Glycogen synthase kinase-3 inhibition enhances translation of pluripotency-associated transcription factors to contribute to maintenance of mouse embryonic stem cell self-renewal

Abstract

Maintenance of embryonic stem cell (ESC) self-renewal and pluripotency are controlled by extrinsic factors, molecular signaling pathways and transcriptional regulators. While many of the key players have been studied in depth, how the molecular signals interact with transcription factors of the pluripotency network to regulate their action remains less well understood. Inhibition of glycogen synthase kinase 3 (Gsk-3) has been implicated in the maintenance of mouse ESC pluripotency, although there is contradictory data on its role, with enhancement of cell survival and metabolism, stabilisation of c-Myc and activation of Wnt signalling proposed as potential mechanisms. We have discovered that suppression of Gsk-3 activity leads to enhanced protein levels of key transcriptional regulators of the pluripotency network, notably Nanog, Tbx3 and c-Myc. Protein stability was unchanged following Gsk-3 inhibition, although interestingly, Nanog and Tbx3 proteins were found to have half-lives of 1-3 h, while that of Oct4 protein was longer, at 6 h. We demonstrate that the effects on protein levels seen following inhibition of Gsk-3 are due to both enhanced de novo synthesis of Nanog protein and increases in the proportion of Nanog and Tbx3 RNAs bound to polysomes, findings consistent with Gsk-3 regulating translation of these factors. These effects were not due to changes in regulators of general translation initiation machinery nor mediated via the 5' or 3' UTR sequences of Nanog alone. The data we present provide both new conceptual insight into the mechanisms regulated by Gsk-3 that may contribute to ESC self-renewal and, importantly, establish control of protein translation as an additional mechanism involved in modulation of ESC pluripotency.

Figure 1. Inhibition or ablation of Gsk-3 enhances expression of Nanog, Tbx3 and c-Myc proteins in ESCs.

E14tg2a wild-type (WT) and Gsk-3α/ß double knockout (DKO) mESCs were cultured in the presence of serum plus LIF (A) or chemically defined medium (N2B27) plus LIF and BMP4 (B). Gsk-3 inhibitors 1 m or CHIR99201 (CHIR) were added to WT ESCs at 2 and 3 µM respectively, as indicated in both A and B. (i) Images show colonies formed from WT ESCs that were untreated (WT), or cultured in the presence of 2 µM 1 m or 3 µM CHIR99201 for 48 h and Gsk-3 DKO ESCs. Protein and RNA were extracted at the times indicated. (ii) 12 µg of nuclear protein extracts were immunoblotted with the antibodies specified. (iii) Antibody signals were quantified and normalised to Gapdh. A value of one was given to normalised protein levels in WT ESCs at 8 hours and values for other samples related to these. The data show relative protein levels and are the average and SD of duplicate experiments for A and a single representative experiment in B. (iv) Quantitative RT-PCR was carried out and gene expression normalized relative to ß-actin levels. A value of one was given to normalised RNA levels in WT ESCs at 8 hr and other samples related to these. The data show relative gene expression levels and are the average and S.E.M of quadruplicate samples. *, p<0.05; **, p<0.01, ***, p<0.005. * for Nanog, #for Tbx3 and+for c-Myc (2-way ANOVA with Bonferroni post-hoc test). (C) Nanog-TNG reporter ESCs were cultured in serum and LIF in the presence of 3 µM CHIR99201 (CHIR) or DMSO (CON) and imaged over a period of 24 h using confocal microscopy, with images taken every 10 mins. Representative still images of ESC colonies are shown for the times specified. 10 fields of view chosen at random and within each field at least 15 colonies were imaged. White arrows indicate cells where Nanog reporter expression increases at early time points. Complete time-lapse series for control and CHIR-2 are available Videos S1 and S2.

Figure 2. Gsk-3 inhibition or ablation can maintain the expression of pluripotent markers upon LIF withdrawal.

(A) WT and Gsk-3 DKO ESCs were grown in serum-containing medium in the presence or absence of LIF and WT ESCs were also cultured in the absence of LIF in medium supplemented with 2 µM 1 m. Protein and RNA were extracted at the times indicated. (i) Immunoblotting of 15 µg of nuclear protein was performed with the indicated antibodies. (ii) Antibody signals were quantified and values normalised to Gapdh. A value of 1 was given to normalised protein levels in WT+LIF 24 h sample and values for other samples were related to these. The data show relative protein levels and are the average and SD of duplicate experiments. (iii) Quantitative RT-PCR was carried out and a value of 1 was given to normalised RNA levels in WT ESCs +LIF at 24 h and other samples related to these. The data show relative gene expression levels and are the average and S.E.M of quadruplicate samples. (B) (i) E14tg2a ESCs were grown for 24 hours in basal N2B27 medium without LIF or BMP4 in the presence or absence of PD (1 µM), CHIR (3 µM), 1 m (2 µM), CHIR plus PD or 1 m plus PD. Immunoblotting was performed with the antibodies specified. (ii) ESCs were grown for 16 h in basal N2B27 with 1 µM PD before either 3 µM CHIR or vehicle were added for 4, 24 and 48 hrs. Samples were immunoblotted with the antibodies indicated.

Figure 3. The ability of Gsk-3 inhibition to promote ESC proliferation is dependent upon the culture environment.

ESCs were grown in (A) GMEM supplemented with Serum and LIF; (B) N2B27 supplemented with BMP4 and LIF or (C) N2B27 without extrinsic stimuli, in the presence of Gsk3 inhibitor CHIR (CHIR99201) or MEK inhibitor PD (PD0325901) alone or in combination (2i) for 3 days and their growth measured. Control ESCs had DMSO vehicle added instead of inhibitors (CON). The data are the average and S.E.M of triplicate experiments. **, p<0.01, ***, p<0.005 (2-way ANOVA with Bonferroni post-hoc test).

Figure 4. Gsk-3 inhibition does not alter Nanog, Tbx3, Oct4 or c-Myc protein stability.

(A) E14tg2a WT ESCs and (B) WT and Gsk-3 DKO ESCs were grown in the presence of N2B27 plus LIF and BMP4 for 24 h. In (A) 1 m (2 µM) or vehicle alone control (CTL) were added to the cells for this 24 h period. All samples were then incubated with cycloheximide (CHX) to halt protein synthesis for the times indicated. (i) Protein samples were extracted after 0, 1, 3 and 6 hours of CHX treatment and immunoblotting performed with the indicated antibodies. (ii) Antibody signals were quantified and protein levels normalised to Gapdh. In each case a value of 100% was given to the t = 0 samples to enable direct comparisons to be made. The data are the average relative protein expression levels and S.E.M of triplicate experiments.

Figure 5. Gsk-3 inhibition or ablation increases Nanog protein synthesis in the absence of increased transcription.

WT and Gsk-3 DKO ESCs were incubated in (A) GMEM supplemented with LIF and serum or (B) N2B27 medium supplemented with LIF and BMP4 prior to addition of 10 mg/ml CHX for 4 hours to halt protein synthesis. CHX was then washed out and fresh media (A, GMEM plus LIF plus serum and B, N2B27 plus LIF and BMP4) added back. (i) & (ii) Protein and (iii) RNA were extracted between 1 and 4 hours after CHX washout. (i) Immunoblotting was performed with the antibodies indicated. (ii) Antibody signals were quantified and normalised to Gapdh. A value of 1 was given to CHX-treated samples. The data are the average relative protein expression levels and SD of duplicate experiments. (iii) Quantitative RT-PCR was carried out and Nanog expression normalized relative to ß-actin levels. A value of 1 was given to CHX-treated samples in (iii). The data are the average gene expression levels and S.E.M of quadruplicate samples. (C) Nanog reporter Nd ESCs were sorted by flow cytometry and cells with low to no VNP expression collected and grown in GMEM supplemented with LIF and serum in the presence or absence of DMSO (controls) or in the presence of 3 µM CHIR99201. The data shown are the average Nanog:VNP expression levels (relative to DMSO control) and standard deviation of at least 3 biological replicates. All p-values were calculated using a two-tailed distribution, two-sample equal variance t-test (** p-value <0.01; *** p-value <0.0001).

Figure 6. Association of Nanog and Tbx3 RNAs with polysomes increases upon inhibition of Gsk3.

WT ESCs grown in the absence (WT) or presence of 2 µM 1 m (1 m) and Gsk-3 DKO ESCs (DKO) were cultured in serum supplemented with LIF for 4 and 8 hours before extracting cell lysates. Sedimentation through sucrose gradients was used to separate the polysomal-enriched fractions from the monosomal fractions. The levels of RNA bound to polysomal or monosomal fractions were measured using quantitative RT-PCR, with the primers shown in Table S1. In each case gene expression was normalized relative to ß-actin levels. A. Expression of (i) Nanog, (ii) Tbx3 and (iii) Oct4 in individual fractions across the sedimentation gradient. A representative experiment is shown in each case (4 h time points for (i) and (ii) and 8 h time point for (iii)). Values show expression relative to ß-actin. B. Gene expression in pooled monosomal and polysomal fractions. Values show the proportion of RNA bound to the polysome fraction (Bound/Total RNA). The data are the average and S.E.M of three independent experiments run in duplicate for the 8-hour time point and the average and S.E.M of two independent experiments run in duplicate for the 4-hour time point. *, p<0.05; **, p<0.01, ***, p<0.005 (Student’s t-test).

Figure 7. Gsk-3 does not mediate its effects via general translational machinery or the 5′UTR of Nanog.

(A) E14tg2a wild-type (WT) and Gsk-3α/ß double knockout (DKO) ESCs were cultured in the presence of LIF and serum. Gsk-3 inhibitor 1 m was added to WT cells at 2 µM and protein samples taken at the times indicated. Cell lysates were blotted with the antibodies indicated. (B) WT and Gsk-3 DKO ESCs were grown in N2B27 plus BMP4 and LIF. WT were treated with 2 µM 1 m or 3 µM CHIR for 8 and 24 hours before lysis. Immunoblotting was performed with antibodies against pT389 S6K1 and Gapdh. (C) E14tg2a ESCs were grown overnight in the presence of MEK inhibitor (PD) before being treated with 3 µM CHIR or 2 µM 1 m for the times indicated. ESCs were also grown in either CHIR or 1 m alone overnight prior to preparation of cell lysates. Immunoblotting to detect phosphorylation of T389 on S6K1 and Gapdh were performed. (D)(i) The 215bp 5′UTR sequence of the mouse Nanog gene (RefSeq NM_028016). Polypyrimidine tracts, shown in blue, were identified at bp positions 122, 137 and 191, with the tract at 137 containing 1 CCT motif and the tract at position 191 containing 2 CCT motifs (in bold font). One potential uORF, underlined, with an ATG start codon was identified between bp133 to 156. WT E14 ESCs were transfected with (ii) pGL3’ or pGL3’-Nanog-5′UTR in the presence of pRenilla luciferase plasmid or (iii) pGL3-MCS, pGL3-Nanog-3′UTR R1 or pGL3-Nanog-3′UTR R2 in the presence of pRenilla luciferase plasmid. After 24 h, transfected cells were treated with vehicle alone (C) or 2 µM 1 m. Cells were harvested at 4, 8 and 16 h and Firefly luciferase activity determined and normalised relative to Renilla luciferase activity. Luciferase activity is expressed as fold-change over pGL3’ (ii) or pGL3-MCS (iii) controls (DMSO) for a given time point. Data are the average and S.E.M. of three independent experiments.