Control of embryonic stem cell self-renewal and differentiation via coordinated alternative splicing and translation of YY2

Abstract

Translational control of gene expression plays a key role during the early phases of embryonic development. Here we describe a transcriptional regulator of mouse embryonic stem cells (mESCs), Yin-yang 2 (YY2), that is controlled by the translation inhibitors, Eukaryotic initiation factor 4E-binding proteins (4E-BPs). YY2 plays a critical role in regulating mESC functions through control of key pluripotency factors, including Octamer-binding protein 4 (Oct4) and Estrogen-related receptor-β (Esrrb). Importantly, overexpression of YY2 directs the differentiation of mESCs into cardiovascular lineages. We show that the splicing regulator Polypyrimidine tract-binding protein 1 (PTBP1) promotes the retention of an intron in the 5'-UTR of Yy2 mRNA that confers sensitivity to 4E-BP-mediated translational suppression. Thus, we conclude that YY2 is a major regulator of mESC self-renewal and lineage commitment and document a multilayer regulatory mechanism that controls its expression.

Keywords: 4E-BPs; PTBP; YY2; embryonic stem cell; mRNA translation.

Fig. S1.

Translation control in WT and 4E-BP1/2–null mESCs. (A) Lysates from WT and Eif4ebp1 and Eif4ebp2 DKO mESCs were subjected to m⁷GTP pull-downs and analyzed for the indicated proteins. Numbers indicate the ratio of eIF4G1 in each pull-down to that in WT cells (B) Polysome profiles of WT and DKO mESCs treated with 100 µg/mL cycloheximide for 5 min. Absorbance light was set at 254 nm. (C) [³⁵S] methionine/cysteine incorporation into newly synthesized proteins from WT and DKO mESCs grown in 10% (vol/vol) dialyzed FBS and pulsed for 30 min with [³⁵S] methionine/cysteine. Data are mean ± SD (n = 3). (D) Correlation between replicates in mRNA-Seq and ribosome profiling datasets. R² indicates the Pearson correlation. (E) Metagene analysis of randomly fragmented mRNAs and RFPs in mESCs. Normalized read counts are averaged across the entire transcriptome and aligned at the annotated start codons and stop codons.

Fig. 1.

The lack of 4E-BPs deregulates the expression of pluripotency factors in mESCs. (A and B) The log₂ abundance of RFPs (A) and mRNA (RNA-Seq) (B) of transcripts that were included in Babel analysis are plotted for WT and Eif4ebp1 and Eif4ebp2 DKO mESCs. (C) Babel analysis of transcripts with a significant change in RFPs independent of the corresponding change in mRNA abundance (black dots; FDR < 0.1). Triml2 and Trmt61b, respectively, are mRNAs with the highest and the lowest RFP ratios in DKO compared with WT mESCs; FC, fold change. (D) Western blot analysis of NANOG expression in a WT mESC and in two independent DKO mESC clones (C1 and C2). Numbers indicate the ratio of NANOG expression in each clone to that in the WT mESC followed by normalization with β-actin. (E) Western blot analysis of YY2 and YY1 expression in a WT mESC and two independent DKO mESC clones. Numbers indicate the ratio of YY2 expression in each clone compared with the WT mESC followed by normalization with α-tubulin. (F) DKO mESCs carrying the doxycycline-inducible 4E-BP1-4A mutant construct were treated with 0, 0.005, 0.01, 0.05, 0.1, or 0.2 μg/mL doxycycline for 24 h and were subjected to Western blot analysis. Numbers indicate the ratio of YY2 expression in each treatment compared with no doxycycline followed by normalization with β-actin.

Fig. S2.

Characterization of mESCs in the presence or absence of 4E-BP1/2. (A) Western blot analysis of mESCs transduced by lentivirus expressing shRNAs against 4E-BP1 and 4E-BP2 (shEif4ebp1/2) or control (shCTR). LE, long exposure; SE, short exposure. Numbers indicate the ratio of NANOG expression in each condition to that in control followed by normalization with α-tubulin. (B) Cell growth assay for WT and Eif4ebp1 and Eif4ebp2 DKO mESCs. Data are mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001. (C) WT and DKO mESCs cultured in the absence of feeder layer. (D) Control and 4E-BP1 and DKD mESCs cultured in the absence of feeder layer. (E and F) Western blot (E) and RT-qPCR (F) analysis of expression of selected pluripotency factors in WT and DKO mESCs cultured in the absence of feeder layer and LIF for the indicated time. RT-qPCR values are normalized to β-actin. Data are mean ± SD (n = 3). Numbers in E indicate the ratio of NANOG or OCT4 expression in each condition to that at day 0 of differentiation of WT mESCs followed by normalization with β-actin. (G) RT-qPCR analysis of selected pluripotency markers in WT and DKO mESCs. Values are normalized to β-actin. Data are mean ± SD (n = 3). **P < 0.01, ***P < 0.001. (H) RT-qPCR analysis of WT and DKO EBs 2 wk postdifferentiation for expression of selected differentiation markers. Values are normalized to β-actin. Data are mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001; ns, nonsignificant.

Fig. S3.

Role of YY2 in mESCs and blastocysts formation. (A) RT-qPCR analysis of Yy2 mRNA expression in WT and DKO mESCs. Values are normalized to β-actin. Data are mean ± SD (n = 3); ns, nonsignificant. (B) WT and DKO mESCs were transduced with three independent shRNAs against Yy2 or control shRNA. Resistant colonies were selected with puromycin (5 μg/mL) for 2 d and subjected to alkaline-phosphatase staining. (C) Western blot analysis of mESCs transduced with three independent shRNAs against Yy2 or control shRNA. Numbers indicate the ratio of cleaved caspase-3 in each condition to that in control after normalization with α-tubulin. (D) Schematic presentation of the CRISPR-Cas9 target site in the mouse Yy2 gene. The guide RNA was specifically designed to avoid regions with high homology to mouse Yy1. PAM, protospacer-adjacent motif. Numbers indicate the distance from the start codon. (E) The sequence of the Yy2 alleles in the WT and two mutant male blastocysts depicted in Fig. 2D. (F) Western blot analysis of mESCs overexpressing GFP or v5-YY1. (G) Graphical representation of selected YY2-binding peaks obtained from the UCSC browser. Twenty-kilobase windows are displayed.

Fig. 2.

Stringent regulation of YY2 levels is critical for mESC survival and differentiation. (A and B) Western blot (A) and RT-PCR (B) analysis of WT mESCs carrying the doxycycline-inducible YY2 construct and treated with 0, 0.2, 1, or 4 μg/mL doxycycline for 24 h. Numbers indicate the ratio of the expression of the identified protein in each treatment compared with no doxycycline followed by normalization with β-actin. (C) RT-qPCR analysis of mESCs carrying doxycycline-inducible shRNA against Yy2 (shYy2) and treated with doxycycline (0, 0.2, 1, or 4 µg⋅mL⁻¹⋅d⁻¹) for 72 h. Values are normalized to β-actin. Data are mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001; ns, nonsignificant. (D) Blastocyst outgrowth assay in a WT embryo and in two independent CRISPR/CAS9-mediated Yy2-knockout embryos. Cas9 mRNA and sgRNAs targeting Yy2 were injected into zygotes. The blastocysts derived from injected embryos were subjected to the blastocyst outgrowth assay. The mutagenesis strategy and the sequence of mutant alleles are provided in Fig. S3 D and E. (E) RT-PCR analysis of EBs carrying the doxycycline-inducible YY2 construct and treated with 0, 0.002, 0.02, or 0.2 μg/mL doxycycline every other day for 4 wk.

Fig. 3.

YY2 controls the ESC transcriptional regulatory network and development-related genes. (A) Pie chart displaying the distribution of YY2 ChIP-Seq peaks across the genome based on the distance of the peaks from the nearest RefSeq gene: proximal, <2 kb upstream of the TSS; gene, exon or intron; distal, 2–10 kb upstream of the TSS; 5d, 10–100 kb upstream of the TSS; gene desert, >100 kb from a RefSeq gene; and other, anything not included in the above categories. (B) Histogram depicting the distance of YY2 ChIP-Seq peaks relative to the TSS of the nearest gene. (C) De novo motifs enriched in YY2-binding events. Enrichment P values and percentage of targets containing each motif are displayed, as generated by HOMER software. (D) Plots showing the average density of selected motifs in a window 2 kb from the YY2 peak center. (E) The most significantly enriched canonical pathways in genes associated with YY2 ChIP-Seq peaks, as identified by IPA. (F) Standard ChIP-qPCR validation of YY2-binding regions. Data are normalized to IgG. The Lmnb2 gene was used as a negative control. (G) Graphical representation of selected YY2-binding peaks, obtained from the University of California, Santa Cruz (UCSC) browser. Twenty-kilobase windows are displayed.

Fig. 4.

The regulatory network and distinct mode of action of YY2 compared with YY1. (A) RT-qPCR analysis of selected YY2 targets in control and YY2-overexpressing mESCs. mESCs carrying the doxycycline-inducible YY2 construct were treated with 0 or 0.2 μg/mL doxycycline for 24 h. Values are normalized to β-actin. Data are mean ± SD (n = 3). **P < 0.01, ***P < 0.001; ns, nonsignificant. (B) RT-qPCR analysis of YY2 targets in mESCs carrying a doxycycline-inducible shYy2 and treated with doxycycline as described in Fig. 2C. Values are normalized to β-actin. Data are mean ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001; ns, nonsignificant. (C) Comparison of YY2 and YY1 ChIP-Seq targets in mESCs. Only peaks with at least a 1-nt overlap were considered as common targets. (D) EMSA with a radioactive-labeled dsDNA oligonucleotide probe derived from the promoter region of the mouse Arid1a gene (32) and purified recombinant mouse triple Flag-tagged (3xF)-YY1 and 3xF-YY2 proteins. The probes were incubated in the presence of increasing amounts of recombinant proteins and in the presence or absence of antibodies as indicated. (E) EMSA with a radioactively labeled single-stranded RNA oligonucleotide probe derived from the promoter region of the mouse Arid1a gene (32) and purified recombinant mouse 3xF-YY1 and 3xF-YY2 proteins.

Fig. S4.

YY1, but not YY2, binds RNA through its N-terminal domain. (A) Pairwise alignment of mouse YY1 and YY2 protein sequences using Geneious software. The protein motifs are clustered according to their similarity. (B) Purification of 3xF-YY1 and 3xF-YY2 proteins from HEK293H cell lysate under nondenaturing conditions with anti-Flag M2 agarose beads and elution with 3xFlag-peptide. (C) Coomassie-stained 10% (wt/vol) SDS/PAGE gel of purified 3xF-YY1 and 3xF-YY2 proteins. (D) Prolonged exposure (3 wk) of the EMSA with the radioactively labeled single-stranded RNA oligonucleotide probe described in Fig. 4E. (E) RNA-binding prediction at the N-terminal domains of YY1 and YY2 using the BindN server (39). Predicted binding residues (pink) are labeled “+”; nonbinding residues (green) are labeled “−”. Numbers denote confidence from level 0 (lowest) to level 9 (highest).

Fig. 5.

Retention of the 5′-UTR intron renders Yy2 sensitive to 4E-BP–mediated translation suppression. (A) Sequence of the promoter region of the mouse Yy2 gene. The two alternative TSS are marked by arrowheads; the boxed sequence shows the retained intron; the two hexamers highlighted in red are the consensus PTBP-binding motifs; and the underlined ATG is the translation start codon for Yy2 mRNA. (B) A cartoon depicting the four variants of Yy2 5′-UTR. vB and vΔB represent a long variant with and without intron retention, respectively; vA and vΔA represent a short variant with and without intron retention, respectively. CD, coding DNA sequence. (C) RT-PCR using the primer pair Fw2 and Rv designed to recognize all four possible variants to estimate the splicing efficiency of the 5′-UTR intron in mESCs and mEBs on days 4 and 6 postdifferentiation. Gapdh mRNA was used as the control. (D) RT-PCR analysis of intron retention (IR) in the Yy2 5′-UTR in different mouse embryonic stages and adult tissues using the primers described in C. Gapdh mRNA was used as the control. (E) RT-PCR analysis of intron retention in the Yy2 5′-UTR using the primers described in C upon depletion of PTBP1 expression in mESCs by two independent shRNAs. Yy2-ORF primers amplifying a segment of the coding region of Yy2 transcript were used to demonstrate the change in overall expression of Yy2 mRNA. β-actin mRNA was used as the internal control. (F) RT-PCR amplification (primers Fw2 and Rv) of the in vitro splicing products of the A, B, and ΔA variants in WERI retinoblastoma cell extracts with different amounts of recombinant PTBP1 protein. Recombinant BSA was used as a negative control. (G) Luciferase reporter assay with Firefly (Fluc) and Renilla (Rluc) luciferase reporter mRNAs, as described in Fig. S5J. The in vitro-transcribed mRNAs were purified and transfected into WT and DKO mESCs. (Left) The normalized luciferase activity of each construct in WT mESCs. (Right) Comparison of the luciferase activity of each construct in DKO and WT mESCs. Fluc mRNA was cotransfected with Rluc mRNA as a transfection control. Data are mean ± SD (n = 3). **P < 0.01, ***P < 0.001; ns, nonsignificant. CTR, control; RLU, relative luminescence units.

Fig. S5.

PTBP1 regulates Yy2 5′-UTR intron retention. (A) RT-PCR analysis of mESCs and day 4 (D4) and day 6 (D6) mEBs using the primer pair designed to recognize the B and ΔB variants (Fw1 and Rv) to estimate the splicing efficiency of the Yy2 5′-UTR intron. (B) RT-PCR analysis (primers Fw1 and Rv) of intron retention in variant B of the Yy2 5′-UTR at different embryonic days (Left) and in different mouse tissues (Right). (C) Consensus PTBP1- or MBNL1-binding motifs in the 5′-UTR of mouse Yy2 mRNA, identified by the RBPmap web server. (D) Western blot analysis of PTBP1 expression in mESCs stably expressing two independent shRNAs against Ptbp1 or control shRNA (shGFP); this panel is related to Fig. 5E. (E) Representative images of the mESC colonies described in D. (F and G) RT-PCR analysis of the in vitro splicing products of the A, B, and ΔA variants in the presence (+) or absence (−) of the WERI retinoblastoma cell extract. Primers Fw2 and Rv were used in F to recognize all four possible variants, and primers Fw1 and Rv were used in G to recognize B and ΔB variants. (H) Coomassie-stained 10% (wt/vol) SDS/PAGE to detect the level of recombinant GST-PTBP1 protein used for the in vitro splicing assay in I and Fig. 5F. One microgram of BSA was loaded as control. (I) RT-PCR analysis (primers Fw1 and Rv) of the in vitro splicing products in WERI retinoblastoma cell extracts with different amounts of recombinant PTBP1 protein. Recombinant BSA was used as a negative control. (J) Schematic diagrams of the Renilla (Rluc) and Firefly (Fluc) luciferase reporter mRNA constructs used in G.

Fig. S6.

Yy2 5′-UTR intron retention influences its secondary structure and expression in mouse tissues. (A) Effect of intron retention on the sequence and structure of Yy2 5′-UTR. The mfold web server (unafold.rna.albany.edu/?q=mfold/RNA-Folding-Form) was used to predict the secondary mRNA structures and ΔG (Gibbs free energy). Nucleotides are numbered starting from the relevant TSS. The intronic sequence is highlighted in red. (B) Expression of YY2, PTBP1, and PTBP2 proteins in different tissues of a 4-wk-old female mouse. LE, long exposure; SE, short exposure. Numbers indicate the ratio of YY2 expression in each tissue compared with lung followed by normalization with α-tubulin.

Fig. S7.

Expression profiles of Ptbp1, Eif4ebp1, Eif4ebp2, and Sox2 in a cohort of embryonic and adult mouse samples. The references for the RNA-Seq studies used in this analysis are listed in Dataset S4.

Fig. 6.

Regulation of YY2 expression by splicing and mRNA translational control. A proposed model depicting the regulation of YY2 expression at two different steps: (i) alternative splicing (via PTBP1) and (ii) mRNA translation (via 4E-BP1/2). A basal level of YY2 expression is required for maintenance of mESC self-renewal, but increased translation of Yy2 mRNA directs cardiovascular lineage commitment.