N6-methyladenosine (m6A) depletion regulates pluripotency exit by activating signaling pathways in embryonic stem cells

Abstract

N6-methyladenosine (m⁶A) deposition on messenger RNA (mRNA) controls embryonic stem cell (ESC) fate by regulating the mRNA stabilities of pluripotency and lineage transcription factors (TFs) [P. J. Batista et al., Cell Stem Cell 15, 707-719 (2014); Y. Wang et al., Nat. Cell Biol. 16, 191-198 (2014); and S. Geula et al., Science 347, 1002-1006 (2015)]. If the mRNAs of these two TF groups become stabilized, it remains unclear how the pluripotency or lineage commitment decision is implemented. We performed noninvasive quantification of Nanog and Oct4 TF protein levels in reporter ESCs to define cell-state dynamics at single-cell resolution. Long-term single-cell tracking shows that immediate m⁶A depletion by Mettl3 knock-down in serum/leukemia inhibitory factor supports both pluripotency maintenance and its departure. This is mediated by differential and opposing signaling pathways. Increased FGF5 mRNA stability activates pErk, leading to Nanog down-regulation. FGF5-mediated coactivation of pAkt reenforces Nanog expression. In formative stem cells poised toward differentiation, m⁶A depletion activates both pErk and pAkt, increasing the propensity for mesendodermal lineage induction. Stable m⁶A depletion by Mettl3 knock-out also promotes pErk activation. Higher pErk counteracts the pluripotency exit delay exhibited by stably m⁶A-depleted cells upon differentiation. At single-cell resolution, we illustrate that decreasing m⁶A abundances activates pErk and pAkt-signaling, regulating pluripotency departure.

Keywords: formative stem cells; m6A; pluripotency; signaling; single-cell resolution.

Fig. 1.

Immediate m⁶A depletion promotes Nanog-state heterogeneity in mESCs. (A–J) mESCs were transduced with control shRNA (Scr) or shRNAs targeting Mettl3 (shM3-1 and shM3-2). (A) Mettl3, Mettl14, and Wtap immunoblots. GAPDH is a loading control. (B) Reduction in m6A levels on mRNA upon Mettl3 KD by m6A dot blot (Left) and quantification (Right). Methylene blue is a loading control. (C) Mettl3 KD resulted in flatter colony morphology in NanogVENUS mESCs. (Scale bars, 100 μm.) (D) Flow cytometric analysis of Nanog^VENUS-negative and -positive cell percentages upon Mettl3 KD in Nanog^VENUS mESCs. “Relative counts” indicates normalized cell counts as cumulative percentages up to 100%. (E) Proportions of Nanog^VENUS-negative cells upon Mettl3 KD, as quantified by flow cytometry using n = 9 independent repeats. (F) Representative immunostaining for Oct4 (cyan), Sox2 (magenta), Nanog (yellow), and DAPI (blue, cell nuclei) for Scr and shM3-1 cells. (Scale bars, 50 μm.) (G and H) Multispectral quantitative analysis of immunostaining in F. Scatter plot showing fluorescence intensity per cell in arbitrary units (a.u.) for Oct4 versus Nanog (G) and Sox2 versus Nanog (H). Dashed gray lines mark electronic gates (Methods) delineating positive and negative cells. Contour lines indicate probability distribution. Percentages of cells in each compartment are annotated. Analysis of n = 6,239 (Scr), n = 5,169 (shM3-1), and n = 5,257 (shM3-2) cells. (I and J) Relative mRNA levels (Left) and mRNA half-lives (Right) for Nanog (I) and Esrrb (J) in control and m6A-depleted mESCs. mRNA lifetime was determined by monitoring transcript abundance after transcription inhibition (TI), detailed in Methods. All statistics include error bars indicating mean ± SD and were calculated using two-tailed independent t test and unequal variance, *P < 0.1, **P < 0.05, and ***P < 0.01. (A) n = 3 (B), n = 9 (D), n = 3 (F–H), and n = 3 (I and J) independent repeats. Unless stated otherwise, Nanog^VENUS cells were used for experiments and analyzed 10 d after transductions.

Fig. 2.

m⁶A depletion sustains Nanog-positive and -negative state accumulation. (A) Targeting strategy for Nanog^KATUSHKA/Oct4^VENUS double knock-in reporter mESCs. White rectangles denote exons, and asterisks denote stop codons. Southern blotting verified correct targeting of FPs to the C termini of Nanog and Oct4; NK: Nanog^KATUSHKA, OV: Oct4^VENUS, and WT: wild-type. (B) Cells were transduced with control shRNA (Scr) or shRNAs targeting Mettl3 (shM3-1 and shM3-2). Proportions of Nanog^KATUSHKA-negative (Left) and Oct4^VENUS-positive (Right) cells upon Mettl3 KD as measured by flow cytometry. Error bars indicate mean ± SD from n = 5 independent repeats. (C–G) Cells were transduced with Scr or shM3-1. (C) Representative contour plots showing the intensity of Nanog^KATUSHKA against Oct4^VENUS in a representative live-cell imaging experiment. Cell cycle–corrected intensity data (Methods) is shown. (D) Transition kernels showing Nanog intensity transitions within one generation between mother and daughter cells. Contour lines indicate the probability distribution of all mother-to-daughter transitions. Representative data showing 844 (Scr) and 956 (shM3-1) mother-to-daughter transitions. (E) Positive and negative compartment probability matrices summarize the probabilities for all four types of mother-to-daughter intensity transitions. (In total, 3,058 [Scr] and 3,231 [shM3-1] transitions were analyzed from three independent experiments). (F) Nanog^KATUSHKA-positive and -negative compartment exit dynamics over time. Dotted and solid lines denote Nanog^KATUSHKA-positive and -negative cells, respectively. Each line shows an independent experiment. (G) Representative heat trees of Nanog^KATUSHKA/Oct4^VENUS FP levels in single cells over many cell generations. Each square denotes a cell, and gray scale intensity shows the concentration of Oct4VENUS or Nanog^KATUSHKA intensity (Methods) for a representative Scr or shM3-1 cell lineage. Colonies from Nanog^KATUSHKA-negative founder cells were observed for up to 85 h. At time point 55 h, representative live-cell imaging is shown. White circles denote cell nuclei. Numbered nuclei correspond to numbered cells in the heat tree. The iRFP channel denotes nucmem-iRFP, a constitutively expressed nuclear membrane marker C-terminally fused to the iRFP713 FP. Nucmem-iRFP was engineered into all lentiviral shRNA constructs and used to delineate nuclear area. a.u.: arbitrary unit. All live-cell imaging experiments were performed on Nanog^KATUSHKA/Oct4^VENUS reporter mESCs with three independent experimental repeats.

Fig. 3.

m⁶A depletion activates Mek-pErk promoting Nanog-negative state accumulation. (A–H) mESCs were transduced with control shRNA (Scr) or shRNAs targeting Mettl3 (shM3-1 and shM3-2). (A) Relative mRNA level (Left) and mRNA half-life (Right) for Fgf5. (B) Immunoblot for total Fgfr1 and phosphorylated Fgfr1 (pFgfr1). Quantification of immunoblot shown on the right. (C and D) Immunoblot for total Mek1/2 and phosphorylated Mek1/2 (pMek1/2) (C) and total Erk1/2 and phosphorylated Erk1/2 (pErk1/2) (D). Quantification of immunoblots is shown on the Right. (E and F) Immunostaining for pErk1/2 (magenta), Nanog (yellow), and DAPI (blue, cell nuclei). White arrows indicate only pErk-positive cells; yellow arrows indicate only Nanog-positive cells; light-blue arrows indicate pErk and Nanog double-positive cells. (Scale bars, 50 μm.). (F) Corresponding contoured, quantitative scatter plot of imaging data. Percentages of cells in each compartment are annotated. Analysis of n = 169 (Scr), n = 186 (shM3-1), and n = 205 (shM3-2) cells. A representative plot is shown. a.u.: arbitrary unit. (G) Cells were treated with 1 μM of Mek inhibitor PD0325901 for 4 d and analyzed by flow cytometry. Proportions of Nanog^VENUS-negative cells are labeled. (H) Flow cytometry quantification of Nanog^VENUS-negative cells upon PD0325901 treatment over 4 d. Statistical significance was calculated using two-tailed independent t test with unequal variance, *P < 0.1, **P < 0.05, and ***P < 0.01. Error bars indicate mean ± SD from (A–D) n = 5 and (E–H) n = 3 independent experiments. pMek1/2: phospho-Mek1/2(Ser217/221); and pErk1/2 (pErk): phospho-Erk1/2(Thr202/Tyr204). All experiments were performed with Nanog^VENUS mESCs. DMSO, dimethyl sulfoxide.

Fig. 4.

m⁶A depletion activates signaling regulating epiblast cell identity. (A and B) mESCs were transduced with control shRNA (Scr) or shRNAs targeting Mettl3 (shM3-1 and shM3-2). Multispectral protein-level quantification of immunostaining for TFs Nanog against Otx2 n = 5,118 cells (Scr), n = 3,382 cells (shM3-1), and n = 4,326 cells (shM3-2) (A) and Oct6 n = 3,545 cells (Scr), n = 3,815 cells (shM3-1), and n = 6233 cells (shM-3-2) (B). (A) Dashed rectangle and percentage denotes the proportion of Otx2-positive cells. (B) Dashed lines separate quadrants of positive and negative cell proportions denoted by percentages. (C) Immunoblot analysis for Otx2 and Oct6 protein levels after Mek-pErk inhibition (PD0325901, 1 μM) or vehicle control (dimethyl sulfoxide [DMSO]). (D) Immunoblots showing Oct6 and/or Otx2 KD using siRNAs in m⁶A-depleted cells. (E) Representative flow cytometry histograms for Nanog^VENUS mESCs treated with siRNA-targeting Oct6 (siOct6) and/or siOtx2 (siOtx2) in m⁶A-depleted cells. Nanog^VENUS-negative cell fractions are labeled. (F) Quantifications of siOct6 and siOtx2 effects on Nanog^VENUS-negative cell fractions. (G and H) Relative mRNA levels (Left) and half-lives (Right) for Otx2 (G) and Oct6 (H) in control and m⁶A-depleted mESCs. Analyses were performed on Nanog^VENUS cells. (I) FS cells were induced from 2i/LIF grown R1 wild-type mESCs and cultured for 10 d. Thereafter, FS cells were transduced with Scr or shM3-1 and shM3-2 followed by m⁶A depletion for a further 10 d in FS maintenance conditions. Immunoblot showing Mettl3, Mettl14, Oct4, Otx2, and Oct6 protein expression levels upon Mettl3 KD. (J) Immunoblot showing total Akt, phosphorylated Akt (pAkt^S473), total Erk1/2, phosphorylated Erk1/2 (pErk1/2), and GAPDH in m⁶A-depleted FS cells. (K) mRNA levels quantified for mesendodermal markers Mixl1, Mesp1, and Brachyury T. Levels are compared in FS maintenance conditions (FS) and after 24 h of mesendodermal differentiation (Mesendoderm). The magnitude of change in gene expression is plotted relative to the mesendodermal Scr control condition delineated as 1. Error bars indicate mean ± SD from (D–F) n = 4, (G–J) n = 3, and (K) n = 5 independent experimental repeats. Significance is measured by two-tailed independent t test using unequal variance, **P < 0.05, and ***P < 0.01. (I–K) was performed on R1 wild-type mESCs.

Fig. 5.

m⁶A depletion coactivates PI3K-pAkt promoting Nanog expression. (A–E) Nanog^VENUS mESCs were transduced with control shRNA (Scr) or shRNAs targeting Mettl3 (shM3-1 and shM3-2). (A) Immunoblotting for total Akt and phosphorylated Akt (pAkt) on two amino acid residues, Ser473 and Thr308 (Left), and corresponding protein quantifications (Right). (B) Intracellular flow cytometry analysis for Alexa Fluor 647–conjugated anti-pAkt^S473 antibody. A solid rectangle denotes the Nanog-positive population, and a dashed rectangle denotes the Nanog-negative population. Representative of n = 5 independent experiments. (C) Control and m⁶A-depleted cells were treated with 5 μM PI3K inhibitor (LY294002) or vehicle control (dimethyl sulfoxide [DMSO ]) for 4 d and analyzed by flow cytometry. Proportions of Nanog^VENUS-negative cells are labeled. Representative from n = 4 independent repeats. (D) Quantification of the increase in Nanog^VENUS-negative cell fraction upon PI3K inhibition in C. Statistical significance is measured by two-tailed independent t test with unequal variance, *P < 0.1, **P < 0.05, and ***P < 0.01. pAkt^S473: phospho-Akt (Ser473); and pAkt^T308: phospho-Akt (Thr308). (E) Flow cytometry analysis was used to measure the proportions of Nanog^VENUS-positive cells every 2 d in Serum/LIF for a total of 30 d. All experiments shown were performed with at least three independent experimental repeats.

Fig. 6.

Stably m⁶A-depleted cells activate Mek-pERK facilitating pluripotency exit. (A–E) J1 wild-type (WT), Mettl3 KO clone 1 (KO1), and Mettl3 KO clone 2 (KO2) cell lines were used in experiments (Methods). (A) Multispectral, single-cell, protein-level quantification showing Nanog against Oct4 expression in Serum/LIF. WT = 10,112, KO1 = 11,399, and KO2 = 9,413 cells shown. (B) Immunoblotting for total FGFR1, phosphorylated FGFR1 (pFGFR1), total Akt, phosphorylated Akt (pAkt^S473), total Erk1/2 and phosphorylated Erk1/2 (pErk1/2), Mettl3, Nanog and GAPDH protein levels are shown. (C) Multispectral, single-cell, protein-level quantification showing Nanog against Oct4 expression intensity in 2i/LIF. WT = 14,215, KO1 = 13,865, and KO2 = 12,283 cells shown. (D) Immunoblotting for total Akt, phosphorylated Akt (pAkt^S473), total Erk1/2 and phosphorylated Erk1/2 (pErk1/2), Mettl3, Nanog, and GAPDH protein levels are shown. Protein levels in B and D are directly comparable, and GAPDH is used as a loading control. (E) Multispectral, single-cell, protein-level quantification showing Nanog against Oct4 expression intensity after 48 h of retinoic acid (RA) induced differentiation. WT = 14,445, KO1 = 14,390, and KO2 = 12,506 cells shown. (A–E) Representatives from n = 3 independent experimental repeats. (F) A model of signaling activation in m⁶A-depleted cells. Arrows (blue or red) indicate signaling stimulation, and blunted arrows indicate inhibition of cell state, respectively. Black arrows delineate cell-state transitions, and greater thickness indicates an increase in transition propensity. In Serum/LIF, higher pErk (blue arrows) stimulates the expression of lineage TFs like Otx2 and Oct6, facilitating pluripotency exit. This is countered by increased pAkt activation (red arrows), reinforcing Nanog expression. Prolonged m⁶A depletion may gradually strengthen the pluripotency TF network due to increased pluripotency TF mRNA stability effects. This leads to delayed pluripotency exit upon differentiation or “hyper-pluripotency.” While pErk activation counteracts hyper-pluripotency in stably m⁶A-depleted cells in Serum/LIF, pErk blockade in 2i/LIF exacerbates hyper-pluripotency. In FS cells, m⁶A depletion activates both pAkt and pErk, tipping the signaling balance toward a state more poised for differentiation. Consequently, lineage induction in m⁶A-depleted FS cells results in a greater propensity for mesendodermal commitment.