Cell Death and the p53 Enigma During Mammalian Embryonic Development

Abstract

Twelve forms of programmed cell death (PCD) have been described in mammalian cells, but which of them occurs during embryonic development and the role played by the p53 transcription factor and tumor suppressor remains enigmatic. Although p53 is not required for mouse embryonic development, some studies conclude that PCD in pluripotent embryonic stem cells from mice (mESCs) or humans (hESCs) is p53-dependent whereas others conclude that it is not. Given the importance of pluripotent stem cells as models of embryonic development and their applications in regenerative medicine, resolving this enigma is essential. This review reconciles contradictory results based on the facts that p53 cannot induce lethality in mice until gastrulation and that experimental conditions could account for differences in results with ESCs. Consequently, activation of the G2-checkpoint in mouse ESCs is p53-independent and generally, if not always, results in noncanonical apoptosis. Once initiated, PCD occurs at equivalent rates and to equivalent extents regardless of the presence or absence of p53. However, depending on experimental conditions, p53 can accelerate initiation of PCD in ESCs and late-stage blastocysts. In contrast, DNA damage following differentiation of ESCs in vitro or formation of embryonic fibroblasts in vivo induces p53-dependent cell cycle arrest and senescence.

Graphical Abstract

Neither cell cycle arrest nor programmed cell death requires p53 prior to gastrulation, at which stage DNA damage induces p53-dependent cell cycle arrest and senescence.

Significance Statement.

Programmed cell death (PCD) and survival are inherent components of mammalian development. In addition to its role as “guardian of the genome”, the p53 transcription factor and tumor suppressor has been reported to regulate at least six different forms of PCD. Given the importance of pluripotent stem cells as models of embryonic development and their applications in regenerative medicine, identifying which of the 12 forms of PCD respond to stress imposed at the beginning of embryogenesis and the role of p53 in regulating them is essential.

Introduction

Cell death is a normal event in mammalian development, as well as a cellular response to stressful conditions. The mechanisms that cause cell death are categorized as either necrosis or programmed cell death (PCD). Necrosis results from the progressive degradative action of enzymes and is typically followed by inflammation. Necrosis requires neither energy nor effector proteases¹; it is simply a response to physical damage or pathology that does not occur during normal animal development.² In contrast, PCD is a sequence of genetically programmed events by which a cell provokes its own demise in response to a stimulus. PCD occurs in mammals as early as the blastocyst stage during preimplantation development and as late as tissue homeostasis in adulthood.³ During organogenesis, both PCD and programmed cell senescence are involved in sculpting structures by eliminating interdigital webbings, converting solid structures into hollow tubes, and removing excess cells from nervous, immune, and reproductive systems.^4,5 In postimplantation embryos, the proamniotic cavity is formed by PCD of the ectodermal cells in the core of the developing egg cylinder⁶ (Fig. 1).

Figure 1.

Early mouse embryonic development. (A) The number of cells, days post-coitum (E2.5-E12), and morphogenetic events are indicated. ALL, allantois; AMN, amnion; AVE, anterior visceral endoderm; BC, blastocyst cavity; DVE, distal visceral endoderm; ECT, ectoderm; EPI, epiblast; ExE, extraembryonic ectoderm; ICM, inner cell mass; MES, mesoderm; N, node; NF, neural fold; PAC, proamniotic cavity; PrE, primitive endoderm; PS, primitive streak; TE, trophectoderm; VE, visceral endoderm; ZP, zona pellucida. Adapted from Ref. ¹¹⁶. Preimplantation development begins with totipotent blastomeres (1-8 cell stage) encapsulated by the zona pellucida. Totipotent cells can give rise to both placental and embryonic cells. When the blastomeres develop cell-to-cell adhesion (compaction), the outer blastomeres differentiate into the trophectoderm while the remaining blastomeres form the inner cell mass. The epithelial trophoblast cells (trophectoderm) are multipotent; they differentiate only into cells required for implantation and placentation. The inner cell mass (recognized upon formation of a blastocoel cavity) differentiates into the epiblast and the primitive endoderm. Postimplantation development begins when the primitive endoderm differentiates into multipotent visceral and parietal endoderm. Mesoderm and ectoderm are derived from the epiblast during gastrulation. Gastrulation begins at the primitive streak, from which mesoderm and endoderm progenitor’s ingress and begin to differentiate.¹¹⁷ Mouse embryonic fibroblasts (MEFs) are derived from E12-E14 embryos. Ablation of the Mdm2, Rbbp6, or Mdm4 gene is lethal in embryos at the indicated times. Mouse embryonic stem cells (mESCs) are derived from the epiblast in blastocysts.⁴⁹ mESCs cultured in the presence of serum and LIF interleukin-6 are considered “naïve” pluripotent cells, because they can give rise to all the cells of the embryo, but not to the trophectoderm. Naïve mESCs cultured in defined medium (no serum) containing 2 metabolic inhibitors are considered totipotent “ground-state” ESCs (2iESCs), because they give rise to both extraembryonic and embryonic cells. Naïve mESCs cultured in the presence of activin and fibroblast growth factor generate pluripotent “primed” ESCs, because they give rise to the same cells as “naïve mESCs,” but they cannot generate chimeric animals.¹¹⁸ Human embryonic stem cells (hESCs) and mouse epiblast stem cells (EpiSCs) are derived from the epiblast of post-implantation blastocysts.⁴⁸ (B) Images of LysoTracker Red stained oocytes and preimplantation embryos revealed that the number of lysosomes increased after fertilization.³⁶ Scale bar is 10 µm.

Remarkably, the mechanism of PCD during mammalian development, as well as the role of the p53 transcription factor in PCD, remain controversial. Pluripotent embryonic stem cells (ESCs) respond to stressful conditions, such as DNA damage, by arresting cell proliferation and undergoing PCD, but the form of PCD and the role of p53 remain unresolved. Studies not reviewed herein reveal 2 roles of p53 during mouse embryonic development are promoting genomic stability and maintaining a differentiated state by suppressing pluripotent gene expression.⁷ However, in differentiated cells, the primary role of the p53 transcription factor is regulating cell cycle arrest, senescence, and PCD in response to DNA damage and other stresses. p53 operates primarily through transcriptional activation of the cyclin-dependent kinase inhibitor CDKN1A/p21 to prevent cells from entering S-phase, and the proapoptotic proteins BBC3/PUMA and NOXA/PMAIP1^8,9 to induce PCD. p53 also regulates transcription of genes involved in DNA repair (DDB2, XPC, GADD45A)¹⁰ and cell senescence (p21, PAI1, PML).^11,12 Low levels of p53 trigger cell cycle arrest and induction of DNA repair pathways specific to the lesion of concern, but as p53 continues to accumulate with time, it eventually triggers PCD, thereby removing cells.^13,14

Remarkably, neither p53, p21, nor PUMA is required for mouse development.^15,16 Therefore, either p53 is not required for specific events, or events that require p53 are themselves not required for mammalian development. Alternatively, in the absence of p53, cells utilize a different form of PCD. Therefore, the goal of this review is to summarize the status of PCD at the beginning of mammalian development and to reconcile disparate data based on differences in experimental conditions with mouse ESCs and the fact that p53 cannot induce lethality during mouse embryogenesis until gastrulation.

Programmed Cell Death

Of the 12 forms of PCD characterized in human cells, 4 are associated with mammalian development and 8 with disease states (Table 1). Apoptosis is the most reported PCD and the only one that employs caspases 3, 6, and 7. Apoptosis can be induced by DNA damage, unfolded proteins, reactive oxygen species (ROS), or disruption of cell division using either an intrinsic or an extrinsic pathway. The intrinsic pathway regulates mitochondrial permeability via the Bcl-2 family of cytokines. The extrinsic pathway is triggered by ligand binding to tumor necrosis factor-family receptors in the plasma membrane that activate caspases. Both pathways activate initiator caspase (CASP) 2, 8, 9, or 10, which then activate effector CASP3, 6, and 7, which then degrade cellular proteins indiscriminately. Expression of proapoptotic genes Bax, Puma, Bid, and Bcl-2, as well as Casp6 and Apaf1, a coactivator of Casp9, are upregulated by the p53 transcription factor.

Table 1.

Hallmarks of programmed cell death in mammals.

Form	p53	Morphology	Biochemistry
Associated with development
Apoptosis18,19	p53	Cell rounding, nuclear condensation, membrane blebbing, apoptotic bodies	Activates CASP3 and PARP1, DNA fragmentation and loss, ΔΨm dissipation, phosphatidylserine exposure
Anoikis20	p53	Anchorage-dependent cells detach from the extracellular matrix	Cleaved EMC proteins (laminin, fibronectin, vitronectin) → apoptosis
Lysosome dependent21		Plasma membrane repair, lysosome membrane permeabilization	Release of lysosomal hydrolytic enzymes (cathepsins), lysosomal iron-induced oxidative injury
Autophagy dependent22,23	p53	Autophagic vacuolization	LC3-I to LC3-II conversion, increases autophagic flux and lysosomal activity
Associated with disease
Necroptosis24	p53	Cell swells, PMR, moderate chromatin condensation	Activates RIPK1, RIPK3, and MLKL, cytosolic necrosome formation
Oxeiptosis25		Apoptosis-like	ROS-dependent, activates KEAP1 and NFE2L2. caspase-independent, no AIFM1 translocation
Ferroptosis26	p53	Small mitochondria (mt), reduced mt-crista, elevated mt-membrane densities, mt-membrane rupture	Iron accumulates, lipid peroxidation, ΔΨm dissipation, LC3-I to LC3-II conversion, glutaminolysis, caspase-independent
Parthanatos27	p53	Chromatin condensation, large DNA fragments, no cell swelling, apoptotic bodies or small DNA fragments, PMR	Oxidative stress (ROS)-induced, PARP1 activation, ΔΨm dissipation, caspase-independent, NAD+ and ATP depletion, accumulates PARP polymers, AIFM1 translocation
Alkaliptosis28		Necrosis-like	Intracellular alkalinization, activates NF-κB, caspase-independent
Pyroptosis29		No cell swelling, PMR, bubbling, moderate chromatin condensation	Activates CASP1, CASP3, and GSDMD, GSDMD-N-induced pore formation, IL1B released
Entotic30		One cell invades another	Activates adhesion proteins and actomyosin, LC3-associated phagocytosis
Netotic31		PMR, nuclear membrane collapse, chromatin fiber release	Forms NETs, release and translocation of granular enzymes, histone citrullination

Note: ΔΨm is mitochondrial membrane potential.³⁷ Reactive oxygen species is reactive oxygen species. Neutrophil Extracellular Traps (NETs) are neutrophil extracellular traps. LC3 is MAP1LC3B. Dying cells release small vesicular apoptotic bodies.³⁸ Plasma membrane rupture releases intracellular molecules that propagate inflammatory response.³⁹ Apoptosis-inducing factor 1 (AIFM1) translocates from mitochondria to nucleus.⁴⁰ EMC is extracellular matrix. Adapted from Ref. ¹⁷.

NMRs, neutrophil extracellular traps; PARP, poly(ADP-ribose) polymerase; PMR, plasma membrane rupture; ROS, reactive oxygen species.

Apoptosis is recognized by DNA fragmentation, accumulation of cells containing <2N DNA, binding of annexin-V to detect phosphatidylserine exposure in the plasma membrane, propidium iodide, or trypan blue staining to detect plasma membrane permeability, accumulation of γH2AX to confirm double-strand DNA breaks, and cleavage of poly(ADP-ribose) polymerase (PARP) and CASP3. Apoptosis can be either p53-dependent or p53-independent.

Anoikis is the induction of apoptosis when cells lose their attachment to the extracellular matrix (EMC) and neighboring cells. Anoikis suppresses tumor metastasis and eliminates ectopic proliferation of misplaced progenitor cells during tissue development.

Autophagy-dependent PCD is an unlikely death mechanism because autophagy is primarily a survival mechanism. Autophagy is a process in which cytoplasmic organelles, proteins, and macromolecules are degraded to produce new macromolecules and energy. Thus, starvation activates autophagy to maintain homeostasis and viability, and the autophagy gene ATG7 inhibits p53-dependent cell cycle arrest and PCD.^32,33

This paradox could be resolved in 3 ways. First, mild stress might induce autophagy for survival, whereas severe stress induces autophagy for PCD. This hypothesis is analogous to the fact that mild DNA damage induces p53-dependent cell cycle arrest in differentiated human cells, whereas severe DNA damage induces p53-dependent PCD or senescence.³⁴ Alternatively, PCD might disrupt autophagic flux, which results in accumulation of autophagosomes (LC3-I, LC3-II, and p62 proteins),³⁵ a phenomenon that could be misinterpreted as increased autophagy. Finally, lysosome-dependent PCD might be mistaken for autophagy-dependent PCD, because lysosomes are present in zygotes within 2 to 4 h after fertilization and then enriched during reimplantation development³⁶ (Fig. 1B). In autophagy, autophagosomes collect cellular trash and then fuse with lysosomes to degrade the trash within the autolysosome vesicle. In lysosome-dependent PCD, permeabilized lysosomes release their hydrolytic enzymes into the cytoplasm in a process facilitated by p53-upregulation of cathepsin synthesis.¹⁷

PCD in Preimplantation Mouse Embryos

During in vitro development of zygotes into blastocysts, the polar bodies in 1-cell and 2-cell embryos and one or 2 of the cells in the morula at the junction between inner cell mass and trophectoderm undergo PCD.⁴¹ However, although caspase activity was required for preimplantation development, these early events were not caspase-dependent. Therefore, PCD appeared to occur via a non-canonical form of apoptosis (Fig. 1A).

γH2AX is easily detected throughout preimplantation development, thereby revealing the presence of double-strand DNA breaks even in the absence of any induced DNA damage.⁴² In contrast, the p53-binding protein “53BP1” that is recruited to sites of double-strand DNA breaks⁴³ was not detected. Thus, p53-dependent PCD does not appear to be induced by low levels of double-strand DNA breaks. In fact, regardless of the presence or absence of p53, induction of double-strand DNA breaks by X-irradiation of two-cell embryos retarded their development to the late blastocyst stage but did not prevent it.⁴⁴ Wild-type blastocysts exhibited 2 to 3-times more cells with DNA damage than p53^−/− blastocysts, thereby revealing that p53 accelerates PCD, but is not required for PCD.

Autophagy-dependent PCD is unlikely because autophagy is essential for preimplantation development.⁴⁵ Autophagy-defective oocytes derived from oocyte-specific Atg5 (autophagy-related 5) knockout mice failed to develop beyond the 4- to 8-cell stages if fertilized by Atg5^−/− sperm, but they did develop if fertilized by wild-type sperm. However, lysosomes rapidly accumulate after fertilization (Fig. 1B) and lysosome accumulation is required for development,³⁶ suggesting that induced stress might activate lysosome-dependent PCD. In fact, PIKfyve, a phosphoinositide kinase essential for maintaining lysosome homeostasis and autophagic flux,³⁵ is essential for preimplantation mouse development.⁴⁶

PCD in ESCs

Embryonic stem cells can exhibit multiple physiological states (Fig. 1 legend). Naïve mESCs correspond to transient populations in pre- or peri-implantation embryonic epiblast whereas primed mESCs (termed EpiSCs) isolated from post-implantation blastocysts model the postimplantation epiblast.⁴⁷ hESCs are similar to primed mESCs.⁴⁸ Ground state mESCs are produced by culturing naïve mESCs with inhibitors of MAP2K1/MEK1 and FRAT2/GSK-3, and therefore termed 2iESCs.⁴⁹ An intermediate state between naïve and primed ESCs has recently been described.⁵⁰

Doxorubicin/Adriamycin is an anticancer drug commonly used to induce PCD in mammalian cells by causing double-strand DNA breaks. Doxorubicin induces PCD equally well in either p53^+/+ or p53^−/− naïve ESCs, as evident from visual inspection of cultured cells, DNA loss, annexin-V binding, propidium iodide, and trypan blue staining, cleaved PARP, and CASP3.⁵¹ Robust PCD did not require p53 or its primary targets, the CDK2 inhibitor p21 and pro-apoptotic protein PUMA, to cleave PARP and CASP3 (Fig. 2A), arrest cell proliferation (cells accumulate with 4N DNA content, Fig. 2C) and complete apoptosis (cells accumulate with <2N DNA content) (Fig. 2C and D). Thus, DNA damage in naïve ESCs induced a p53-independent form of noncanonical apoptosis. This conclusion was confirmed by the translocation of AIFM from mitochondria to the nucleus in p53^−/− as well as p53^+/+ ESCs (Fig. 2B).

Figure 2.

Cell cycle arrest and apoptosis in naïve ESCs are not dependent on p53. (A) Doxorubicin/Adriamycin (Dox) induced DNA damage (γH2AX expression), DNA damage response (PARP to c-PARP cleavage) and apoptosis (CASP3 to c-CASP3 cleavage) in ESCs derived from p53^+/+ and p53^−/− mouse blastocysts (BD-ESCs, “chronic phenotype”). ESCs were cultured with or without 500 nM Dox. At the times indicated, attached and unattached cells were combined, and total cellular proteins analyzed by immunoblotting. (B) PCD was detected by translocation of AIFM (red) from cytoplasm to nuclei (blue) in BD-ESCs cultured with 500 nM Dox for 16 h. Scale bar is 15 µm. (C) A transient accumulation of cells with 4N DNA content is characteristic of a DNA damage-induced G2-arrest. The G2-checkpoint was activated within 24 h and apoptosis within 72 h by 50 nM Dox in both conditional knockout p53^−/− ESCs and their p53^+/+ parent (cKO-ESCs, “acute phenotype”). Attached and unattached cells were combined, and their DNA content quantified by fluorescence-activated cell sorting. Cells with <2N DNA content (apoptotic cells) and cells with 4N DNA content (G2/M phase cells) are indicated. Equivalent results were obtained with BD-ESCs. (D) Cells with <2N DNA content were quantified as a function of time cultured with Dox and normalized to 0% at zero hours. Error bars indicate ±SEM. Panels A and B are from Fig. 3B and C in Ref. ⁵¹ and panels C and D are from Figs. 2 and S2 in Ref. ⁵¹.

Similar results occurred using etoposide to cause double-strand DNA breaks.⁵² DNA fragmentation was accompanied by annexin-V binding, plasma membrane permeabilization, and cleavage of PARP, but not caspase-3. Instead, PCD was accompanied by increased levels of cathepsins. Pifithrin-α reduced PCD, suggesting these effects were p53-dependent, but pifithrin-α also suppresses ESC self-renewal⁵³ via mechanisms unrelated to p53.⁵⁴ These results are consistent with a p53-independent form of lysosome-dependent PCD.

PIKfyve inhibitors induce non-canonical apoptosis (no caspase-3 cleavage) in mouse and human ESCs,^51,55 as well as in autophagy-dependent human cells,^35,56 thereby confirming the dependence of pluripotent stem cells on either lysosome homeostasis or autophagic flux. Because PIKfyve inhibitors alter lysosome permeability and cathepsin maturation,^35,57 these results suggest lysosome-dependent PCD.

Necroptosis initiates cell death in the absence of caspase cleavage by activating death receptors in the plasma membrane that trigger assembly of a “necrosome complex” followed by permeabilization of the plasma membrane and an inflammatory response. High levels of autophagosome-associated proteins ATG5, ATG8/LC3, or SQSTM1/p62 promote necrosome assembly and activation in human cancer cells,^58-62 suggesting that disruption of autophagy by inhibition of PIKfyve, which causes accumulation of LC3 and p62, might trigger necroptosis in ESCs. Thus, different cellular stresses appear to trigger different forms of PCD.

PCD in Postimplantation Mouse Embryos

Leukemia inhibitory factor (LIF) deprivation of naïve ESCs induces differentiation, and comparison of LIF deprived p53^+/+ with p53^−/− ESCs revealed that the roles of p53 in cell cycle regulation, apoptosis, and senescence are acquired during pluripotent stem cell differentiation.⁵¹ Senescence prevents cell proliferation permanently, while retaining cell function. Wild-type mouse embryonic fibroblasts (MEFs, Fig. 1) treated with doxorubicin undergo G1-arrest and senescence, but p53^−/− MEFs fail to do so and ultimately undergo noncanonical apoptosis.^51,63-66 As little as 50 nM doxorubicin induces p53-independent apoptosis in ESCs (Fig. 2D), but MEFs require excessive concentrations of doxorubicin (Fig. 3B).^51,67 Only MEFs expressing an oncogene exhibit p53-dependent apoptosis.^68-70 Thus, senescence rather than apoptosis is the normal response to DNA damage in differentiated cells.

Figure 3.

p53 activity and MEF PCD response at the beginning of mouse development. (A) p53 activity assayed in embryos isolated from mice homozygous for reporter genes expressing enhanced green fluorescence protein driven by either the Cdkn1a/p21 or the Bbc3/Puma gene’s p53 response element.⁷⁵ At embryonic day E3.5, fluorescence was detected in the inner cell mass (ICM), and trophectoderm (TE) of blastocysts. The large blastocoel cavity identifies these examples as late-stage blastocysts containing early epiblast (Fig. 1). At embryonic day E6.5, fluorescence was detected in the epiblast but not in the extraembryonic tissue of gastrula. (B) MEFs cultured for 24 h with doxorubicin and then stained for “apoptosis-inducing factor” AIFM.⁵¹ Scale bar is 15 μm. Translocation of AIFM from mitochondria to nucleus occurred in both p53^+/+ and p53^−/− cells, thereby confirming non-canonical apoptosis in MEFs treated with 500 nM doxorubicin, but not in MEFs with 50 nM doxorubicin.

In the absence of pro-apoptotic genes Bax and Bak, MEFs appear to undergo autophagy-dependent PCD in response to either etoposide or Staurosporine.^71,72p53^−/− MEFs were not characterized. Alternatively, DNA damage in MEFs lacking both Bax and Bak induced “parthanatos” (programable necrosis), a mechanism largely controlled by p53-mediated transcription of cathepsin Q in cooperation with DNA damage-induced ROS.^73,74Parthanatos is a PARP1-dependent form of PCD that relies on the AIFM1-macrophage migration inhibitory factor (MIF) pathway. MIF is an AIFM1-binding protein with nuclease activity that produces large DNA fragments. Thus, alternative forms of PCD can be activated in different types of cells.

p53 Regulation of Cell Proliferation, PCD, and Senescence

p53-dependent transcription is first detected during mouse development at the late blastocyst stage, but p53 levels are not great enough to induce embryonic lethality until gastrulation. Therefore, a role for p53 in arresting cell proliferation or activating PCD or senescence begins with cell differentiation.

p53 Activity

Expression of an ectopic EGFP reporter gene driven by a p53-dependent response element demonstrated that p53-dependent transcriptional activity exists as early as late-stage blastocysts and is confined to the epiblast in post-implantation embryos⁷⁵ (Figure 3A). Double-strand DNA breaks introduced by X-irradiation of embryos at either E3.5 (blastocysts) or E9.5 (organogenesis) revealed that p53^+/+ embryos die more frequently than p53^−/− embryos, whereas p53^−/− embryos exhibit more developmental anomalies.⁷⁶ X-irradiated p53^+/+ embryos undergoing organogenesis contain a greater number of apoptotic cells than p53^−/− embryos. p53 facilitates apoptosis in X-irradiated embryos only after preimplantation embryos developed into late-stage blastocysts (E5).⁴⁴ No significant change in cell proliferation was observed following X-irradiation, but late-stage p53^+/+ blastocysts exhibited 2 to 3-times more apoptotic cells than p53^−/− blastocysts. Thus, p53-dependent transcriptional activity and apoptosis are first evident in late-stage blastocysts and increase during organogenesis.

p53 Regulation

p53 activity is tightly regulated posttranslationally. Under normal conditions, p53 expression is very low; it is a short-lived protein whose stability and activity are regulated by phosphorylation, methylation, and acetylation events, and by association with specific p53 regulatory proteins such as MDM2 and MDM4/MDMX,^77-79 RBBP6/PACT,^80-82 and PRKRA/RAX/PACT.^83,84 MDM2, MDM4, and RBBP6 are essential for cell viability and embryonic development. Mice lacking p53 and mice lacking both p53 and MDM2 display the same incidence and spectrum of spontaneous tumor formation,⁶⁵ thereby revealing that, in the absence of p53, MDM2 has no effect on cell proliferation, cell cycle regulation, or tumorigenesis. Thus, ablation of the Mdm2, Mdm4, or Rbbp6 gene in mouse embryos is lethal, but only in the presence of p53 protein. Thus, unregulated p53 expression during embryonic development is lethal.

p53-Dependent Lethality

Unregulated p53 activity does not induce embryonic lethality until the onset of gastrulation (Fig. 1). For Mdm2^−/− embryos, demise occurs after implantation of the embryo in the wall of the uterus but before day 7.5 of gestation (≈E5.5).^85,86 Deletion of the Mdm2 gene has no additional effect on cell proliferation, cell cycle control, or tumorigenesis when p53 gene is absent.^65,87 Therefore, lethality in the absence of MDM2 is due solely to p53 activity. For RBBP6/PACT^−/− embryos, lethality occurs after implantation but before E7.5.⁸¹ For Mdm4/Mdmx^−/− embryos, lethality occurs between E7.5 and E12 from the p53-dependent arrest of cell proliferation (presumably senescence).^88-90 All 3 phenotypes could be rescued by transferring the mutated p53 negative regulator gene (Mdm2, RBBP6, or Mdm4) to a p53-nullizygous background, in which case mice develop normally. Thus, embryonic death in the absence of a p53 regulator resulted from activation of p53 protein.

p53-Dependent Senescence

Given that p53 activity is first detectable during the late blastocyst stage and confined to the epiblast in early gastrula, p53 expression is too low to induce either cell cycle arrest or cell death upon release from post-translational regulation until after the blastocyst has implanted (E4.5) and gastrulation has begun (E6.25-E7.5). However, DNA damage begins to accumulate in Mdm2^−/− blastocysts, but less so if embryos also lack the p53-dependent proapoptotic gene Bax, suggesting that unregulated p53 initiates apoptosis in blastocysts.⁹¹ Nevertheless, embryonic lethality still occurs at E6.5-E7.5 due to the arrest of cell proliferation (cell senescence) rather than PCD.

p53 and Naïve ESCs

Cell cycle checkpoints are p53-independent. The G1 checkpoint is a response to cell stress that retards entrance into the S phase.⁹² Naïve mESCs lack a G1 DNA damage checkpoint.^51,93-97 The G2 checkpoint is a transient accumulation of cells with 4N DNA content in response to DNA damage prior to induction of apoptosis.⁹⁸ Double-strand DNA breaks induced by culturing cells with doxorubicin activated the G2 checkpoint in naïve mESCs regardless of the presence or absence of p53,^51,95,97,99 p21, or PUMA⁵¹ (Fig. 2A). In contrast, both hESCs and mouse EpiSCs derived from the epiblast in postimplantation blastocysts exhibit a G1 checkpoint.⁴⁸

PCD is not dependent on p53. Of the 9 studies that investigated the role of p53 in PCD, 3 concluded that p53 is not required^51,93,97 and 6 concluded that p53 is required.^{95,96,100-102} To resolve this paradox, Jaiswal et al⁵¹ quantified the effects of doxorubicin on p53^+/+ and p53^−/− ESCs derived by 2 different methods. To eliminate the possibility that conclusions depended on either the source or derivation of p53^−/− ESCs, both wild-type and p53^−/− ESCs derived directly from blastocysts were characterized in parallel with ESCs in which the p53 genes were ablated in vitro. ESCs isolated from p53^−/− blastocysts exhibit the effects of p53 loss through multiple generations in vivo (chronic phenotype), whereas p53^−/− ESCs engineered in vitro from p53^+/+ ESCs exhibit the effects of immediate p53 loss (acute phenotype).

To eliminate methodology-dependent biases, the rate and extent of cell cycle arrest and cell death were quantified by time-dependent changes in DNA content (Fig. 2C and D), by staining with annexin-V and propidium iodide to distinguish apoptosis from necrosis, by exclusion of trypan blue to distinguish live cells from dead cells, by Western immunoblotting of p53, p21, PUMA, γH2AX, PARP, and CASP3 to confirm genotypes, DNA damage, and caspase cleavage (Fig. 2A), and by cellular localization of AIFM to confirm caspase-independent apoptosis (Fig. 2B).

The results revealed that, regardless of their derivation, naïve mouse ESCs do not require p53, p21, or PUMA either to activate the G2-checkpoint (Fig. 2C) or to undergo robust apoptosis (Fig. 2D). Depending on conditions such as seeding density and doxorubicin concentration, p53 can accelerate initiation of apoptosis in ESCs in response to DNA damage by 8.4 ± 0.5 h, but the rate and extent of apoptosis in ESCs are equivalent and complete PCD within 72 h, regardless of the presence or absence of p53. The inhibitory effect of only 50 nM doxorubicin is evident from visual inspection of cultured cells, and the lethal effect is evident from the accumulation of cells with <2N DNA content. Short exposure (24 h) to a low concentration (50 nM) of doxorubicin to ESCs, then allowing them to recover for 96 h proved that even minimal DNA damage is enough to induce apoptosis in ESCs regardless of presence or absence of p53.

p53 and Ground-State 2iESCs

Naïve ESCs are characterized by hyper-phosphorylated RB1 protein, lack of G1 control, and rapid progression through the cell cycle. In contrast, ground-state 2iESCs, which are derived from naïve ESCs (Fig. 1), have a longer G1-phase with hypo-phosphorylated RB1, implying that they have a functional G1 checkpoint. The RB1-dependent G1 restriction point is active in 2iESCs but abrogated when cultured in serum.¹⁰³ Moreover, the p53-p21 pathway appears active in 2iESCs, and its role in the G1-checkpoint is abolished in naïve ESCs.¹⁰⁴

DNA damage in 2iESCs caused by doxorubicin-induced p53-dependent cell death.¹⁰⁴ DNA damage in 2iESCs caused either by doxorubicin or by aphidicolin inhibition of DNA polymerase-α activated expression of DUX transcription factors that are involved in zygotic gene activation in mouse 2-cell to 4-cell embryos.^105,106 Both studies concluded that this phenomenon is mediated by an ATR and CHK1 response to double-strand DNA breaks. Critical experiments in which p53^+/+ and p53^−/− 2iESCs were compared were carried out in both studies. However, one study concluded that this phenomenon required p53 expression,¹⁰⁶ whereas the other study concluded that it did not.¹⁰⁵ Ironically, even if p53 is essential for DUX expression, loss of p53 would still not affect embryonic development, because DUX is not required for mouse development.¹⁰⁷

Experimental Conditions Could Account for Contradictory Conclusions

Cell Culture

Culture conditions are critical to maintaining the pluripotent state.⁴⁹ Suboptimal conditions promote DNA damage¹⁰⁸ and ESC differentiation, thereby selecting for p53 dependence.¹⁰⁹ ESCs under stress characteristically undergo either differentiation or apoptosis.^110,111 In fact, the culture conditions used to convert naïve ESCs into 2iESCs enforce self-renewal and a dramatic loss of spontaneously differentiating cells; neither primed ESCs nor differentiated somatic cells survive these conditions.¹¹² Remarkably, 2 studies used blastocyst derived-ESCs from the same source (Rudolf Jaenisch, MIT, Cambridge, MA) but reported contradictory results. Culturing ESCs to “sub-confluence” before adding doxorubicin¹⁰¹ might have created conditions in which excessively high concentrations of doxorubicin-induced apoptosis in p53^+/+ cells more rapidly than in p53^−/− cells.^51,93

Time

The only effect of p53 on apoptosis in naïve ESCs was to accelerate its initiation. Once initiated, apoptosis continued at the same rate and to the same extent as in the absence of p53. Experiments with a single time point and a single-drug concentration cannot reveal the relationship between DNA damage and the significance of p53.

DNA Damage

As little as 0.05 µM doxorubicin is sufficient to induce apoptosis in either p53^+/+ and p53^−/− naïve ESCs. Yet most studies used from 0.5 µM to 1.8 µM. With naïve ESCs, high doxorubicin concentrations-initiated apoptosis more quickly in p53^+/+ ESCs than in p53^−/− ESCs, but once initiated, PCD occurred at equivalent rates and to equivalent extents (Fig. 2D). With 2iESCs, one study used 1 µM doxorubicin for 6 h and concluded that the DNA damage response was p53-dependent.¹⁰⁶ A second study used 1 µg/mL (1.84 µM) doxorubicin for 48 h and concluded that the DNA damage response was p53-independent.¹⁰⁵ Still a third study cultured 2iESCs with 1 μM doxorubicin for 16 h and observed that p53^+/+ cells underwent apoptosis more quickly than p53^−/− cells (63% p53^+/+ cells vs. 13% p53^−/− cells).¹⁰⁴ These results might be reconciled if both studies avoided excessively high concentrations of doxorubicin and monitored the effects of doxorubicin over time.

Viability

Two studies concluded that DNA damage-induced PCD was p53-dependent in naïve mESCs⁹⁵ and hESCs,¹⁰² because cells in which p53 was suppressed constitutively by shRNA did not exhibit doxorubicin-induced apoptosis. However, this technology raises 2 caveats. First, isolation of viable clones also selects for “off-target” mutations that promote cell proliferation or prevent cell death, as evidenced by the fact that constitutive suppression of p53-expression ESCs and embryos promotes clonal heterogeneity by disrupting DNA methylation homeostasis.¹¹³ Furthermore, since ESCs under stress characteristically undergo either differentiation or apoptosis,^110,111 changes observed in gene expression and relocalization of p53 from the cytoplasm to the nucleus are characteristics of ESC differentiation as well as apoptosis. The same caveats apply to the application of CRISPR-Cas9 technology to ablate p53 in 2iESCs.^104-106

p53 Null Mutation

Comparing p53^+/+ with p53^−/− ESCs is essential to establish a role for p53. Two studies with contradictory conclusions relied on inadequately characterized ESCs.^97,100 Another study on hESCs based its conclusion solely on changes in p53 expression in response to apoptotic stimuli.¹¹⁴ Studies that rely upon changes in p53 protein in response to stress and p53 inhibitors ignore the fact that the p53 transcription factor regulates at least 343 target genes involved in maintaining genomic stability, cell differentiation, cell senescence, cell cycle regulation, and PCD.⁷ The fact that ectopic over-expression of certain p53 mutations also suppressed doxorubicin-induced apoptosis¹⁰¹ simply reflects the fact that p53 affects expression of hundreds of different genes, some of which affect apoptosis. Many naturally occurring p53 mutations have the opposite effect; they gain additional oncogenic functions that endow cells with growth and survival advantages.¹¹⁵

Reproducibility

Two studies using the same source of ESCs (Yang Xu, Univ. California, San Diego) concluded that p53 is not required for cell cycle arrest^51,96 and their results with p53^+/+ ESCs are indistinguishable.^51,96 However, one study concluded that p53 is essential for doxorubicin-induced apoptosis⁹⁶ whereas the other study concluded that it is not.⁵¹ The first study relied on caspase-3 cleavage to confirm apoptosis, which they detected with a monoclonal antibody specific for the cleaved form. Thus, the fact that the extent of CASP3 cleavage was insignificant was not recognized. Moreover, the time delay for initiation of apoptosis exhibited by p53^−/− ESCs cultured with excess doxorubicin delayed the appearance of cleaved caspase-3, thereby allowing cleaved-caspase-3 to be detected in p53^+/+ cells under conditions where it appeared to be absent in p53^−/− cells. Apoptosis is also delayed in p53^−/− ESCs cultured under stress, such as the extremely high seeding density (260 000 cells/cm²) used in the first study.⁹⁶

Conclusions

Of the 12 forms of PCD described in human cells, only noncanonical apoptosis, autophagy-dependent, and lysosome-dependent PCD have been reported in ESCs, preimplantation, or gastrulating embryos. However, autophagy-dependent PCD might be confused with autophagy disruption which could activate non-canonical apoptosis, lysosome-dependent PCD, or necroptosis. Another candidate is parthanatos.

The importance of p53 in PCD has been characterized extensively, but conclusions are often enigmatic. Based solely on studies comparing wild-type with p53^−/− ESCs, MEFs, or mice, 3 conclusions appear uncontested; p53 is not required for activation of the G2-checkpoint, for embryonic lethality prior to gastrulation, or for embryonic development. The form of PCD and the role of p53 might change as preimplantation embryos develop from totipotent (2iESCs) to pluripotent (naïve ESCs) to primed pluripotent cells in post-implantation embryos (hESCs, mEpiSCs). However, contradictory conclusions concerning the role of p53 during PCD in ESCs can be reconciled by differences in experimental conditions, such as the amount of stress and the length of time stress was induced, culture conditions, and assay conditions.

In mice, p53 dependent transcription is first evident in late-stage blastocysts, and the ability of p53 to induce embryonic lethality is first evident during gastrulation. Depending on experimental conditions, p53 can accelerate initiation of PCD in mESCs and late-stage blastocysts, but once initiated, PCD occurs at equivalent rates and to equivalent extents regardless of the presence or absence of p53. Following either mESC differentiation in vitro or the formation of MEFs in vivo, DNA damage induces p53-dependent cell cycle arrest and senescence. Given the sensitivity of MEFs to p53-dependent senescence, failure of embryonic development likely results from cell senescence rather than PCD, although excessive DNA damage induces PCD.

Funding

The work was funded by the National Institute of Child Health and Human Development (NICHD) Intramural Research Program, Grant/Award Numbers: ZIA HD000506, ZIA HD000507.

Conflict of Interest

The authors declare no potential conflicts of interest.

Author Contributions

All the authors wrote and revised the manuscript. All the authors have read and agreed to the published version of the manuscript.

Data Availability

No new data were generated or analyzed in support of this research.