Large-Scale Analysis of Loss of Imprinting in Human Pluripotent Stem Cells  Shiran Bar, Maya Schachter, Talia Eldar-Geva, Nissim Benvenisty  Cell Reports  Volume 19, Issue 5, Pages 957-968 (May 2017) DOI: 10.1016/j.celrep.2017.04.020 Copyright © 2017 The Author(s) Terms and Conditions

Cell Reports 2017 19, 957-968DOI: (10.1016/j.celrep.2017.04.020) Copyright © 2017 The Author(s) Terms and Conditions

Figure 1 Large-Scale Analysis for Loss of Imprinting and Correlation with DNA Methylation Changes (A) Schematic diagram of the pipeline and analysis developed for the determination of loss-of-imprinting (LOI). SNPs and global gene expression values (FPKM) were extracted from RNA-seq data. For each sample, a gene was categorized as either expressed or not expressed. The expressed genes were further examined for biallelic expression to determine LOI. In most analyses, the genetic background was unknown, therefore as default we classified samples with no biallelic score as having no apparent imprinting aberrations. (B) Full analysis of LOI in PSCs. Top: heatmap of biallelic expression including the complete list of imprinted and control pluripotent genes in 271 hPSC samples. Genes are arranged according to their genomic locations. Bottom: heatmap analysis of the expressed single isoform imprinted genes in 188 unique hPSC samples (merging replicates). Colored dots represent the parental origin for the gene’s DMR (red, maternal; blue, paternal; purple, both; gray, unknown). ES, embryonic stem cells; IPS, induced pluripotent stem cells; NT, somatic cell nuclear transfer derived embryonic stem cells; NE, not expressed (FPKM < 0.2). (C) Correlation between methylation levels of imprinted germline differentially methylated region (gDMR) and biallelic expression for the gene PSIMCT-1. A schematic representation of the gene and the gDMR region is included in the top box. Maternal allele is marked in red and paternal allele in blue. Black circle represents the methylated allele. The average gDMR methylation level (y axis) was plotted as a function of the SNP score (x axis) along with the linear R2 value. (D) Same as (C) for the gene H19; dot color represents a single study. Cell Reports 2017 19, 957-968DOI: (10.1016/j.celrep.2017.04.020) Copyright © 2017 The Author(s) Terms and Conditions

Figure 2 LOI Comparison of Different PSCs and Analysis of Parental Cell Lines (A) Left: Overlapping of the best fitted trend lines for the distribution of the number of biallelically expressed (aberrant) imprinted genes in ESCs and reprogrammed PSCs. Right: boxplot diagram showing the number of biallelically expressed (aberrant) imprinted genes (p = 2 × 10−6, one-tailed equal-variance t test) for ESCs and reprogrammed PSCs. (B) Schematic diagram of imprinting aberrations in three fibroblast parental cell lines (left) and their reprogrammed derivatives (IPS, induced pluripotent stem cells; SCNT-ES, somatic cell nuclear transfer embryonic stem cells; right). The arrangement of the PSC samples reflects the logical cloning process for establishing the loss of imprinting landscape from the parental cell line. X marks the number of samples having the same allelic expression pattern. (C) Correlation between the number of biallelically expressed (aberrant) imprinted genes (y axis) and passage number (x axis) in informative reprogrammed PSCs. Marked in red box are samples with low passage and high (≥2) number of aberrant genes. Cell Reports 2017 19, 957-968DOI: (10.1016/j.celrep.2017.04.020) Copyright © 2017 The Author(s) Terms and Conditions

Figure 3 LOI Differences between Imprinted Genes Exhibit a Higher Sensitivity in Paternal DMRs (A) Left: Classification of imprinted genes according to the percentage of hPSC samples expressing them biallelically. Right: Detailed analysis of LOI level in imprinted genes. Shown are percentages of PSCs biallelically expressing every gene. (B) Average LOI differences between genes with paternal and maternal gDMRs (p = 0.04, one-tailed equal-variance t test). Error bars indicate SD. (C) Comparison of LOI in imprinted genes controlled by gDMRs from different parental origins either in ESCs (purple) or in reprogrammed PSCs (green). Shown is the percentage of aberrant samples biallelically expressing imprinted genes governed by a paternal iDMR (blue background), maternal iDMR (red background), or both. NA, genes with no known DMR. (D) Comparison of the difference in ESCs and reprogrammed PSCs LOI between genes with paternal and maternal gDMRs (p = 0.003, one-tailed equal-variance t test). Error bars indicate SD. Cell Reports 2017 19, 957-968DOI: (10.1016/j.celrep.2017.04.020) Copyright © 2017 The Author(s) Terms and Conditions

Figure 4 Effect of LOI on Expression Level and Differentiation Analysis (A) Volcano plot displaying the fold change of expression (FPKM) for biallelic (expressing SNP) and non-biallelic samples (x axis) versus the p value (y axis calculated by one-tailed equal-variance t test). (B) Fold change of expression level for biallelic and non-biallelic samples for significant genes from (A). Blue line represents the expected 2-fold upregulation. (C) Percentage of cells differentiated to neural lineage which biallelically expressed each imprinted gene. (D) Schematic diagram of imprinting aberrations in ES parental cell lines (left) and their differentiated beta cell lineage derivatives (right), according to the differentiation time point. X marks the number of samples having the same allelic expression pattern and differentiation time point. (E) Same as (D) for neural differentiation. Cell Reports 2017 19, 957-968DOI: (10.1016/j.celrep.2017.04.020) Copyright © 2017 The Author(s) Terms and Conditions