A genome-wide RNAi screen reveals MAP kinase phosphatases as key ERK pathway regulators during embryonic stem cell differentiation

Abstract

Embryonic stem cells and induced pluripotent stem cells represent potentially important therapeutic agents in regenerative medicine. Complex interlinked transcriptional and signaling networks control the fate of these cells towards maintenance of pluripotency or differentiation. In this study we have focused on how mouse embryonic stem cells begin to differentiate and lose pluripotency and, in particular, the role that the ERK MAP kinase and GSK3 signaling pathways play in this process. Through a genome-wide siRNA screen we have identified more than 400 genes involved in loss of pluripotency and promoting the onset of differentiation. These genes were functionally associated with the ERK and/or GSK3 pathways, providing an important resource for studying the roles of these pathways in controlling escape from the pluripotent ground state. More detailed analysis identified MAP kinase phosphatases as a focal point of regulation and demonstrated an important role for these enzymes in controlling ERK activation kinetics and subsequently determining early embryonic stem cell fate decisions.

Figure 1. A genome-wide RNAi library screen identifies factors involved in signal-dependent embryonic stem cell differentiation.

(A) Schematic representation of primary screen and secondary validation screens using mouse embryonic stem cell lines containing GFP reporter constructs under the control of either the endogenous rex1 (Rex1GFPd2) and/or oct4 (Oct4GFP) promoters. “−2i” indicates that the two kinase inhibitors (CHIR99021 and PD0325901) were removed for 28 hrs (Rex1GFPd2 cells) and 72 hrs (Oct4GFP cells) before quantifying the GFP-positive population of cells. The Venn diagram shows 316 high confident hits resulting from the overlap of both validation screens. (B) The z-score of each of two biological replicates from the primary screens are plotted against each other (left panel). The average of ranked z-scores from the knockdown of individual siRNA pools is shown (right panel). The arrows indicate the z-score threshold (2 and −2). The green and red boxes mark the hits with z-score >2 and <−2, respectively. (C) Representative FACS profiles from control non-targeting and positive siRNA hits from the primary rex1-GFP(d2) screen. The high GFP expressing population is depicted in dark green and numbers above each graph are the corresponding GFP high/GFP low ratios. (D and E) The log₂ ratio of GFP(+)/GFP(−) of each of the two biological replicates are plotted against each other in either the rex1-GFP(d2) (D) or oct4-GFP (E) validation screens (left panels) and graphical representations of the ranked log₂ values of these ratios from the average of two independent experiments upon knockdown of individual genes are shown (right panels). The yellow shaded boxes indicate the positive hits which scored as a GFP(+)/GFP(−) ratio above 1.25× standard deviation of the controls in each of the sub-screens. (F and G) Enriched KEGG (F) and “molecular function” level GO terms (G) amongst the high confidence hits identified in both of the secondary validation screens.

Figure 2. Secondary screening and association of genes to the ERK and/or GSK3 pathways.

(A) Schematic representation of the strategy used to stratify the high confidence positive hits via either the ERK and/or GSK3 pathways (1i-screens) using the rex1-GFP reporter system. A counter screen was performed in the presence of the two kinase inhibitors (CHIR99021 and PD0325901)(“+2i”) and remaining hits were tested when either the GSK3 inhibitor (CHIR99021; -GSKi) or the MEK inhibitor (PD0325901; -MEKi) was withdrawn. The Venn diagram illustrates the four different hit categories. (B) Heatmap summary depicts the stratification of the hits according to their effects on the GFP(+)/GFP(−) ratio upon withdrawal of the GSK3 inhibitor (-GSK3i) or MEK inhibitor (-MEKi). The numbers of genes in each category are indicated. The colour scale represents the ratio of high to low GFP expressing cells for each siRNA pool in each of the “1i” screens. (C and D) Heatmaps of the enriched GO terms identified for genes corresponding to hits specific to the total dataset from the “2i” screen (274), or hits from the “1i” screens; ERK (133), or GSK3 (168) pathways. Each GO term is scored by −log₁₀(P-value). The associated GO term descriptions are indicated (the GO terms enriched in only the ERK or GSK3 categories are indicated in red or green font respectively). Distinct functional groups corresponding to terms associated with cell signaling (C) and gene expression (D) are manually clustered. (E) STRING network analysis of the core network formed by the 274 genes associated with signal-dependent loss of pluripotency and promoting early differentiation processes in the mouse embryonic stem cells. Genes are grouped according to common biological processes. The coloured lines of edges represent confidence scores of interconnectivity. Dark blue lines represent 0.8–1, light blue lines represent 0.6–0.8, and light grey lines represent 0.4–0.6 confidence levels, respectively.

Figure 3. The role of ERK pathway-specific hits in the expression of pluripotency and early differentiation marker genes.

(A) Venn diagram (top) and heatmap summary (bottom; see Figure 2B for details) illustrating the number and a list of selected screen hits used in the subsequent studies. These selected hits are distributed within three categories as indicated on the heatmap. The colour scale represents the ratio of high to low GFP expressing cells for each siRNA pool in each of the “1i” screens. (B) rex1 (x-axis) and nanog (y-axis) mRNA expression levels following 2i withdrawal for 36 hrs are plotted upon knockdown of individual genes (see Figure S9B for details). Data are shown for each siRNA duplex relative to the maximal expression exhibited in the presence of an siRNA pool (taken as 100). Dotted lines represent the expression values >2 standard deviations above the mean of the negative control siRNAs. Red dots represent siRNA duplexes which promote elevated expression of both genes (quadrant 1), whereas green (quadrant 3) and black (quadrant 4) dots represent siRNAs that cause changes at or below this threshold cut-off value for only one gene. The brown dot represents the negative control siRNAs. (C) rex1 (y-axis) and the reciprocal of fgf5 (x-axis) mRNA expression levels upon 2i withdrawal for 36 hrs and 48 hrs, respectively, are plotted following knockdown of individual genes (see Figure S9B and C for details). The labeling is as indicated in (B), except that red dots represent siRNA duplexes which promote elevated expression of rex1 and lower levels of fgf5 (quadrant 1). Blue dots (quadrant 2) represent siRNAs that cause elevated rex1 expression but fail to show reductions in fgf5 expression. (D) Alkaline phosphatase staining of Rex1GFPd2 ES cells following treatment of cells with siRNAs against gmnn or 3830406c13rik or a non-targeting control (ctrl) and release from “2i” for 5 days. Data are means ± SEM (n = 2) (E) RT-PCR analysis of the expression of the indicated lineage marker genes following treatment of cells with siRNAs against gmnn or 3830406c13rik or a non-targeting control (ctrl) and release from “2i” for 3 (top) or 5 days (bottom). Data are presented as means ± SEM (n = 2).

Figure 4. Association of siRNA screen hits with the ERK signaling pathway.

(A) ERK activation levels following 2i withdrawal (−2i) for 20 mins are plotted as the ratio of phospho-ERK2 and ERK2 signals upon depletion of selected genes as indicated. The blue dashed line indicates the threshold level (1.5× SD above the mean of the negative controls) and levels below this are indicated by red bars. The average activity in the presence of control siRNA (ctrl) is shown by the solid grey line and data are plotted relative to the siRNA giving the highest levels of phospho-Erk (taken as 100). The data are presented as means ± SEM and are the average of three biological replicates (n = 3). A heatmap summary of the effect of each siRNA duplex in the “1i” screens is shown on the left. (B) Summary of the points of action of the siRNA screen hits with respect to the ERK pathway. Genes are partitioned according to which class of siRNA hits they belong. (C) Ras activity levels upon depletion of the indicated genes upon 2i withdrawal (−2i) for 2 mins. The blue dashed line indicates the threshold level (2× SD above the mean of the negative controls) and levels below this are indicated by red bars. The average activity in the presence of control (ctrl) siRNA is shown by the solid grey line and data are plotted relative to the siRNA giving the highest levels of Ras activity (taken as 100). Data are presented as means ± SEM and are the average of three biological replicates (n = 3). A heatmap summary of the effect of each siRNA duplex in the “1i” screens is shown on the left. (D) Summary diagram illustrating the point of action of upstream ERK effectors in the ERK pathway as either upstream of Ras or between Ras and Erk. The hit lists are grouped into the ERK-unique or ERK/GSK-shared hit categories. (E) Summary of the points of action of genes encoding transcriptional regulators respect to the ERK pathway.

Figure 5. The regulation of Dusp1 and Dusp6 activity and ES cell differentiation.

(A–C) RT-qPCR analysis of dusp1 and dusp6 mRNA expression in mouse ES cells. Data are normalised by the average values of three reference genes (ref) and are presented as means ± SEM and are the average of three biological replicates (n = 3). (A) The kinetics of dusp1 and dusp6 expression at the indicated times following 2i withdrawal. (B) The effects of the indicated inhibitors, either alone or in combination, on the expression of dusp1 and dusp6 at the indicated times following inhibitor withdrawal. (C) The effects of depletion of the indicated genes on dusp1 and dusp6 mRNA expression in the presence of 2i. The blue dashed line indicates the threshold level (2× SD above the mean of the negative controls) and levels below this are indicated by red bars. The average activity in the presence of control siRNA (ctrl) is shown by the solid grey line (taken as 100). (D and E) Active ERK levels were determined by the ratio of phospho-ERK (pERK)/total ERK (ERK) levels at the indicated times following 2i release in the presence of the indicated siRNAs (red lines) or control siRNA (blue lines). The data are plotted relative to maximal levels with the control siRNA (taken as 100) and are presented as means ± SEM from the average of two biological replicates (n = 2). (F and G) The change in the ratio of GFP negative to GFP positive Rex1GFPd2 cells 28 hrs after 2i withdrawal in the presence of the siRNAs against the indicated dusps relative to control siRNAs is shown. Data are the average of two biological replicates. (H) RT-qPCR analysis of the changes in nanog mRNA expression in Rex1GFPd2 cells upon 2i withdrawal for 28 hrs and depletion of the indicated dusps. The data are normalised by the average of three reference genes, and presented relative to control siRNAs. Data are presented as means ± SEM and are the average of two biological replicates (n = 2). (I) RT-qPCR analysis of the expression of the indicated lineage marker genes following treatment of cells with siRNAs against dusp1 or dusp6 or a non-targeting control (ctrl) and release from “2i” for 3 (top) or 5 days (bottom). Data are presented as means ± SEM (n = 2). (J) Summary diagram illustrating the key regulatory role of the Dusps in mediating the action of the ERK pathway in early cell fate decisions during loss of pluripotency and onset of differentiation.