood cell are incompletely understood. LSD1 (Lysine-Specific Histone Demethylase 1) is a widely expressed histone demethylase that plays an important role in erythroid maturation (Kereyni, elife, 2013). Although LSD1 is important for a number of biologic processes ranging from embryonic development to leukemogenesis, the molecular mechanisms underlying the influence of LSD1 on gene expression are incompletely understood. The goal of our study is to elucidate the molecular mechanisms by which LSD1 regulates erythroid gene expression and influences erythroid maturation. We hypothesize that LSD1 promotes specific patterns of histone and DNA methylation that facilitate gene expression changes necessary for normal erythroid maturation to occur. To address this hypothesis, the functional and molecular consequences of LSD1 knockdown were assessed in Extensively Self Renewing Erythroblasts (ESREs), a non-transformed, karyotypically normal model of terminal erythroid maturation (England, Blood, 2011). Primary fetal liver was cultured in the presence of EPO, SCF, IGF1 and dexamethasone to derive ESREs. The ESREs were capable of extensive ex-vivo expansion, doubling daily at the proerythroblast phase, however when matured, >90% of cells became benzidine positive and >65% enucleated within 3 days. Lentiviral-mediated shRNA was used to knock down LSD1 in expanding ESREs. Imaging flow cytometry done on maturation day 3 demonstrated that the knockdown cells had impairments in multiple facets of maturation, with larger cell and nuclear areas, higher kit expression, and lower rates of enucleation than the scramble control. LSD1 knockdown was also associated with impaired hemoglobin accumulation (78% vs. 95% benzidine positive; p<0.005). Treatment of ESREs with an inhibitor to LSD1 (Tranylcypromine; TCP) resulted in similar abnormalities in cell and nuclear size, kit expression, hemoglobin accumulation, and enucleation (40% vehicle vs.1% TCP). The functional deficits in maturation, including abnormal kit expression and low rates of enucleation, persisted on maturation day 4.

To delineate the molecular mechanisms underlying this maturation impairment, RNA-seq was done in LSD1 knockdown and scramble control samples, and 230 differentially expressed genes (FDR<0.01) were identified using cuffdiff (Trapnell, Nat Biotech, 2013). Consistent with LSD1’s role in erythroid maturation, Ingenuity Pathway Analysis identified multiple networks involving hemoglobin synthesis, and GATA1, EPO, and KLF1 were all predicted as upstream regulators (p-values of 8.24e10-11, 7.25 e10-6, and 3.86e10-4, respectively). To better understand how LSD1 influences gene expression, chromatin immunoprecipitation coupled with high throughput sequencing was used to identify sites of H3K4me2 binding in the differentially expressed genes. 214/230 differentially expressed genes were associated with sites of H3K4me2 occupancy. Quantitative ChIP demonstrated that LSD1 inhibition was associated with increases in H3K4me2 levels at a subset of these sites, however consistent with previous studies, global levels of H3K4me2, determined by Enzyme Linked Immunosorbent Assay (ELIZA), did not change significantly.

Although it is known that LSD1 demethylates and stabilizes the maintenance DNA methyltransferase DNMT1 (Wang, Nat Genet 2009), the consequences of LSD1 loss on DNA methylation (5-methyl cytosine; 5-mC) have yet to be investigated. To gain a comprehensive understanding of how LSD1 regulates erythroid gene expression, changes in the level of 5-mC were assessed after knockdown or inhibition of LSD1. Global 5-mC levels, determined by ELIZA assay, were ∼30% lower in TCP treated samples than vehicle treated control (p<0.02) and western blot demonstrated a 3-fold decrease in DNMT1 protein in the TCP treated samples. Both methyl binding domain pull-down coupled with quantitative PCR and genome-wide bisulfite sequencing were utilized to assess changes in 5-mC levels in the differentially expressed genes. Loss of LSD1 was associated with significantly lower levels of 5-mC at several differentially expressed, erythroid-specific genes, such as bh1. Taken together, these data support the hypothesis that LSD1 influences both histone and DNA methylation at genes important for erythroid maturation.