A genome-wide RNAi screen identifies a new transcriptional module required for self-renewal

Abstract

We performed a genome-wide siRNA screen in mouse embryonic stem (ES) cells to identify genes essential for self-renewal, and found 148 genes whose down-regulation caused differentiation. Many of the identified genes function in gene regulation and/or development, and are highly expressed in ES cells and embryonic tissues. We further identified target genes of two transcription regulators Cnot3 and Trim28. We discovered that Cnot3 and Trim28 co-occupy many putative gene promoters with c-Myc and Zfx, but not other pluripotency-associated transcription factors. They form a unique module in the self-renewal transcription network, separate from the core module formed by Nanog, Oct4, and Sox2. The transcriptional targets of this module are enriched for genes involved in cell cycle, cell death, and cancer. This supports the idea that regulatory networks controlling self-renewal in stem cells may also be active in certain cancers and may represent novel anti-cancer targets. Our screen has implicated over 100 new genes in ES cell self-renewal, and illustrates the power of RNAi and forward genetics for the systematic study of self-renewal.

Figure 1.

Genome-wide RNAi screen for self-renewal genes in mouse ES cells. (A) Outline of the screen. (Dark wells) Differentiated cells that are GFP-negative; (green wells) self-renewing cells that are GFP-positive. (B) Functional categorization of candidate self-renewal genes identified from the screen. (Left panel) Molecular functions (Panther Classification System). (Right panel) Biological processes (Ingenuity Pathway Analysis). Only selected categories were shown. For complete lists, see Supplemental Table S2. (C) Expression of candidate self-renewal genes during ES cell differentiation. (Right panel) Differentiation after 4 d in RA. (Left panel) Differentiation after 4 d of EB formation. (D) Relative expression level of candidate self-renewal genes in 61 different tissues. Genes were clustered based on their expression levels in embryonic tissues, nervous system, and other tissues.

Figure 2.

Validation of selected self-renewal genes identified in the screen. (A) Oct4GiP reporter assays after knockdown of candidate self-renewal genes. Oct4GiP ES cells were transfected with siRNAs against candidate self-renewal genes (two different siRNAs for each gene). Percent of differentiation (% Differentiation) was defined by percentage of GFP-negative cells, and was measured 96 h after transfection by FACS. (B) Knockdown of candidate self-renewal genes causes down-regulation of ES cell marker genes. ES cells were transfected with siRNAs against candidate genes and cells were collected 96 h after transfection. Level of Oct4, Nanog, and Sox transcripts were measured by qPCRs from cDNAs prepared from the transfected cells.

Figure 3.

Morphological changes and loss of alkaline phosphatase staining after the silencing of candidate genes. ES Oct4GiP cells were transfected with siRNAs against candidate self-renewal genes and replated at low density on the second day. Cells were cultured in ES cell medium, and alkaline phosphatase staining was performed 5 d after plating.

Figure 4.

Identification of Cnot3 and Trim28 target genes. (A) Chromosome view of examples of Cnot3 and Trim28 target genes. (B) Distribution of Cnot3- and Trim28-binding sites from target gene TSSs. (C) Consensus binding motif of Cnot3 and Trim28. (D) H3K4 and H3K27 methylation in promoters of Cnot3 and Trim28 target genes. (Fold enrichment) Percentage of H3 methylation in Cnot3 or Trim28/percentage of H3 methylation in the genome. (Asterisks) P < 10⁻⁸, calculated by hypergeometric distribution. (E) Venn diagram of target genes co-occupied by Cnot3 and Trim28. (F) Histogram of distance between Cnot3- and Trim28-binding sites in genes occupied by both. (G) GO analysis (by Ingenuity Pathway Analysis) of Cnot3 and Trim28 target genes. Only selected functional categories were shown. For a complete list, see Supplemental Table S3.

Figure 5.

The Cnot3, Trim28, Myc, and Zfx module in the self-renewal network. (A) Correlations of target gene occupancies between pluripotency genes. (B) Distances between Cnot3-, Trim28-, c-Myc-, and Zfx-binding sites in gene promoters occupied by any two of the four genes. Distance between Trim28- and Nanog-binding sites in their common targets was shown as a negative control. (C) Expression pattern of genes co-occupied by Cnot3, Trim28, c-Myc, and Zfx during EB differentiation. Genes were ordered based on their expression level in ES cells (0 h of EB differentiation). (D) Change in expression levels of genes occupied by all four factors (Cnot3, Trim28, Myc, and Zfx) between ES cells and 14-d EB. (Left panel) Heat map of log2 ratios between 14-d EB and ES for all genes detected on the array (All) and for genes co-occupied by all four factors (CTMZ). (Right panel) Average log2 ratios between 14-d EB and ES for all genes (All) and for genes co-occupied by four factors (CTMZ). (E) GO analysis (Ingenuity Pathway Analysis) of common target genes of Cnot3–Trim28–c-Myc–Zfx (left) and Nanog–Sox2–Oct4 (right). Only selected functional categories were shown. For complete lists, see Supplemental Table S4.