p53: emerging roles in stem cells, development and beyond

The gene TP53 (Trp53 in mice), which encodes the transcription factor p53, is the most frequently mutated tumor suppressor gene in human cancers (Bouaoun et al., 2016; Vousden and Prives, 2009). Since its discovery more than three decades ago, the molecular mechanisms involved in the selection and execution of the many functions of p53, as well as how they culminate in safeguarding genomic stability and suppressing tumor development, continue to unfold. Loss of p53 transcriptional activity, by mutations in TP53 or the activation of pathways that repress p53, are major contributing factors to malignant transformation. p53 safeguards the genome by restricting chromosomal instability through its ability to eliminate cells at risk of aberrant mitoses (Eischen, 2016). Accordingly, numerous in vivo and in vitro studies have revealed that the loss of p53 function both facilitates the accumulation and permits the survival of aneuploid cells. Genomic instability fueled by p53 loss also leads to the acquisition of additional cancer driver events with the potential to accelerate transformation, metastasis and drug resistance (Eischen, 2016).

In normal cells, p53 expression levels are low, and an initial response to stress-induced signaling to p53 is disruption of the activity of E3-ubiquitin ligases, such as MDM2, that maintain low levels of p53 by ubiquitylation and protein degradation (recently reviewed by Pant and Lozano, 2014). The consequences of unchecked p53 activity during embryonic differentiation, and support for tight regulation of p53 during development, were illustrated by the early embryonic lethality of Mdm2^−/− mice at implantation, a phenotype that is rescued by deletion of Trp53 (de Oca Luna et al., 1995; Jones et al., 1995). The subsequent creation of elegant mouse models that expressed mutated forms of p53, with or without wild-type p53, revealed that p53 functions at multiple stages of embryonic development (Van Nostrand et al., 2014), as well as in aging (Tyner et al., 2002b). These mice also provided models of human genetic diseases and of pathologies associated with defective ribosome biogenesis (McGowan et al., 2008). Together, these studies showed that normal differentiation, development and aging require p53 levels to be precisely regulated in a spatial and temporal manner.

Despite these early findings, the precise roles of p53 in differentiation and development have remained relatively under-studied. However, recent genome-wide profiling studies of embryonic stem cells (ESCs) and adult stem cell populations, together with a more in-depth analysis of the developmental defects in mice devoid of Trp53 (Clarke et al., 1993; Donehower et al., 1992; Lowe et al., 1993), have revealed that p53 functions appear to be intertwined with stem cell biology and differentiation in the soma of higher organisms. Here, we review these studies, providing an overview of the modes of action of p53 and its function in development and stem cells, and highlighting how the developmental roles of p53 relate to its well-known functions in tumor suppression.

p53 functions as a regulatory node. It receives signals, which are modulated and relayed in a cell- and context-dependent manner, to direct a variety of downstream outcomes, including cell cycle arrest, apoptosis, senescence, metabolic regulation and other responses that promote the repair and survival or death and elimination of abnormal cells (Vousden and Prives, 2009). Adding complexity to the p53 regulatory network are the potential modulatory roles played by other p53 family members, p63 and p73, that are found in mammals. TP53, TP63 and TP73 arose from a common ancestral gene, first detected in the evolution of modern-day anemones to protect the germline from genomic instability (Belyi et al., 2010; Yang et al., 2002). TP53 of higher eukaryotes diverged from TP63/TP73 before the appearance of bony fishes (Lane et al., 2011) and acquired tumor-suppressive activities not shared by TP63 and TP73, both of which display clear involvement in embryonic development (reviewed by Belyi et al., 2010). All p53 family members have a conserved protein domain structure (Fig. 1) that includes: an N-terminal transactivation (TA) domain (Lin et al., 1994); a proline-rich (PR) region, which is implicated in apoptosis and protein-protein interactions (Walker and Levine, 1996); a DNA-binding domain (DBD) that recognizes a core sequence motif of 10-base pairs (PuPuPuCA/TA/TPyPyPy, where Pu=purine, Py=pyrimidine), repeated with varied nucleotide spacing within p53-regulatory elements of genes (El-Deiry et al., 1992); and an oligomerization domain (OD) that mediates p53 tetramer formation (Jeffrey et al., 1995). p63 and p73 have an additional sterile α motif (SAM), which is involved in protein-protein interactions.

Further adding to the complexity, it has been shown that p53 family members exist as various isoforms. The alternative splicing of the C-terminal exons of TP63 and TP73 results in at least three isoforms of TP63 (α, β, γ) and at least seven isoforms of TP73 (α, β, γ, δ, ε, ζ, η) (Bénard et al., 2003; Bourdon et al., 2005). A conserved feature of the p53 family is the presence of potential transcription start sites from intronic alternative promoters that generate N-terminal truncated isoforms (ΔNp53 or Δ40p53, ΔNp63 and ΔNp73), which can exert dominant-negative effects on p53, p63 and p73 (Bénard et al., 2003; Courtois et al., 2002). In addition, an internal promoter in intron 4 of TP53 results in an N-terminal truncated isoform of p53 (Δ133p53) that is devoid of the transactivation and proline-rich domains (Bourdon et al., 2005). N-terminal variants of TP53 are also expressed in a cell-type specific manner and can activate or repress the transactivation of specific p53-target genes (Engelmann and Putzer, 2014). For example, Δ40p53 is highly expressed in mouse ESCs (mESCs), as the major isoform during early stages of mouse embryogenesis, and haploinsufficiency of Δ40p53 causes loss of pluripotency and acquisition of a more somatic cell cycle (Ungewitter and Scrable, 2010). These truncated isoforms of the p53 family have been detected in late-stage tumors and metastases, in correlation with cancer cell survival and tumor growth (reviewed by Wei et al., 2012). Isoform-specific functions of p53 family have been delineated using elegant mouse models (Table 1); however, the specific or interactive contributions of p53, p63 or p73 family members and their respective isoforms to malignant transformation remain elusive, adding to the complexity of p53-mediated tumor suppression.

The major phenotype of Trp53-null mice is tumor development and resilience to radiation-induced apoptosis (Clarke et al., 1993; Donehower et al., 1992; Lowe et al., 1993). Given its importance in tumor suppression, the ability of the Trp53-null mouse to gastrulate, develop and apparently thrive, until succumbing to a strain-dependent profile of tumors at 3-5 months of age, was surprising and supported a limited role for p53 in stem cell differentiation and development (Bieging et al., 2014; Jacks et al., 1994). However, a more-detailed analysis reveals that a considerable fraction of female Trp53^−/− embryos exhibit failure in neuronal tube closure, leading to exencephaly in 23% of mutant embryos on a 129/Ola background, or cranio-facial abnormalities, including ocular abnormalities and defects in upper incisor tooth formation (Armstrong et al., 1995; Kaufman et al., 1997; reviewed by Shin et al., 2013) (Table 1). In addition, p53 deletion in C57BL/6J background mice results in a lower than expected number of surviving homozygotes (14.3%) and these animals suffer from severely abnormal lung architecture, cleft palate (Tateossian et al., 2015), craniofacial defects in skeletal, neuronal and muscle tissues (Rinon et al., 2011), and a spectrum of congenital abnormalities in the urinary tract and kidney (Saifudeen et al., 2009). Both sexes of Trp53^+/− and Trp53^−/− mice show significant dwarfism or under-development (Baatout et al., 2002), and Trp53^−/− female mice that live to adulthood exhibit low fecundity due to loss of p53-dependent expression of Lif1 (Hu et al., 2007). Knock-in mouse strains that express a transcriptionally dead variant of p53 (p53^25,26,53,54) along with a wild-type Trp53 allele suffer late-gestational lethality associated with phenotypes consistent with the human CHARGE syndrome (Van Nostrand et al., 2014). Thus, p53 functions clearly impinge on normal mouse development and, when perturbed, give rise to distinct phenotypes, although these are perhaps not as dramatic as expected given the roles of p53 in tumorigenesis.

A role for p53 in development has also been demonstrated in studies using other model organisms. For example, Xenopus laevis embryos that lack p53 expression have severe gastrulation defects, in sharp contrast to the mostly normal early stages of development of Trp53-null mice (Wallingford et al., 1997). This species-specific difference may be due to lack of the p53 mammalian family homologues, p63 and p73, in Xenopus laevis (Stiewe, 2007). Furthermore, p53-mutant zebrafish have abnormal gut and neuronal development, whereas p53 inhibition in the salamander prevents limb formation; in planaria (which lack p63/p73), p53 loss disrupts stem cell-dependent regeneration (reviewed by Levine and Berger, 2017). As mentioned above, p53, p63 and p73 exhibit similarity with regard to their amino acid sequence and domain structure (Fig. 1), and possess the ability to bind the same consensus DNA-binding sites to varying degrees at overlapping sets of target genes (Levrero et al., 2000). Thus, in mammals, p63 and p73 may largely compensate for lack of p53 during development. Indeed, p53 wild-type mice that are null for either Trp63 or Trp73 have much more severe developmental phenotypes compared with Trp53-null mice (Table 1). Mice null for Trp63 are born alive but display severe developmental defects, such as absent or truncated limbs, due to the aberrant relaying of signals at the apical ectodermal ridge of the limb (Mills et al., 1999; Yang et al., 1999). Trp73-null mice are viable at birth but exhibit runting and a high mortality rate within the first 2 months (Yang et al., 2000). Together, these results from model organisms that lack multiple p53 isoforms or p53 family members support the idea that p53 family members partially compensate for each other and have isoform-specific and shared target genes (Danilova et al., 2008; Levine et al., 2011). However, using compound knockout mouse models, it was recently reported that the combined loss of p53 and p63 in mouse embryos does not appear to significantly compromise mouse development beyond simple p63 or p53 deficiency, and that Trp63^−/−; Trp73^−/− embryos show no dramatic developmental defects beyond those observed in single knockout embryos. These observations suggest that p53 family members may play redundant roles in specific development processes (Van Nostrand et al., 2017). This species and isoform specific disparity in p53 activities is most likely due to the complex layers of p53 modulation and the array of biological processes regulated by p53.

In response to a variety of cellular stresses, including DNA damage and replication stress often produced by deregulated oncogenes, p53 protein is stabilized. Through its DNA-binding ability, p53 governs a complex anti-proliferative transcriptional program, corresponding to an array of biological responses (Fig. 2). Mechanisms leading to the stabilization and activation of p53 are mostly stimulus specific. DNA-damage promotes post-translational modifications (PTMs) on p53, such as phosphorylation, acetylation or methylation (Dai and Gu, 2010), blocking MDM2-mediated degradation, whereas signaling as a result of oncogenic stress activates the ARF tumor suppressor to inhibit MDM2 (Zhang et al., 1998). The tumor suppressor activities of p53 are largely attributed to its ability to promote cell cycle arrest and apoptosis depending on cell type and stimulus, a context specificity not completely understood. However, this historic view of the effects of widespread TP53 mutations in tumors is changing with recent revelations of broadly diverse consequences of p53 activation to safeguard our genome. An ever-growing body of evidence suggests that, in addition to regulating cell cycle arrest and apoptosis, p53 controls ‘non-canonical’ programs such as autophagy, metabolism, repression of pluripotency and cellular plasticity, and ferroptosis, all of which contribute to its tumor suppressor functions (Fig. 2). Consistent with this notion, mice deficient for p53 target genes responsible for cell cycle arrest and apoptosis, such as p21 (Cdkn1a), Puma (Bbc3) and Noxa (Pmaip1) (i.e. p21^−/−Puma^−/−Noxa^−/− mice) do not develop lymphomas or other malignancies, as observed in Trp53-null mice, suggesting that p53-mediated cell cycle arrest and apoptosis are not sufficient for tumor suppression (Valente et al., 2013).

This diversity in p53 functions depends on several factors, including cell or tissue type, epigenetic state, differentiation state, stress conditions, collaborating environment signals, specific-PTMs on p53 and associated transcription co-regulatory factors that dictate the choice of target genes. Moreover, the crosstalk between p53-PTMs also suggests that one PTM may enhance the acquisition of another, unlocking additional layers of p53 stability and biasing p53 binding to DNA on selected target genes. It is fair to speculate that the tumor suppressive activities of p53 are not limited to DNA-damage response in differentiated cells. Consistent with this notion, based on recent genome-wide analyses comparing p53-DNA associations with gene expression profiles, a better picture of the functions of p53 in somatic cells versus stem cells is now emerging (Fig. 2): in somatic cells under conditions of stress, p53 regulates a plethora of genes that lead to a variety of cellular outcomes, including cell cycle arrest and apoptosis; conversely, p53 in stem cells regulates pathways that target the pluripotency network (described below), all of which likely contribute to tumor suppression.

The possibility that p53 functions in early development and cell differentiation arose from the discovery that, unlike in somatic cells, p53 is expressed at relatively high levels in all cells of day 8.5 post coitum (p.c.) and day 10.5 p.c. mouse embryos, with expression declining in terminally differentiated cells (Schmid et al., 1991). Multiple studies have since shown that p53 is implicated in cell differentiation but that cellular context plays a major and, as yet, not fully defined role in this p53 function. For example, p53 negatively regulates the proliferation and self-renewal of neural stem cells and hematopoietic stem cells to maintain their quiescent state (Liu et al., 2009; Meletis et al., 2006), whereas, under the influence of specific hormonal or chemical inducers, p53 promotes the differentiation of both mouse and human ESCs (Akdemir et al., 2014; Jain et al., 2012, 2016; Li et al., 2012; Lin et al., 2005) (Fig. 3). The transcriptional activity of p53 is also crucial for regulating the status of both ESCs and adult stem cells (discussed in detail below), facilitating specific differentiation programs while inhibiting others (Aylon and Oren, 2016; Spike and Wahl, 2011). ESCs possess robust mechanisms to preserve their genomic integrity and to avoid the propagation of genetic aberrations to their descendent somatic cells (Hong et al., 2007). Given that pluripotent and self-renewing ESCs share some but not all of the cellular and molecular phenotypes of aggressive, p53-mutant cancers [e.g. an ESC-like gene signature is observed in p53 mutant breast cancer (Mizuno et al., 2010)], p53 might function to promote differentiation pathways of ESCs via mechanisms that likewise safeguard genomic stability and DNA fidelity to prevent cancer development.

Numerous studies have reported that inactivation of p53, or disruption of the pathways that activate p53, can increase the efficiency with which mature somatic cells can be reprogrammed to pluripotency to generate induced pluripotent stem cells (iPSCs) (Hong et al., 2009; Kawamura et al., 2009; Marion et al., 2009; Utikal et al., 2009). These findings further underscore the ability of p53 to prevent ‘backsliding’ or the dedifferentiation of somatic cells (reviewed by Krizhanovsky and Lowe, 2009). As the generation of iPSCs and the malignant transformation of somatic cells share many common characteristics, such as unlimited proliferation, similar metabolic status and transcriptional activity of pluripotency factors Oct4 and Myc (Semi et al., 2013), it is not surprising that p53 acts as a barrier to reprogramming. However, the constant suppression of p53 during reprogramming allows widespread genomic instability to occur in the resulting iPSCs (Marion et al., 2009), whereas its reactivation during reprogramming disrupts the formation of iPSCs and induces their differentiation once formed (Yi et al., 2012). Several pathways have been implicated in the ability of p53 to counteract the reprogramming of somatic cells. These include the p53 gene targets miR-34a, which represses several pluripotency genes [including Sox2, Myc and Nanog (Choi et al., 2011), and CDKN1A, which attenuates cell division (Hanna et al., 2009). In addition, p53 restricts mesenchymal-to-epithelial transition (MET) during early reprogramming, which is primarily mediated by the ability of p53 to inhibit the activation of specific epithelial genes (Brosh et al., 2013). Collectively, these studies suggest that p53 governs the transition between cell states and limits the ability of somatic cells to undergo epigenetic reprogramming into iPSCs to play a direct role in regulating cellular plasticity.

ESCs, which are derived from the inner cell mass of mouse embryos, exhibit self-renewal, an unlimited potential to proliferate in vitro and pluripotency, the capacity to develop into all the cell types of the embryo proper (Thomson et al., 1998). Several core transcription factors (including Oct4, Sox2, Nanog and Klf4) act in an intricate gene regulatory circuitry that maintains the pluripotent status of ESCs by regulating specific transcriptional programs (Boyer et al., 2005; Jaenisch and Young, 2008). Deregulation of the core pluripotency network, coupled with alterations to the epigenetic status of ESCs, then contributes to their transitioning from pluripotency to differentiation. In recent years, a number of studies have shown that p53 plays multiple functions, and hence must be tightly regulated, in regulating the pluripotent state and the transition to differentiation in both ESCs and iPSCs.

Several early studies have reported that a measurable decrease in p53 RNA and protein levels occurs during mESC differentiation and mouse development in vivo (Lin et al., 2005; Rogel et al., 1985; Sabapathy et al., 1997). Furthermore, the re-expression of p53 in undifferentiated Trp53-null mESCs drives them towards a more differentiated state (Komarov et al., 1999; Sabapathy et al., 1997). One possible mechanism for this is the direct repression of Nanog expression by p53, which is sufficient to drive mESCs towards differentiation (Fig. 3) (Lin et al., 2005). Nanog is essential for maintaining the self-renewal and pluripotency of ESCs, and its expression is controlled by members of the core ESC circuitry (including Oct4 and Sox2) (Silva et al., 2009; Young, 2011), which are displaced from the Nanog promoter by p53 during differentiation. In order to directly repress Nanog, p53 binds a consensus regulatory element (p53RE) on the Nanog promoter and recruits the mSin3a-HDAC complex to bring about histone deacetylation and chromatin repression (Lin et al., 2005). Recently, using an in-depth genome-wide analysis of p53 chromatin-interactions and gene expression, it was shown that p53 both activates differentiation-associated genes and represses stem cell-specific genes in response to DNA damage in mESCs (Fig. 3) (Li et al., 2012). However, as discussed below, these p53-regulated outcomes in mESCs are in contrast to those observed in human ESCs (hESCs), where p53-regulated chromatin interactions and gene expression changes in response to DNA damage versus differentiation are distinct.

One way in which p53 activity is held in check by iPSCs and ESCs is via post-translational modifications. These modifications include the ubiquitylation, acetylation, phosphorylation, methylation or sumoylation of specific residues of p53 (Jain and Barton, 2010). For example, aurora A kinase phosphorylates mouse p53 at Ser212 and Ser312, resulting in the inactivation and inhibition of p53-directed ectodermal and mesodermal gene expression (Lee et al., 2012). This phosphorylation of p53 impairs p53-induced ESC differentiation and p53-mediated suppression of iPSC reprogramming. Thus, aurora kinase-p53 signaling regulates self-renewal, differentiation and somatic cell reprogramming (Fig. 3). The treatment of mESCs containing a humanized p53 allele knock-in (p53hki) with retinoic acid (RA) stabilizes p53 as a result of Ser315 phosphorylation by cyclin-dependent kinase (CDK), resulting in differentiation. Accordingly, a S315A missense mutation into the endogenous p53hki allele in these mESCs results in loss of S315 phosphorylation and abrogation of p53 transcriptional activity, including Nanog repression, during RA-induced ESC differentiation (Lin et al., 2005). On the other hand, the DNA damage-induced phosphorylation of Ser18 stabilizes and activates p53 to elicit its effect in antagonizing pluripotency and promoting differentiation of mESCs (Li et al., 2012). Phosphorylation at Ser18 by ATM kinase is a prerequisite for p53 activation during DNA damage and cellular protection. A knock-in mouse model of Trp53 harboring a mutant Ser18 that cannot be phosphorylated suffers from several malignancies, including fibrosarcoma, leukemia, leiomyosarcoma and myxosarcoma, which are unusual in p53 mutant mice, suggesting that the phosphorylation of p53 at Ser18 contributes to tumor suppression in vivo (Armata et al., 2007). Together, these studies indicate that the post-translational modifications incorporated into p53 during the in vitro differentiation of ESCs play an essential role in enforcing the tumor suppressive activities of p53 in vivo.

Unlike mESCs, which express relatively high levels of (mostly cytoplasmic) p53 protein (Han et al., 2008; Sabapathy et al., 1997), human ESCs have comparatively low levels of p53 due to its control by the negative regulators, MDM2 and TRIM24, which ubiquitylate p53 and trigger its degradation, potentially acting together with other E3 ligases (Jain et al., 2012; Setoguchi et al., 2016). Increasing the levels of p53 in human ESCs, either by using a small molecule inhibitor of MDM2 (Nutlin) (Maimets et al., 2008) or by stimulating DNA damage (Qin et al., 2007), induces apoptosis and differentiation. It has also been shown that p53 protein in hESCs is located in the nucleus but is inactive, owing to a lack of acetylation at lysine residue K373 (Jain et al., 2012). This reduced acetylation of p53 in hESCs occurs as a result of the OCT4-mediated transcriptional activation of SIRT1, an NAD⁺-dependent deacetylase that deacetylates p53 (Zhang et al., 2014). This inactive form of p53 is vulnerable to degradation by its negative regulators MDM2 and TRIM24. The induction of hESC differentiation by the addition of RA leads to the acetylation of K373 on p53 by the CBP/p300 histone acetyl transferases and to the dissociation of p53 from its negative regulators, resulting in its stabilization and activation (Jain et al., 2012). In order to achieve sustained differentiation, transcriptionally active p53 activates the transcription of CDKN1A (p21), which impedes cellular proliferation, and two small non-coding RNAs, miRNA-34a and miRNA-145, which negatively regulate a set of transcription factors (Oct4, Sox2, Lin28a and Klf4) that favor stem cell maintenance (Fig. 3) (Jain et al., 2012). In line with these findings, it has been shown that the exogenous expression of p53 in hESCs induces their spontaneous differentiation, while the expression of a DNA-binding-deficient p53 mutant does not (Jain et al., 2012). Overall, these findings indicate that p53 transcriptionally regulates hESC differentiation, in part by activating both protein-coding and non-coding genes.

Genome-wide comparisons of p53 chromatin interactions in mESCs and hESCs (via ChIPseq) show that many p53 target genes are evolutionarily conserved, although their inductive signaling, developmental timing and dominant pathways may differ (Akdemir et al., 2014). Recently, p53 ChIPseq and global gene expression analyses of pluripotent, RA-treated and doxorubicin (adriamycin, Adr)-treated hESCs were performed to compare p53-transcriptional activity during differentiation and DNA damage (Akdemir et al., 2014). This analysis revealed that the RA-mediated transcriptional outcomes of p53 activity are quite distinct from its stress-responsive (DNA damage) regulation in hESCs. In particular, the results showed that p53 promotes hESC differentiation by activating the expression of developmental transcription factor genes that are involved in patterning, morphogenesis and organ development (Fig. 3). This activated cascade of transcription factors amplifies the outcomes of p53 induction beyond the transient time period when p53 protein is elevated (Jain et al., 2012). RA-induced, differentiation-specific p53 gene targets in hESCs include those encoding homeodomain proteins such as HOX (homeobox), LHX (LIM homeobox), DLX (distal-less-like) and PAX (paired box) (Gudas and Wagner, 2011; Hobert and Westphal, 2000), and specific forkhead (Lehmann et al., 2003), SOX (Sry-related HMG box) (Schepers et al., 2002) and TBX (T-box) family members (Showell et al., 2004). In addition, p53 targets CBX2 (chromobox) and CBX4, which are part of the polycomb group multi-protein PRC1-like transcription repression complex and are crucial for cell fate determination (Morey and Helin, 2010). One mechanism by which p53 achieves differentiation-specific gene activation is by recruiting UTX, a histone H3K27me3-specific demethylase, to the promoters of these developmental genes. This, along with alterations in the genomic profiles of H3K4me3 and H3K27me3 on p53-regulated gene targets, suggests that p53, by recruiting specific epigenetic regulators, plays a role in modifying chromatin structure to activate a developmental transcriptional program.

Recently, a role for the p53-mediated regulation of long non-coding RNAs (lncRNAs) during ESC differentiation and has also been uncovered. Non-coding RNA transcripts that are longer than 200 nucleotides, often poly-adenylated and devoid of evident open reading frames (ORFs) are called lncRNAs, and these RNAs are generally poorly conserved among species (Fatica and Bozzoni, 2014). LncRNAs are involved in diverse biological processes and often interface with epigenetic machinery as effector molecules that fine-tune transcriptional regulation (Rinn, 2014). Their expression is highly cell- and tissue-specific and regulated by mechanisms that continue to be uncovered (Perry and Ulitsky, 2016). LncRNAs can function as ligands of proteins and can guide lncRNA-containing RNA/protein complexes to specific RNA and DNA sites (Guttman and Rinn, 2012; Wang and Chang, 2011). Thus, lncRNAs can operate through different modes, as signals, scaffolds for protein-protein interactions, molecular decoys or guides to target elements in the genome (Wang and Chang, 2011). Recently, by integrating genome-wide expression data with p53 chromatin-enrichment profiles in hESCs undergoing differentiation, a high-confidence signature of 40 p53-regulated lncRNAs that are regulated by p53-binding to their promoters during hESC differentiation was identified (Fatica and Bozzoni, 2014; Jain et al., 2016). This included differentiation-specific lncRNAs, such as HOTAIRM1, and several pluripotency-specific transcripts that were named lncPRESS (p53-regulated and ESC-associated) transcripts, the expression of which correlated with their histone modification profiles and with p53 regulation (Fig. 3). LncRNA transcripts that are highly expressed in pluripotent hESCs and repressed by p53 during hESC differentiation play a modulatory but significant role in stem-cell maintenance. For example, lncPRESS1, a human-specific lncRNA, acts as a ‘molecular decoy’ that sequesters SIRT6, a NAD⁺-dependent deacetylase that specifically deacetylates H3K56ac and H3K9ac in the chromatin of ESCs (Jain et al., 2016). Acetylation of H3K56 is linked to the core pluripotency circuitry and to the transcriptional activation of pluripotency genes in hESCs (Xie et al., 2009). In addition, LncPRESS4, another p53-repressed lncRNA identified in hESCs (Jain et al., 2016), also known as TUNA (Tcl1 upstream neuron-associated or linc86023), is required for the maintenance of pluripotency (Lin et al., 2014). Thus, p53 in ESCs has multiple functions (summarized in Fig. 3): (1) it represses pluripotency by transcriptionally activating non-coding RNAs (both miRNAs and lncRNAs); (2) it promotes differentiation by directly activating developmental genes and transcription factors; and (3) it provides surveillance for secure, genome-stable differentiation (Hamilton and Brickman, 2017).

One issue concerning p53 and its family members in promoting ESC differentiation is whether p53 plays a role in specifying particular germ layers, i.e. endoderm, mesoderm or ectoderm. Indeed, although the gain- or loss-of-function of p53 in various in vitro models support the notion that p53 regulates specific genes in many differentiation processes, support is strongest for roles for p53 in mesenchymal differentiation programs (reviewed by Molchadsky et al., 2010). For example, the exogenous expression of p53 in an undifferentiated pre-B cell line induces differentiation to B cells (Aloni-Grinstein et al., 1993). Likewise, p53 is essential for skeletal muscle differentiation and for the osteogenic re-programming of skeletal muscle-committed cells (Molchadsky et al., 2008). In addition, monolayer cultures of p53-null mESCs, when subjected to LIF withdrawal, fail to undergo mesodermal differentiation (Shigeta et al., 2013). Finally, by leveraging methods that enable hESCs to be differentiated in the direction of specific lineages (Gifford et al., 2013) specifically toward definitive ectoderm, endoderm and mesoderm, it was shown that p53 preferentially activates lineage-identity genes, including lncRNAs, to promote a mesendodermal state, i.e. a bipotential state that can give rise to both mesoderm and endoderm (Jain et al., 2016).

Previous in vivo studies have established that synergy between Wnt and Nodal-related TGF-β signaling within the mouse primitive streak is involved in the formation of mesendoderm (Conlon et al., 1991). The only direct link previously established between p53 and mesoderm induction was made in Xenopus laevis embryos: p53-depleted Xenopus embryos fail to gastrulate and the overexpression of p53 induces mesoderm specification (Cordenonsi et al., 2007). This study revealed that the regulation of mesoderm gene expression by p53 in Xenopus involves cooperation between TGFβ and p53 signaling. In particular, it was shown that the RTK/Ras/MAPK (mitogen-activated protein kinase)-mediated phosphorylation of p53 during development promotes p53 interaction with TGFβ-activated Smads to regulate a specific class of p53-dependent Nodal/Smad gene targets that play roles in mesoderm formation (Cordenonsi et al., 2007). However, the mechanism by which p53 regulates mesoderm specification in other contexts and organisms is unclear.

How functional redundancy between p53 family members affects lineage specific differentiation is also not fully understood, although a recent study has provided some insight into this issue (Wang et al., 2017). This work, which used a compound triple-knockout (TKO) of all three p53 family members in mouse and human ESCs, showed that all three p53 family members are required for the induction of Nodal-responsive genes and for mesendoderm specification, and that this requirement is conserved between mouse and human ESCs (Wang et al., 2017). Members of the p53 family control a gene regulatory network that integrates Wnt and TGFβ nodal inputs for mesendoderm specification (Wang et al., 2017). This network is composed of multiple layers of regulation: the p53 family activates Wnt signaling by directly activating Wnt3 and Fzd1 as cells exit from pluripotency, and Wnt3 and Fzd1 are absolutely required for the induction of a class of Smad2/3-responsive, mesendoderm-specific genes (Fig. 3) (Wang et al., 2017). These studies using compound knockout ESC models confirm that p53 family members are functionally redundant with each other during mesendoderm specification, and that this redundancy might also mask their roles in gastrulation in single or double knockout mice.

As we have highlighted above, a closer look at the consequences of Trp53 loss indicate that p53 plays important roles in embryonic development. A growing body of evidence also now shows that p53 has additional functions in regulating tissue homeostasis. For example, p53 restricts the self-renewal of various stem and progenitor cells when subjected to oncogenic stress, probably by activating differentiation-inducing checkpoints (Tschaharganeh et al., 2014; Wang et al., 2012; Zhao et al., 2010). The self-renewal capacity of stem cells and the existence of stem-like cells in cancers (cancer stem cells) suggest that these cell types are likely candidates to initiate tumor formation and/or promote metastases. How the physiological and developmental functions of p53 intersect with the cancer-associated phenotype of p53 loss is quite intriguing. Malignant transformation proceeds by evading terminal differentiation, and p53 loss is likely one route to abate this innate barrier to tumorigenesis. Consistent with this notion, striking similarities between the gene expression signatures of aggressive breast cancer tumors that contain TP53 mutations and those of pluripotent ESCs have been observed (Kim and Orkin, 2011; Mizuno et al., 2010). In addition, it has been reported that mutations of TP53 facilitate the expansion of hematopoietic stem cell (HSC) clones in otherwise healthy individuals, sometimes taking over the entire hematopoietic system and resulting in hematological malignancies (Xie et al., 2014), while the expansion of p53-mutant HSC clones by genotoxic chemotherapy results in therapy-related acute myeloid leukemia (t-AML) (Wong et al., 2015). Loss of p53 can also facilitate lineage switching as a mechanism of resistance to anti-androgen therapy in prostate cancer (Mu et al., 2017). Thus, by promoting cellular homeostasis and regulating proliferation, p53 activity can lead to tumor suppression.

The regulation of differentiation or de-differentiation (reprogramming) by p53, either by employing its canonical functions of controlling cell cycle arrest and apoptosis or by directly regulating various lineage-specific programs, challenges the notion that p53 has no role in development. As aberrations in differentiation and de-differentiation programs can promote cell transformation, the abrogation of p53 activity, through deficiency or mutation, might result in the accumulation of oncogenic events and/or in the arrest of progenitor/stem cell differentiation, both of which are associated with tumor formation. Thus, by promoting directed differentiation and development, and by regulating cellular state, p53 contributes to the repertoire of functions that protect healthy cells and achieve tumor suppression.

In recent years, a growing body of evidence, aided by the development of new technologies, has revealed that the functions of p53 extend beyond cellular surveillance and apoptosis to include the regulation of cellular homeostasis and metabolism (Gottlieb and Vousden, 2010), inflammation (Cooks et al., 2014), reproduction (Hu et al., 2007; Levine et al., 2011), ageing (Tyner et al., 2002a), regeneration (Pomerantz and Blau, 2013) and more (Vousden and Prives, 2009). The broad involvement of p53 in promoting programs of differentiation and in antagonizing the de-differentiation of somatic cells relies on its fundamental role in regulating the maintenance of the cellular state, leading to the recent renaming of p53 as the ‘guardian of homeostasis’ (Aylon and Oren, 2016). Findings from numerous in vivo and in vitro studies now suggest that p53 likely plays a gatekeeper function to ensure high-fidelity development. This ‘gate-keeping’ role, which is essential for normal development, connects to the better understood roles of p53 as a tumor suppressor. However, despite decades of studies on the significance of p53 in tumor suppression, the identification of p53 mutations in a wide range of tumors and the determination of pathways that negatively regulate p53 in tumors, much remains to be learned about the roles and regulation of p53.

Exploiting our knowledge of p53 pathways in translation to clinical applications also remains a challenge. Targeting proteins that negatively regulate p53 seems to be an ideal choice for drug design. However, although many small molecules that disrupt the interactions of p53 with its negative regulator MDM2 have been designed (Vassilev et al., 2004), progress beyond the research bench has been slow; clinical outcomes may require wild-type p53 to respond to such inhibitors. Furthermore, although molecules that target mutant p53, either by changing its confirmation to wild-type p53 or promoting its degradation, show some promise (Parrales and Iwakuma, 2015), none of these has yet been approved for the treatment of cancer. It should also be noted that not all mutant forms of p53 are equivalent, hence each may require distinct approaches for targeting. Considering the complexities of p53 regulation and the multiple biological outcomes of p53 activation and inactivation, drug discovery to regulate this pathway thus remains complex. Future work will undoubtedly reveal gene targets (both coding and non-coding) and additional pathways that are controlled by p53 to fine-tune cellular protection. New technologies, together with an increased understanding of the complexities of p53 action in normal developmental, homeostatic and pathological contexts, will hopefully set the stage for clinical advances and drug discovery to combat p53-dependent cancers.

We thank all current and former members of Barton laboratory for helpful discussions. We regret not being able to cite all the work related to this Primer due to space limitations.