The Histone Variant MacroH2A Blocks Cellular Reprogramming by Inhibiting Mesenchymal-to-Epithelial Transition

Abstract

Transcription factor-induced reprogramming of somatic cells to pluripotency is mediated via profound alterations in the epigenetic landscape. The histone variant macroH2A1 (mH2A1) is a barrier to the cellular reprogramming process. We demonstrate here that mH2A1 blocks reprogramming and contributes to the preservation of cell identity by trapping cells at the very early stages of the process, namely, at the mesenchymal-to-epithelial transition (MET). We provide a comprehensive analysis of the genomic sites occupied by the mH2A1 nucleosomes in human fibroblasts and embryonic stem (ES) cells and how they affect the reprogramming of fibroblasts to pluripotency. We have integrated chromatin immunoprecipitation sequencing (ChIP-seq) data with transcriptome sequencing (RNA-seq) data using cells containing reduced levels of mH2A1 and have inferred mH2A1-centered gene-regulatory networks that support the fibroblast and ES cell fates. We found that the exact positions of mH2A1 nucleosomes in regulatory regions of specific network genes with key regulatory roles guarantee the functional robustness of the regulatory networks. Using the reconstructed networks, we can predict and validate several components and their interactions in the establishment of stable cell types by limiting progression to alternative cell fates.

Keywords: chromatin structure; gene expression; histone variants; reprogramming.

FIG 1

MacroH2A1.2 blocks cellular reprogramming. (A) Western blot depicting mH2A1.2 protein levels in mH2A1-KD human fibroblasts (shmH2A1) compared to the corresponding levels in a control sample (scramble). GAPDH was used as a loading control. (B) Representative AP staining of iPSC colonies formed after reprogramming of mH2A1-KD human fibroblasts (shmH2A1) or control cells (scramble). (C) Relative reprogramming efficiency of mH2A1-KD human fibroblasts (shmH2A1) compared to the control sample (scramble), which was taken as 100%. AP-positive iPSC-like colonies were counted, and their percentage was plotted relative to the scramble. The mean value and standard deviation (SD) (n = 3) are shown. **, P ≤ 0.01. (D) Phase-contrast images taken from a time course reprogramming experiment showing the morphological changes and colony formation from day 0 to day 18 of mH2A1-KD human fibroblasts (shmH2A1) (bottom) and control sample cells (scramble) (top). (E) Western blot using the anti-mH2A1.2 antibody in extracts prepared from human fibroblasts overexpressing mH2A1.2 (mH2A1.2/myc-tag) or from control cells (empty). GAPDH was used as a loading control. (F) Relative efficiencies of reprogramming of human fibroblasts overexpressing H2A (H2A), mH2A1.1 (mH2A1.1), mH2A1.2 (mH2A1.2), or control vector (empty). AP-positive colonies were counted, and their percentage was plotted in relation to the empty vector. The mean values and SD (n = 3) are shown. **, P ≤ 0.01. (G) Phase-contrast and fluorescent images taken from a time course reprogramming experiment showing the morphological changes and iPSC colony formation, from day 0 to day 18, of human fibroblasts overexpressing H2A (H2A) (middle) or mH2A1.2 (mH2A1.2) (bottom). The empty vector was used as a control (empty) (top).

FIG 2

mH2A1.2 blocks the cadherin switch required for cellular reprogramming. (A) Western blot analysis of hFBS extracts prepared from mH2A1.2-KD (sh) or control (scr) cells undergoing reprogramming at the indicated time points using an antibody against mH2A1.2. GAPDH was used as a loading control. (B) Normalized levels of N-CAD, E-CAD, SNAI1, and NANOG mRNA at the indicated time points during the reprogramming process of control hFBSs (scramble) or mH2A1-KD cells (shmH2A1) as quantified by qPCR. The data were normalized to the levels of GAPDH and plotted in relation to the median expression value. The error bars indicate SD.

FIG 3

mH2A1.2 blocks MET transition. (A) Confocal images, taken from a time course reprogramming experiment, of human fibroblasts overexpressing mH2A1.2 or control cells (empty). The cells were fixed at the indicated time points and stained for E-CAD (red), and nuclei were stained with DAPI (blue). GFP was coexpressed with mH2A1.2/myc-tag or the empty vector. The insets on day 18 depict human fibroblasts overexpressing mH2A1.2-GFP, which were not stained for E-CAD. However, colocalization of GFP with E-CAD was observed in empty transduced human fibroblasts, indicating that the presence of mH2A1.2 prevents the reprogramming process at the stage of MET. (B) Same as panel A except that the cells were transduced with a vector expressing shmH2A1. The insets for day 18 indicate the colocalization of GFP and E-CAD in cells expressing either scramble or shmH2A1.

FIG 4

Distinct mH2A1.2 nucleosome occupancy in human ESCs and human fibroblasts. (A) Numbers of genes occupied by single nucleosomes and polynucleosomes containing mH2A1.2 in regions spanning 5 kb on either side of the TSS in human ESCs and human FBSs. (B) Comparison of numbers of genes occupied by mH2A1.2 single nucleosomes and polynucleosomes in regions spanning 5 kb on either side of the TSS in human ESCs and human FBSs. (C) Snapshots (IGV) of genes bound by mH2A1.2, depicting the cell-type-specific density profiles of human ESCs (blue track) and human FBSs (red track) (compare top and bottom views for each corresponding region). The gene models (black) depict exons as boxes and introns as lines. (D) Comparison of genes occupied by mH2A1.2 single nucleosomes and polynucleosomes in regions spanning 5 kb on either side of the TSS in human and mouse ESCs. (E) Comparison of genes occupied by mH2A1.2 single nucleosomes and polynucleosomes in regions spanning 5 kb on either side of the TSS in hFBSs and MEFs.

FIG 5

mH2A1.2 nucleosomes are enriched at the promoters of expressed genes in hESCs. (A) (Top) Distribution of mH2A1.2 mononucleosomes and polynucleosomes in the region spanning 5 kb on either side of the TSS of target genes in human ESCs and human FBSs. (Bottom) Distribution of mH2A1.2 mononucleosomes in the region spanning 1 kb on either side of the TSS of target genes in human ESCs and human fibroblasts. (B) (Top) Graph showing the categorization of mH2A1.2 target genes (within 5 kb on either side of TSS) as expressed or nonexpressed genes in human ESCs and human FBSs. (Bottom) Pie charts depicting the percentages of the nonexpressed mH2A1.2 target genes as a fraction of the total number of nonexpressed genes (red line). Similarly, the fractions of expressed mH2A1.2 target genes are depicted compared to the total number of expressed genes (green line). (C) mH2A1.2 is preferentially located at the promoters of active genes in hESCs. mH2A1.2 ChIP aggregated enrichment profiles around TSS in hESCs (top) and hFBSs (bottom). All expressed genes were identified by RNA-seq and divided into five groups according to their expression levels. Each line represents the average number of reads per transcript plotted relative to the TSS for each expression group. At the top is an enlargement of the region within 1 kb on either side of the TSS of all mH2A1.2 targets in hESCs. mH2A1.2 is depleted from the promoters of all expression categories in hFBSs (bottom).

FIG 6

The histone variants mH2A1.2 and H2A.Z cooccupy TSS of highly expressed genes in ESCs. (A) Percentages of all mH2A1.2 target genes cobound by H2A.Z or marked by the H3K27me3 repressive mark in hESCs and hFBSs. (B) (Top) Distribution of H2A.Z nucleosomes and H3K27me3 modification in the region spanning 1 kb on either side of the TSS of expressed mH2A1.2 target genes in hESCs and hFBSs. (Bottom) Distribution of H2A.Z nucleosomes and H3K27me3 modification in the region spanning 1 kb on either side of the TSS of nonexpressed mH2A1.2 target genes in hESCs and hFBSs.

FIG 7

Distinct transcriptional-regulatory elements are masked by mH2A1.2 nucleosomes in hESCs and hFBSs. (A) Main TF DNA binding consensus motifs corresponding to either activators or repressors identified within the footprint of the mH2A1.2 mononucleosomes for expressed or nonexpressed genes in hESCs. The relative enrichment (P value) for each predicted motif is shown. (B) Scatterplot depicting the probabilities calculated for each predicted motif (individual P values) corresponding to activators or repressors found in either expressed or nonexpressed genes in hESCs. (C) Sums of all activator and repressor putative DNA binding sites masked by mH2A1.2 mononucleosomes in expressed and nonexpressed genes in hESCs illustrating a higher proportion of activator masking sites in expressed genes. (D) Same as panel A, but for hFBSs. (E) Same as panel B, but for hFBSs. (F) Same as panel C, but for hFBSs, except that in hFBSs a higher proportion of repressor binding sites are masked in expressed genes.

FIG 8

mH2A1.2 nucleosomes occupy the promoters of a specific set of genes and direct opposite expression programs in hESCs and hFBSs. (A) Snapshots of ChIP-seq profiles (IGV) of genes bound by mH2A1.2 depicting the distinct nucleosome positioning in hESCs (blue track) and hFBSs (red track) (compare top and bottom views for each corresponding region). The ChIP-seq profiles of the key MET regulators, SNAI1 and TWIST1, are shown on the right. The gene models (black) depict exons as boxes and introns as lines. (B) Comparison of the expression levels of the commonly bound mH2A1.2 target genes and the key regulators of MET (SNAI1 and TWIST1 [bottom]) as determined by RNA-seq in hESCs and hFBSs. The heat map (right) displays the ChIP-seq data for mH2A1.2 nucleosomes (within 5 kb on either side of the TSS) of the common target genes with opposite expression patterns in hESCs and hFBSs and the corresponding analysis of the same genes in mESCs and MEFs. The right part of the table displays our RNA-seq analysis in WT mESCs and MEFs, as well as in mH2A1.2-KD mESCs and MEFs. The asterisks denote genes with similar expression patterns in both hESCs/mESCs and in hFBSs/MEFs. (C) Relative positioning of mH2A1.2 nucleosomes in the region within 5 kb on either side of the TSS of the common target genes in hESCs and hFBSs that are differentially expressed in these cell types.

FIG 9

mH2A1.2-centered gene regulatory networks in hESCs and hFBSs define cell fates. GRNs for hFBSs (A) and hESCs (B) were constructed by integrating the experimentally verified interactions of key molecules of cellular reprogramming with direct downstream targets of mH2A1.2 based on our ChIP-seq and RNA-seq analyses.