Research ArticleGENETICS

Genome-wide kinetic properties of transcriptional bursting in mouse embryonic stem cells

See allHide authors and affiliations

Science Advances  17 Jun 2020:
Vol. 6, no. 25, eaaz6699
DOI: 10.1126/sciadv.aaz6699
  • Fig. 1 Genome-wide analysis of the kinetic properties of transcriptional bursting.

    (A) Schematic diagram of gene expression with stochastic switching between ON and OFF states. (B) Schematic representations of the dynamics of transcript levels of a gene with or without transcriptional bursting. (C) Transcriptional bursting induces inter-allelic and intercellular heterogeneity in gene expression (left). Scatter plots of the individual allele-derived transcript numbers (right). (D) Schematic representation of scRNA-seq using hybrid mESCs. (E) Scatter plot of mean normalized read counts and normalized intrinsic noise of individual transcripts revealed by scRNA-seq. (F) Representative scatter plots of normalized individual allelic read counts of high and low intrinsic noise transcripts. N. int. noise, normalized intrinsic noise. (G) Scatter plot of burst size and burst frequency of individual transcripts. (H) Schematic representation of KI of GFP and iRFP gene cassette into individual alleles of mESC derived from inbred mice. Targeted genes are listed in the lower panel. Asterisks indicate genes in which KI cassettes were inserted immediately downstream of the start codon. (I to L) Scatter plots of the mean number of transcripts of targeted genes in KI cell lines counted by smFISH versus mean normalized read counts of corresponding genes in hybrid mESCs revealed by scRNA-seq (I) or versus mean expression levels of targeted genes in KI cell lines revealed by flow cytometry (K). Scatter plots of normalized intrinsic noise of targeted gene transcripts in KI cell lines revealed by smFISH versus that of corresponding genes in hybrid mESCs revealed by scRNA-seq (J) or versus that of targeted genes in KI cell lines revealed by flow cytometry (L).

  • Fig. 2 Exploring molecular determinants of transcriptional bursting.

    (A to C) Kinetic properties of transcriptional bursting of genes either with or without a TATA box. (D) Schematic representation of calculating reads per million (RPM) at the promoter and gene body from ChIP-seq data. In addition, similar calculations were also performed for enhancers (see Materials and Methods). (E) Heat maps of Spearman’s rank correlation between promoter-, gene body–, or enhancer-associated factors and either normalized intrinsic noise (N. int. noise), burst size, or burst frequency (burst freq.). (F) Effect of the Pol II pause release inhibitor, DRB, and flavopiridol treatment on the kinetic properties of transcriptional bursting. Δnormalized intrinsic noise, Δburst size, and Δburst frequency are residuals of normalized intrinsic noise, burst size, and frequency of inhibitor-treated cells from that of control cells, respectively. Error bars indicate 95% confidence interval. (G) Effect of Suz12 K/O on normalized intrinsic noise. Suz12 K/O cell lines derived from Dnmt3l, Dnmt3b, Peg3, and Ctcf KI cell lines were established. Upper panel represents the result of Western blotting. In the lower part of the panel, the Δnormalized intrinsic noise, Δburst size, and Δburst frequency compared with the control (cont1) are shown. Error bars indicate 95% confidence interval. Asterisks indicate significance at P < 0.05.

  • Fig. 3 Normalized intrinsic noise is determined by combinations of promoter- and gene body–binding factors.

    The left side of the panel shows a heat map of promoter and the gene body (GB) localization of various factors with high and low intrinsic noise transcripts. The high intrinsic noise transcripts were classified into 10 clusters, and each cluster of high intrinsic noise transcripts and low intrinsic noise transcripts was subjected to OPLS-DA modeling. The right side of the panel represents score plots of OPLS-DA [the first predictive component (t1) versus the first orthogonal component (to1)] and S-plots constructed by presenting the modeled covariance (p[1]) against modeled correlation {p(corr)[1]} in the first predictive component. In clusters 5 and 6, the first orthogonal component was not significant (NS).

  • Fig. 4 CRISPR library screening of genes involved in intrinsic noise regulation.

    (A) Schematic diagram of CRISPR lentivirus library screening. Screening was performed independently for each of the three (Nanog, Trim28, and Dnmt3l) KI cell lines. (B) Ranked differentially expressed (DE) score plots obtained by performing CRISPR screening on three cell lines. The higher the DE score, the more the effect of enhancing intrinsic noise. (C) KEGG pathway enrichment analysis. KEGG pathway enrichment analysis was performed using clusterProfiler (see Materials and Methods), with the upper or lower 100 genes of DE score obtained from the CRISPR screening (referred as posi and nega, respectively). The pathways shown in red indicate hits in multiple groups of genes. Genes corresponding to these pathways are labeled in (B). (D) Simplified diagram of MAPK, Akt, and mTOR signaling pathways. These pathways are included in the pathways highlighted in red in (C) and cross-talk with each other. (E) Western blot of cells treated with signal pathway inhibitors. (F) Δnormalized intrinsic noise of cells treated with signal pathway inhibitors against control [dimethyl sulfoxide (DMSO)–treated] cells. Error bars indicate 95% confidence interval. (G) Twenty-four KI cell lines were conditioned to 2i or PD-MK conditions and subjected to flow cytometry analysis. Δnormalized intrinsic noise, Δburst size, and Δburst frequency against control (DMSO-treated) cells are shown. Error bars indicate 95% confidence interval.

  • Fig. 5 PD-MK conditioned mESCs retain pluripotency.

    (A) Growth curve of mESC conditioned to Std, 2i, and PD-MK conditions. Error bars indicate SD (n = 3). (B) Percentage of apoptotic cells of mESC conditioned to Std, 2i, and PD-MK conditions. Error bars indicate SD (n = 3). (C) Cell cycle distribution of mESCs conditioned to Std, 2i, and PD-MK conditions. (D) Immunofluorescence of pluripotency markers (NANOG, OCT4, and SSEA1) of mESCs conditioned to Std, 2i, and PD-MK conditions. The images show maximum intensity projections of stacks. Scale bar, 50 μm. (E) Chimeric mice with black coat color generated from C57BL/6NCr ES cells conditioned to the PD-MK condition and then to the Std condition before injection into albino host embryos. Photo credit: T. Okamura (National Center for Global Health and Medicine).

  • Fig. 6 Increase in the expression level of transcription elongation factors in the PD-MK condition.

    (A) Comparison of transcriptome of cells conditioned to Std, 2i, and PD-MK conditions. (B) GO analysis of genes whose expression significantly increased in the PD-MK condition against the 2i condition. BP, biological process; CC, cellular components; MF, molecular function. (C) Expression levels of genes encoding transcriptional elongation factors were elevated under the PD-MK condition as compared to the 2i condition. (D) Effect of RNA Pol II pause release inhibitor flavopiridol and SEC inhibitor KL-2 treatment on normalized intrinsic noise, burst size, and frequency. The KI cell lines conditioned to PD-MK were treated with flavopiridol or KL-2 for 2 days and analyzed by flow cytometry, and normalized intrinsic noise, burst size, and frequency were calculated. Δnormalized intrinsic noise, Δburst size, and Δburst frequency are residuals of normalized intrinsic noise, burst size, and frequency of inhibitor-treated cells from that of control cells (PD-MK condition), respectively. Error bars indicate 95% confidence interval. (E) Schematic summary of the determination of kinetic properties of transcriptional bursting (including intrinsic noise, burst size, and burst frequency) by a combination of promoter- and gene body–binding proteins, including PRC2 and transcription elongation factors.

Supplementary Materials

  • Supplementary Materials

    Genome-wide kinetic properties of transcriptional bursting in mouse embryonic stem cells

    Hiroshi Ochiai, Tetsutaro Hayashi, Mana Umeda, Mika Yoshimura, Akihito Harada, Yukiko Shimizu, Kenta Nakano, Noriko Saitoh, Zhe Liu, Takashi Yamamoto, Tadashi Okamura, Yasuyuki Ohkawa, Hiroshi Kimura, Itoshi Nikaido

    Download Supplement

    The PDF file includes:

    • Figs. S1 to S6
    • Legends for tables S1 to S4

    Other Supplementary Material for this manuscript includes the following:

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article