CRISPR reveals a distal super-enhancer required for Sox2 expression in mouse embryonic stem cells

Abstract

The pluripotency of embryonic stem cells (ESCs) is maintained by a small group of master transcription factors including Oct4, Sox2 and Nanog. These core factors form a regulatory circuit controlling the transcription of a number of pluripotency factors including themselves. Although previous studies have identified transcriptional regulators of this core network, the cis-regulatory DNA sequences required for the transcription of these key pluripotency factors remain to be defined. We analyzed epigenomic data within the 1.5 Mb gene-desert regions around the Sox2 gene and identified a 13kb-long super-enhancer (SE) located 100kb downstream of Sox2 in mouse ESCs. This SE is occupied by Oct4, Sox2, Nanog, and the mediator complex, and physically interacts with the Sox2 locus via DNA looping. Using a simple and highly efficient double-CRISPR genome editing strategy we deleted the entire 13-kb SE and characterized transcriptional defects in the resulting monoallelic and biallelic deletion clones with RNA-seq. We showed that the SE is responsible for over 90% of Sox2 expression, and Sox2 is the only target gene along the chromosome. Our results support the functional significance of a SE in maintaining the pluripotency transcription program in mouse ESCs.

Figure 1. Identification of a distal mouse ESC-specific super-enhancer near Sox2 gene.

In a genome-browser snapshot, 23 tracks of H3K27ac ChIP-seq data are shown. The top 2 tracks are from two different mouse ESC lines. Shadowed region shows the location of the mEnh-Sox2^distal enhancer.

Figure 2. Characterization of the distal Sox2 super-enhancer.

In a genome-browser snapshot, the top Hi-C track is a bar graph showing the Hi-C read number in mouse ESCs using Sox2 as the anchor (highlighted in yellow). Each bar represents one HindIII fragment between two HindIII cutting sites. The bar height represents the Hi-C read count from the anchor to the HindIII fragment. The 2^nd Hi-C track plots the expected value for every HindIII fragment. Therefore the blue highlighted region is a significant looping interaction from the Sox2 promoter to Sox2-SE^distal after comparing the top two tracks. The ChIP-seq tracks show the locations of different histone marks or protein factors in mouse ESC lines. The super-enhancer track shows the locations of two super-enhancers in mESCs identified by Hnisz et al.

Figure 3. Deleting Sox2-SE^distal using CRISPR.

(A) Schematic representation of CRISPR-based SE deletion strategy in a hybrid mouse ESC line, F123. Lower panel shows the location of two small guide RNA flanking the 13kb Sox2-SE^distal. (B) Two PCR primers flanking (see Methods) the deleted region were used to amplify the genomic DNA from mutant clones. Successful deletion (monoallelic or biallelic) generates a small PCR product of ∼420bp with some length variation due to random mutations during non-homologous end-joining. 8 of the 16 clones show a product, meaning at least one copy of the SE is deleted. (C) qPCR was performed using two primers inside the SE (see Methods) to measure the copy number of the SE in WT ESCs and 8 mutant clones. Signal in WT clone is normalized to 2. Error bar: standard deviation. (D) Summary of the genotypes of the mutant ESC clones as confirmed by Sanger sequencing.

Figure 4. Sox2-SE^distal regulates Sox2 expression in cis.

(A) Allele-specific RT-qPCR analysis of mESC clones with different genotypes. Samples are normalized to the average Sox2 signal between two WT biological replicates. Error bar denotes standard deviation. Note one biallelic Sox2-SE^distal deletion clone (Mut B from Fig. 3 ) is missing because we fail to re-expand that clone after freeze and thawing. (B) Genome browser tracks showing the RNA-seq signals near the Sox2 locus. Only RNA-seq reads on SNPs between CAST and 129 genomes are plotted. Red tracks: RNA-seq reads on CAST SNPs, the signal above the base line shows transcription from positive strand and the signal below the base line shows transcription from negative strand. Blue tracks: RNA-seq reads on 129 SNPs. The bottom heat map track shows TAD defined by Hi-C. The TAD containing Sox2 gene is highlighted in yellow. (C) Comparison of allele-specific expression from RNA-seq in the monoallelic 129 and monoallelic CAST clones. Allele specificity is defined as the log2 ratio of RNA-seq reads from CAST and 129 allele after adding a pseudo count of 10. All genes on chromosome 3 including Sox2 are plotted. Only the Sox2 gene shows CAST specific expression in CAST+/129- clone and 129 specific expression in CAST-/129+ clone, suggesting direct regulation by Sox2-SE^distal.

Figure 5. Sox2-SE^distal deletion affects key pluripotency genes.

(A) Alkaline phosphatase staining of wild type (left) and biallelic Sox2-SE^distal deletion (right) clones. (B) RT-qPCR analysis of pluripotency genes in 4 representative wild type (WT) or Sox2-SE^distal deletion clones. (C–E) Scatter plots comparing the expression levels of all mouse genes in different Sox2-SE^distal deletion clones according to RNA-seq data. Knockout clone: biallelic Sox2-SE^distal mutant; haploinsufficient clone: monoallelic Sox2-SE^distal mutant. Plotted are log2 scaled FPKM values for each gene after adding a pseudo-value of 4. Signature up-regulated (red spots) or down-regulated (blue spots) genes were identified from the biallelic mutant clone (C). Expression levels of these signature genes were also compared in (D) and (E). (F) Bar graph comparing the up-regulated and down-regulated signature genes with Sox2 binding at promoters as called by Sox2 ChIP-seq data. P-value: Binomial test. (G) GO analysis identified top GO terms enriched in up-regulated or down-regulated genes upon Sox2-SE^distal deletion. (H) Venn diagrams comparing the Sox2-SE^distal knockout signature genes to signature genes found in Suz12 knockout mouse ESCs. P-value: binomial test.