Abstract
Loss of imprinting (LOI) is the reactivation of the silenced allele of an imprinted gene, leading to perturbation of monoallelic expression. We tested the hypothesis that LOI of PLAGL1, a representative maternally imprinted gene, occurs through an all-or-none process leading to a mixture of fully imprinted and nonimprinted cells. Herein using a quantitative RT-PCR-based experimental approach, we measured LOI at the single cell level in human trophoblasts and demonstrated a broad distribution of LOI among cells exhibiting LOI, with the mean centered at approximately 100% LOI. There was a significant (P < 0.01) increase in expression after 2 days of 5-aza-2'-deoxycytidine (AZA) treatment and a significant (P < 0.01) increase in LOI after both 1 and 2 days of AZA treatment, while the distribution remained broad and centered at approximately 100% LOI. We propose a transcriptional pulsing model to show that the broadness of the distribution reflects the stochastic nature of expression between the two alleles in each cell. The mean of the distribution of LOI in the cells is consistent with our hypothesis that LOI occurs by an all-or-none process. All-or-none LOI could lead to a second distinct cell population that may have a selective advantage, leading to variation of LOI in normal tissues, such as the placenta, or in neoplastic cells.