The protective role of GPX4 in naïve ESCs is highlighted by induced ferroptosis resistance through GPX4 expression

Abstract

Ferroptosis, a form of oxidative cell death mediated by lipid peroxidation, is strictly regulated by glutathione peroxidase 4 (GPX4). Knockout of Gpx4 results in embryonic lethality, highlighting its essential role in development. In vitro, mouse embryonic stem cells (mESCs), which represent the naïve pluripotent state, require β-mercaptoethanol (bME) to prevent cell death, unlike human embryonic stem cells, which represent the primed state. We hypothesized that naïve pluripotency is linked to a heightened susceptibility to ferroptosis due to unique metabolic demands and redox imbalances. In this study, we found that bME deprivation induces ferroptosis in naïve ESCs, as evidenced by lipid peroxidation; ferroptosis, however, is less evident in primed ESCs. Mechanistic analyses revealed that active oxidative phosphorylation (OXPHOS) in naïve ESCs increased mitochondrial reactive oxygen species. Consistent with the upregulation of Gpx4 transcripts and OXPHOS-associated gene sets seen in the inner cell mass of blastocysts, stable GPX4 expression conferred resistance to ferroptosis induced by bME withdrawal. These results suggest that the unique redox and metabolic landscape of naïve ESCs highlits a potential requirement for GPX4 in maintaining naïve pluripotency, providing insights into early developmental processes and vulnerabilities.

Keywords: Ferroptosis; Glutathione peroxidase 4 (GPX4); Naïve pluripotency; Primed pluripotency.

Fig. 1

Cell death of naïve but not primed ESCs triggered by bME depletion (A) Graphical illustration of the culture conditions for obtaining naïve and primed ESCs from mouse ESCs. Representative brightfield images show the dome-shaped colony characteristic of naïve ESCs and the flat epithelial morphology of primed ESCs. LIF, leukemia inhibitory factor. Scale bars, 200 μm. (B) Quantification of mRNA expression for naïve- (left) and primed-specific (right) markers in naïve and primed ESCs. Biological replicates: n = 2 for Dppa3 and n = 3 for all others. (C) Microscopic images (left) and quantification of colony area (right) of naïve ESCs cultured in the presence (Mock) or absence of bME [(−) bME] for 48 h. Colony area was calculated as a percentage of the total culture surface area. Scale bars, 200 μm. Biological replicates: n = 6 per condition. (D) Microscopic images (left) and quantification of colony area (right) of primed ESCs cultured in the presence (Mock) or absence of bME [(−) bME] for 48 h. Colony area was calculated as a percentage of the total culture surface area. Scale bars, 200 μm. Biological replicates: n = 3 per condition. (E–F) Time-lapse images (left) and quantification (right) of colony area (%) of naïve ESCs (E) and primed ESCs (F) cultured in the presence (Mock) or absence of bME [(−) bME]. Representative images show colony morphology at the indicated time points. Colony area was calculated as a percentage of the total culture surface area. Biological replicates: n = 6 for naïve and n = 3 for primed ESCs. Data show mean ± s.d. (B–F). P values were calculated using a two-tailed t-test (C, D) or two-way ANOVA (E, F). ∗∗∗p < 0.001; ns, not significant.

Fig. 2

Cell death of naïve ESCs upon bME depletion occurs through ferroptosis (A) Flow cytometric analyses of 7-AAD negative live populations in naïve ESCs cultured with (Cont) or without bME [(−) bME], treated with vehicle (Mock), pan-caspase inhibitor (zVAD, 20 μM), or ferrostatin-1 (Fer-1, 1 μM) for 24 h. The percentage of live cells is quantified (right). Biological replicates: n = 3 for each group. (B) Time-dependent cell viability of bME-deprived naïve ESCs cultured in the absence (Mock) or presence of Fer-1 (1 μM) for 24 h and 48hrs, assessed using 7-AAD staining. Biological replicates: n = 3 for each group. (C) Lipid peroxidation in bME-deprived naïve ESCs assessed by flow cytometry using C11-BODIPY 581/591 staining. Cells were cultured without (Mock) or with bME [(−) bME], and treated with either zVAD (20 μM) or Fer-1 (1 μM) for 24 h. Biological replicates: n = 3 for each group. (D) Flow cytometric analysis (top left) and corresponding microscopic images (bottom left) of naïve ESCs treated with the apoptosis inducer etoposide (Eto) in the presence or absence of zVAD (20 μM) or Fer-1 (1 μM) for 24 h. Live cells were assessed by 7-AAD and Annexin V staining. The percentage of live cells is shown (right). Biological replicates: n = 3 for each group. (E) Immunoblot analysis of cleaved PARP1 (c.PARP1) and cleaved caspase-3 (c.Cas3) in naïve ESCs cultured with bME (Cont), without bME (-bME), or treated with 1 μM etoposide (Eto, cultured in the presence of bME) for 24 h. β-actin was used as a loading control. Data show mean ± s.d. (A–D). P values were calculated using one-way ANOVA with Tukey test for post-hoc analyses (A, C, D) or two-tailed t-test (B).

Fig. 3

Naïve-specific lipid metabolism: the mevalonate pathway is attenuated in naïve ESCs (A) Enrichment plots from GSEA of RNA-seq data illustrating reduced activity of the mevalonate pathway in naïve ESCs compared to primed ESCs. The gene sets were derived from the Wikipathway database. (B) Schematic of the mevalonate pathway and downstream processes, highlighting key enzymes in cholesterol and isoprenoid biosynthesis. Red boxes indicate enzymes downregulated in naïve ESCs versus primed ESCs, based on RNA-seq analysis. The red intensity reflects fold change magnitude. The left and right halves of each box show fold changes from the J1/P-J1 and OG2/P-OG2 ESC pairs, respectively. (C) Relatively lower mRNA levels of Hmgcr and Fdps in naïve compared to primed ESCs, determined by qRT-PCR. Biological replicates: n = 3 per condition. (D) Relatively lower luciferase activity of sterol regulatory element (SRE) reporters in naïve compared to primed ESCs. Biological replicates: n = 2 for J1/P-J1 pairs and n = 2 for OG2/P-OG2 pairs. Data are normalized to Renilla luciferase and presented as fold change. (E) Schematic of the mevalonate (MVA) pathway and its downstream processes involved in isoprenoid and cholesterol biosynthesis. HMGCR and FDPS (green) are key enzymes downregulated in naïve ESCs, as shown by qRT-PCR. CoQ (red) acts as an antioxidant that inhibits ferroptosis. (F) Reduced amounts of CoQ in naïve ESCs, determined using ELISA (upper) and mass spectrometry (lower). Biological replicates: n = 7 per condition for each method. (G) Viabilities of control and bME-depleted naïve ESCs for 48 h, determined by 7-AAD staining. bME-depleted cells were treated with vehicle, Fer-1 (1 μM), or idebenone (CoQ, 2 μM) for 24 h before harvest. Biological replicates: n = 3 per condition. (H) Representative bright-field images (left) and quantification of lipid peroxidation (right) in naïve ESCs cultured with or without bME and treated with zVAD (20 μM), Fer-1 (1 μM), or idebenone (CoQ, 2 μM) for 24 h. Lipid peroxidation levels (% of cells) were assessed using flow cytometry after staining with C11-BODIPY 581/591. Biological replicates: n = 3 per condition. Data show mean ± s.d. (C, D, F–H). P values were calculated using two-tailed t-test (C, D, F, G) or one-way ANOVA followed by Tukey's post-hoc test (H). ∗p < 0.05, ∗∗<0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Fig. 4

Redox imbalance from oxidative phosphorylation and high TfR1 expression predispose naïve ESCs to ferroptosis (A) Relative GSH/GSSG ratio in naïve and primed ESCs, cultured with and without bME for 24 h. Biological replicates: n = 3 per group. (B) Flow cytometric analysis of mitochondrial reactive oxygen species levels using mitoSOX (left) and quantification of the geometric mean fluorescence intensity (right) in naïve and primed ESCs in standard culture conditions. Biological replicates: n = 8 for naïve, and n = 4 for primed ESCs. (C) Flow cytometric analysis of mitochondrial lipid peroxidation in naïve ESCs at indicated culture conditions using mitoPerOX staining (left). bME-depleted cells were treated with Mock, Fer-1 (1 μM), or idebenone (CoQ, 2 μM) for 24 h. Quantification of geometric mean intensity is shown (right). Biological replicates: n = 3 per condition. (D) Viabilities of bME-depleted naïve ESC treated with Mock or oligomycin (Oligo, 1 μM) for 12 h (left). Quantification of the percentage of live cells is shown (right). Biological replicates: n = 3 per condition. (E)Tfrc and Klf2 expression in naïve and primed ESCs. Biological replicates: n = 3 per each group. (F) Immunoblot of TfR1 in naïve and primed mESCs. β-actin served as a loading control. (G) Representative images of fluorescence-labelled transferrin uptake in naïve (top row) and primed (bottom row) ESCs. Quantification of fluorescence intensity is shown in the bottom graph. Biological replicates: n = 3. (H) Representative microscopic images of control (Cont) and bME-depleted [(−)bME] naïve ESCs. In the bME-depleted group, cells were treated with Mock, Fer-1, control IgG, or anti-TfR1 antibody for 24 h. (I) Cell viability analysis of bME-depleted naïve ESCs treated with Mock, Fer-1, control IgG, or anti-TfR1 antibody for 24 h, assessed by 7-AAD staining (left). The percentage of live cells is shown (right). Biological replicates: n = 3 per condition. (J) Cell death of wild-type (WT) and Trp53 knockout (KO) naïve ESCs treated with Mock or RSL-3 (0.5 μM), assessed by flow cytometry of 7-AAD staining (left). Quantification of the percentage of dead cells is shown (right). Biological replicates: n = 3 per condition. (K)Tfrc expression in WT and Trp53 KO naïve ESCs. Biological replicates: n = 3. (L) Immunoblot of TfR1 in WT and Trp53 KO naïve ESCs. β-actin served as a loading control. Data are presented as mean ± s.d. (A-E, G, I–K). Statistical significance was tested using two-tailed t-test (A, B, G, J) or one-way ANOVA followed by Tukey's post-hoc test (C, D, I). ∗p < 0.05, ∗∗<0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Fig. 5

The high GPX4 dependency of naïve ESCs is associated with oxidative stress (A) Schematic representation of increasing ROS levels of ESCs from the oocyte to the inner cell mass (ICM) of blastocyst, adapted from Deluao J et al., 2022 (PMID: 36111646) (left). The ICM, the in vivo counterpart of naïve ESCs, is highlighted with an arrow. The PCA plot (right) illustrates the trajectory of transcriptional changes (red to blue) spanning embryonic developmental stages. Gene expression data were obtained from the GSE70605 dataset (Liu et al., 2016; PMID: 27462457). (B) Dot plot illustrating the enrichment of oxidative stress response gene sets from the Molecular Signatures Database (MSigDB) across developmental stages. Dot size represents the absolute enrichment score, indicating the magnitude of enrichment, while color indicates the direction of enrichment (blue for negative scores, red for positive scores). (C) Violin plots showing the expression levels (FPKM) of Gpx4 (left) and Gss (right) across embryonic developmental stages. (D) Representative pseudocolor plots of 7-AAD stained naïve and primed ESCs treated with the indicated concentrations of RSL3 for 4 h (left). Red numbers indicate the percentage of 7-AAD negative live cells. Graphical representation of the percentage of live cells is shown (right). (E) Immunoblot analysis of GPX4 protein levels in naïve and primed ESCs cultured with (Cont) or without bME [(−) bME]. β-actin served as a loading control. The numbers represent the relative normalized densities. (F) Immunoblot analysis of GPX4 protein levels in bME-depleted naïve ESCs treated with carfilzomib for 24 h. β-actin served as a loading control. The numbers represent the relative normalized densities.

Fig. 6

Stable overexpression of GPX4 in naïve ESCs favors naïve pluripotency (A) mRNA expression of Gpx4 in wild-type (WT) and GPX4-overexpressing (OE) naïve ESC, assessed by qRT-PCR (left) (n = 3 biological replicates). Immunoblotting of GPX4 with β-actin as a loading control. The relative GPX4 protein level in OE cells (shown in red) was quantified by densitometry analysis using ImageJ. (B) Expression of naïve pluripotency-associated genes (Dppa3, Esrrb, Klf2, Klf4) (left) and primed pluripotency-associated genes (Fgf5, Otx2) (right) in WT and GPX4-OE naïve ESCs, compared with primed ESCs. Biological replicates: n = 3 per each group. (C) Immunoblot of NANOG, pSTAT3, and GPX4 in wild-type (WT) and GPX4-OE naïve ESCs compared with primed ESCs. β-actin served as a loading control. (D) Teratoma formation assay confirming pluripotency in WT and GPX4-OE naïve ESCs. The representative section images show all three germ layers. (E) Microscopic images of WT and GPX4-OE naïve ESCs cultured in the presence (Mock) or absence of bME [(−) bME]. (F) Cell viability of WT and GPX4-OE naïve ESCs in the absence of bME (-bME). Flow cytometric plots (left) show 7-AAD staining, with red numbers indicating the percentage of live cells. Quantification of the percentage of live cells is presented (right). Biological replicates: n = 3 per group. (G) Cell death analysis of WT and GPX4-OE naïve ESCs treated with etoposide (Eto, 1 μM) or etoposide plus zVAD (20 μM) (Eto + zVAD) for 24 h. Representative flow cytometric histograms (left) and quantification of 7-AAD positive dead cells (right) are shown. Biological replicates: n = 3 per group. (H) Quantification of growth area (%) of WT and GPX4-OE naïve ESCs cultured with (Mock, left) or without bME (-bME, right) over the indicated time. Biological replicates: n = 3 per group. Data are presented as mean ± s.d. (A, B, F–H). Statistical significance was tested using two-tailed t-test (F), one-way ANOVA followed by Tukey's post-hoc test (G), or two-way ANOVA (H). ∗∗<0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; ns, not significant.

Fig.S1

Fig.S2

Fig.S3

Fig.S4