Blockage of the Epithelial-to-Mesenchymal Transition Is Required for Embryonic Stem Cell Derivation

Abstract

Pluripotent cells emanate from the inner cell mass (ICM) of the blastocyst and when cultivated under optimal conditions immortalize as embryonic stem cells (ESCs). The fundamental mechanism underlying ESC derivation has, however, remained elusive. Recently, we have devised a highly efficient approach for establishing ESCs, through inhibition of the MEK and TGF-β pathways. This regimen provides a platform for dissecting the molecular mechanism of ESC derivation. Via temporal gene expression analysis, we reveal key genes involved in the ICM to ESC transition. We found that DNA methyltransferases play a pivotal role in efficient ESC generation. We further observed a tight correlation between ESCs and preimplantation epiblast cell-related genes and noticed that fundamental events such as epithelial-to-mesenchymal transition blockage play a key role in launching the ESC self-renewal program. Our study provides a time course transcriptional resource highlighting the dynamics of the gene regulatory network during the ICM to ESC transition.

Keywords: EMT; ESC derivation; mouse.

Figure 1

High-Resolution Transcriptome Trajectories from ICM to ESCs (A) Isolated ICM and whole-blastocyst outgrowths (BOs) on days 3, 5, 7, and 9, and ESCs (in passage 20). (B) Unsupervised hierarchical clustering of time course gene expression profiles for ESC derivation. Three independent biological replicates (nearly 20–30 ICMs and BOs for each replicate) were used except for BO-3 with two biological replicates. Transcript levels are mean-centered log₂ scale values. (C) Principal coordinate analysis (PCoA) of time course gene expression profiles. The red arrow shows the trajectory of cell expression profiles during ESC derivation. (D) The number of differentially expressed genes (DEGs) between each of the consecutive time points. Red and blue bars indicate up- and downregulated genes, respectively. (E) Schematic illustration of the overall sample collection time points for gene expression profiling of ESC line derivation for the second run of microarray analysis. Two culture conditions were used: treatment (R2i) and control (SB). The samples (two biological replicates) included immunosurgically isolated ICMs (day 0), ICM outgrowths (IOs) on days 0.5, 1, 2, 3, and 5, and ESCs of early (P2 and P4) and late (P15) passages. (F) PCoA of the time course transcriptome profiles. The red dotted arrows show the cell-state trajectories and the bifurcation point between R2i and SB groups. (G) The number of DEGs during the indicated time points is shown on the left and significantly upregulated (red bars) and downregulated (blue bars) genes between each pair of consecutive time points are shown on the right. (H) Comparison of the number of DEGs in the first and second run of microarray for BOs versus ESCs (in the first run), and for IOs versus ESCs (in the second run).

Figure 2

Transition from ICM to ESCs Is a Gradual Process (A) The representation of DEGs between different stages of IOs and ESCs versus ICM. (B) Functional annotation of up- and downregulated genes between IOs and ESCs of different stages versus ICM. (C) Experimental schematic and the results of time course dependency of the derived ESCs under R2i culture conditions. For each experiment (rows), the red and green bars indicate the duration of time the cells were cultured in the negative control and R2i media, respectively. The efficiency of deriving ESCs is based on the number of Nanog-positive colonies derived from ten isolated ICM. (D) The number of up- and downregulated genes between ESCs and R2i-treated IOs. (E) Functional annotation of up- and downregulated genes between P15 and IOs of different stages.

Figure 3

Transcriptome Signature of IOs (A) Categorization of DEGs for IOs on days 1 and 3 between the R2i and SB groups (IO-1 versus IOSB-1 and IO-3 versus IOSB-3). The upper panels show the pattern of this classification for the below heatmaps. Heatmaps I–IV represent the pattern of genes upregulated in R2i. Heatmaps V–VIII represent the pattern of genes downregulated in R2i versus SB. The related genes for each cluster are shown. Green and red dashes show significant down- and upregulated genes in SB versus R2i, respectively. (B) Functional annotation of heatmaps I–IV (R2i-specific genes) and V–VIII (SB-specific genes). The high ranking of gene ontology (GO) analysis, overrepresented KEGG pathways and WikiPathway are shown. (C) A set of pluripotency-related genes shared between R2i- and SB-treated IOs (common pluripotency) or specific to R2i pluripotency and SB pluripotency. (D) Efficiency of deriving ESCs upon Nanog overexpression. A Tet-On Nanog-inducible 3.5-day blastocyst from F1 hybrid × OG2 mice was used for this study.

Figure 4

Day 3 Outgrowths Represented the Highest Difference between R2i and SB Groups (A and B) Scatterplots showing the gene expression profiles of (A) SBIO-3 versus SBIO-1 and (B) SBIO-3 versus IO-3. The trend is shown as a blue line. The pink ribbon shows genes that have less than 2-fold expression changes between the two samples (p < 0.05). (C) Unsupervised clustering of gene expression for DEGs for IOs on day 3 between treatment (R2i) and control (SB) groups. Clusters I–III represent the pattern of genes that are downregulated in IO-3 versus SBIO-3. Clusters IV–VII represent the pattern of genes that are upregulated in IO-3 versus SBIO-3.

Figure 5

DNA Methylation Changes during ESC Derivation (A) Time course expression of epigenetic modifier genes during ESC derivation using qRT-PCR. Three biological replicates were used for this experiment. Error bars represent the SD. Here, ^∗ represents significant DEGs between ICM and ESCs, p ≤ 0.01. Statistical analysis was performed by one-way ANOVA and post hoc Tukey test. (B) Hairpin-bisulfite amplicon sequencing of CpG dyads at retrotransposable elements (L1Md_tf, IAP-LTR1, and mSat). The bars sum up the DNA methylation status of all CpG dyads. The map next to the bar represents the distribution of methylated sites. Each column shows neighboring CpG dyads and each line represents one sequence read. The reads in the map are sorted first by fully methylated sites, then by hemi-mCpG dyads. Red, fully methylated; light green and dark green, hemi-mCpG; blue, unmethylated CpG. Two biological replicates for IAP-LTR1 and mSat and one biological replicate for L1Md_tf were used and ^∗ represents significant difference of DNA methylation between ICM, IOs, and ESCs, p ≤ 0.01. The numbers of reads and methylation state are shown in Table S3. (C) Efficiency of ESC derivation upon inhibition of DNA methyltransferases by using RG108 in R2i- and 2i-treated IOs.

Figure 6

Blocking EMT Is Required for ESC Derivation (A) Time course expression of EMT and MET gene markers was measured by qRT-PCR in R2i- and serum + LIF-treated IOs. All data were calibrated to ICM, which is considered 1. Error bars reflect SD of results derived from three biological replicates. Here, ^∗ represents significant deferentially expressed genes in R2i- and serum + LIF-treated IOs as well as ESCs. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, error bar, ± SD. (B) Overexpression of Snail-2A-tdTomato in OG2 ICM, which contain an Oct4-GFP reporter transgene, with R2i treatment. A construct with tdTomato alone was used as a negative control. Induction of EMT through the forced expression of Snail impaired ICM to ESC transition. tdTomato expression correlates with the transduction rate and expression of the gene of interest. (C) Efficiency of ESC derivation upon Snail overexpression. (D) Overexpression of Klf4-2A-tdTomato on E3.5 ICM (B6 × C3H) F1 in N2B27 supplemented with SB + LIF prevents EMT and supports the derivation of ESC lines. (E) Efficiency of ESC derivation upon Klf4 overexpression on SB-treated IOs. (F) The impact of TGF-β1 on the generation of ESCs was assessed in 2i-treated IOs after NANOG immunostaining. Blue, DAPI.

Figure 7

R2i-Treated IOs Exhibited preEpi Cell Characteristics (A) DEGs between in vivo embryonic samples (Boroviak et al., 2015) and ICM to ESC time course samples for this study (≥2-fold expression change, adjusted p ≤ 0.01). (B) Volcano plots show fold change in gene expression (x axis, log₂ fold change, adjusted p ≤ 0.01) between E4.5 preEpi cells (Boroviak et al., 2015), IO-5, and ESCs of different passages (this study). The y axis is the negative log₁₀ of Benjamini-Hochberg adjusted p value. Blue, green, and red dots represent downregulated, non-significant, and upregulated genes, respectively. (C) The comparability of the gene expression for some heatmap II-specific genes (from Figure 3) between in vitro IOs (this study) and single-preEpi cells (EPI1-3) or single-diapaused cells (DIA1-2; Boroviak et al., 2015). (D) The expression of PGC-related genes in R2i- and SB-treated IOs.