p53 is a fundamental determinant of oncogenesis, aging physiology and neurodegenerative pathologies. The complexity of the p53 regulatory network can hinder attempts to fully understand how this oncogenic protein operates within/between cells to constrain growth potential. Orthologs of p53 in non-mammalian models, such as zebrafish, Drosophila and Caenorhabditis elegans, afford simpler models that illuminate core properties of this ancient network. The existence of p53 in short-lived organisms, where cancers do not occur, argues that tumor suppression per se was not the evolutionary pressure shaping p53. Instead, p53 as a constraint against tumor growth was probably co-opted from more ancestral, nonautonomous functions that are entirely unknown. Describing these functional properties will be essential for a comprehensive picture of the p53 regulatory network in normal and disease states. The tractable systems described here offer sophisticated genetic tools and fundamental insights that will continue to be indispensable models for achieving this goal.

Introduction

The tumor suppressor gene p53 is mutated in more than 50% of human cancers. Three domains in p53, corresponding to transcriptional activation, DNA binding, and tetramerization activities, have been well characterized. In human cancer, most of the mutations occur within the DNA-binding domain. Surprisingly, it is also the most conserved domain from invertebrates to mammals. The major function of p53 is a sequence-specific transcription factor, which responds to several stress signals, including DNA damage, physiological stress, and oncogenic stimulation. The activation of p53 ultimately regulates DNA repair, cell cycle progression and apoptosis. In recent years, a vast literature, predominantly focused on transformed cells, has sought to understand the function of p53 as ‘the guardian of the genome’. However, cell culture-based studies provide only limited information about the nonautonomous roles of p53 in living organisms. Therefore, properties of the regulatory network of p53 that extend beyond the single cell level are almost entirely unknown. Our naivety in this area is easily exemplified by a glaring paradox: Everything we know about p53 and its role in damage response pathways predicts that transformed cells (p53−) should exhibit profound resistance to chemotherapeutic agents and radiation therapies relative to p53+ counterparts. And yet, the very fact that these anticancer agents actually have some efficacy in the clinic, combined with the high incidence of p53− state of human tumors, tells us just the opposite. That is, in patients, the p53− genotype of cancer tissues correlates with tumor sensitivity to genotoxic agents. These counter-intuitive lessons indicate that, despite a vast literature, there is still much to learn about this network. At the same time, they also suggest that properties deduced from studies at the single cell level may not be adequate to predict group behaviors of cells in tissues and tumors. The fact that p53 genes are conserved from invertebrate to mammals affords attractive opportunities for researchers to use well-defined genetic models to examine the functions of p53 in tissues and whole animal systems. p53 orthologs have been described in clams,1 squid,2 flies,3, 4, 5 frogs,6 and zebrafish.7 Knowledge of p53 in different genetic research models will be summarized in this review, and by considering this gene family from an evolutionary viewpoint, we highlight some conundrums in the field of p53 research and hope to shed light on how potential insights from these simpler models can advance therapeutic applications in cancer.

Zebrafish

Zebrafish (Danio rerio) p53 shares overall 48% sequence similarity to human p53.7 To study the in vivo function of p53, antisense morpholinos were injected into early embryos.8 In these studies, the native p53 locus remains intact but the gene product is reduced. p53 ‘knockdown’ embryos showed normal development, but had suppressed induction of apoptosis after UV irradiation or drug treatment. Several features of zebrafish p53 regulation are similar to mammals. First, like in mice and human, zebrafish MDM2 protein also negatively regulates p53. Second, the transcription level of p21 is regulated by p53 as a downstream effector. Third, p53 family members, p63 and p73 sequence also occur in zebrafish genome and each evidently has separate roles depending on the context of development, tissue specificity, and stress source. Recently, different splicing isoforms of human p53 mRNA transcripts were reported,9 although the tissue distribution and in vivo function of different isoforms remained unclear. In zebrafish, a truncated transcript, delta113p53 is induced by abnormal development and specifically engages in cell-cycle arrest but not in apoptosis,10 providing an intriguing clue into the distinct physiological roles of p53 isoforms during development. Given the similarities of p53 action between zebrafish and mammals, it has been proposed that zebrafish may become a useful model for future compound and genetic screens. This notion is supported by studies on the tumor suppressor activity of p53, since zebrafish strains harboring missense mutations in the DNA-binding domain of p53 showed higher susceptibility to neuronal tumors.11

Caenorhabditis elegans

C. elegans p53-like protein (CEP-1) is a 429 amino acids protein,12, 13 which does not share obvious overall homology to mammalian p53 members except in the DNA-binding domain. The most frequently mutated sites in DNA-binding domain are well conserved in cep-1. Unlike zebrafish, cep-1 is the only p53-related sequence in genome; no other p53 family members can be found in C. elegans. To understand the function of cep-1, a chromosomal rearrangement mutant, producing a dominant negative from of cep-1, was studied together with RNAi knockdown experiments. Both studies documented an essential role of cep-1 for radiation-induced apoptosis in germ cells. Although the native cep-1 locus still remains intact, the cep-1 phenotypes with RNAi are consistent with other animal models, where the gene is dispensable for normal development. Forced cep-1 expression caused wide-spread caspase-independent death, suggesting that the proper amount of CEP-1 is important for cell survival.12 cep-1 also mediates normal meiotic chromosome segregation and several stress-induced responses, but does not engage in cell-cycle arrest after DNA damage. How does cep-1 specify cell death? Induction of two BH-3 only proteins, EGL-1 and CED-13 in C. elegans, are thought to largely account for radiation-induced CEP-1-dependent cell death,14, 15 although whether direct binding by CEP-1 occurs at these loci remains unclear. As is the case in flies (see below), regulation of p53 by MDM2 is absent in C. elegans, albeit other DNA damage response pathways are still preserved. For example, an ATR ortholog in C. elegans, atl-1, also triggers apoptosis via cep-1/egl-1 pathway after radiation.16 Other recently reported p53 regulators in C. elegans such as GLD-117 and iASPP18 have not been examined in other systems. While the ATM/ATR mode of regulation appears well conserved here, it is not yet known whether these additional upstream regulators function similarly in other models.

Drosophila melanogaster

Drosophila p53 (dmp53) was first described as a 385 amino acids protein, with the highest homology to human p53 in the DNA-binding domain.3, 4, 5 More recently, a longer 495 amino acid isoform and a shorter 110 amino acid isoform were also reported.9 The anatomic structure of dmp53 is similar to mammalian p53. Moreover, the frequently mutated sites in human tumor are also well conserved. As in C. elegans, p63/73 orthologs do not occur in the Drosophila genome. The in vivo apoptotic activity of dmp53 was initially shown by ectopic overexpression in the eye, which induced apoptosis.3, 5 These findings were further supported by genetic loss-of-function studies, which established that dmp53 is required for transcriptional induction of rpr proteins (rpr, hid, skl) and for radiation-induced apoptosis.3 General development was not affected in the dmp53 mutants,19, 20, 21 but mild defects in longevity and fertility were found. Where studied, dmp53 does not engage damage-induced cell cycle checkpoints and, consistent with this, the Drosophila ortholog of p21 is unresponsive to irradiation.22 As in C. elegans, MDM2 is absent from the fly genome, implying that alternative modes of p53 regulation must exist. Although a full panel of upstream activators in the fly p53 regulatory network are not yet defined, we do know that Drosophila CHK2 regulates dmp53 via direct phosphorylation23 and that CHK2 and dATM are required for damage-induced apoptosis.24, 25 Moreover, a recent study reported that dmp53 can be activated by reduced ATP level in a cytochrome oxidase subunit mutant,26 suggesting intriguing links between energy status and dmp53 regulation. This same study implicated cyclin E, in addition to IAP antagonists, as a possible effector of dmp53. To predict additional dmp53 target genes in vivo, genome-wide array experiments compared the radiation responses of normal versus dmp53-null embryos and, along with proapoptotic functions, genes involved in DNA synthesis and repair were also found.23, 27

Ancient Effectors and Activators of p53

By comparing and contrasting p53 regulatory networks in simpler systems (schematized in Figure 1), we might perhaps deduce a set of ancient regulators and primitive effectors that define the core evolutionary circuit for p53 action. In zebrafish, upstream activators include signals generated by suboptimal cell replication, oncogenic stress, and genotoxic stress. While the former two conditions are yet to be fully examined in flies and C. elegans, it seems evident that genotoxic stress qualifies as a commonly shared stimulus that provokes this pathway in all three systems. Likewise, in flies, worms, and probably zebrafish, dmp53 is a downstream effector of CHK2, ATM, and possibly ATR (Figure 1). Therefore, at minimum, these kinases constitute an ancient signaling pathway fundamental to p53 regulation in response to DNA damage. In contrast, neither flies nor worms utilize an MDM2/MDMX-associated regulatory pathway, which is clearly present in zebrafish and essential in mammals. A scenario consistent with these findings proposes that MDM2/MDMX-mediated degradation of p53 appeared in the vertebrate lineage after the divergence of protosomes and deuterosomes (Figure 2). Alternatively, perhaps these regulators were de-emphasized among invertebrates, in favor of alternatives, such as iASPP18 (apoptotic-stimulating proteins of P53), a shared negative regulator of p53 in nematodes and mammals.

Figure 1
figure 1

Known components of the p53 regulatory network in non-mammalian models. Components in blue share conserved regulatory functions with mammalian counterparts

Full size image
Figure 2
figure 2

A parsimonious phyletic tree for the p53 gene family. The common ancestor to vertebrates and invertebrates is proposed to encode an ancient p53-like sequence together with a p63/p73-like paralog containing a SAM domain. This presumed paralog was lost in the ancestor shared by nematodes and flies but retained in mollusks

Full size image

If we apply the same logic to p53 effectors, p21 induction qualifies as a vertebrate (or perhaps exclusive mammalian) invention, since neither fly p21 nor C. elegans p21 appears to be engaged by their respective p53 counterparts. Conversely, BH3-only proteins emerge as strong candidates for ancient proapoptotic effectors since, like mammalian noxa and puma,28 the C. elegans BH3 proteins EGL-1 and CED-13 are p53 regulated and responsible for damage-induced cell death.14, 29 So do these sets of effectors fully explain tumor suppression by p53? The answer here is ‘probably not’ since, in mice, neither the p21− genotype30 nor the noxa−/puma− genotypes31 phenocopy the appearance of cancers in p53− strains. The search for additional proapoptotic effectors might be aided through studies in the Drosophila model since BH3-only proteins have not been seen here. Instead, inhibitors of apoptosis (IAP) antagonists (e.g. rpr, hid, skl) are pivotal effectors in flies and in mice, where comparable pathways similarly antagonize cIAPs.32 Profiling studies in this model have uncovered a limited scope of stimulus-dependent genes controlled by dmp53 and functional analyses on these may illuminate novel effectors relevant to human cancers.23, 27, 33

p53 in Development, Aging, and Disease

p53 activity is dispensable for normal animal development but can modify the aging process. In mice, a hyperactivated p53 mutant exhibits signs of premature aging and reduced longevity,34 prompting suggestions that accelerated aging is the ‘price paid’ for tumor suppression mediated by p53. However, in mdm2 hypomorphic mice, constitutively elevated p53 activity does not cause premature aging.35 If the first observation holds true, then loss-of-function p53 mutations should delay aging and extend lifespan. Testing this premise is not feasible in mice, since p53-deficient animals develop cancers and die prematurely. However, the Drosophila model does provide solid support for this idea since selective loss of p53 in neuronal tissue extended adult lifespan.36 The same genetic model was also applied in studies on mechanisms of neurodegneration in Huntington's disease,37, 38 and here p53 mediates the pathogenesis process in both flies and mice.39, 40 Together, these observations indicate that across the animal kingdom, conserved features of the p53 regulatory network are fundamentally linked to adaptive stress responses governing aging and health.

Evolutionary Lessons

p53 is structurally conserved across different phylums, including nematoda (worms), mollusca (clams), arthropoda (insects), and chordate (vertebrates) while the p63/p73 paralogs occur in the genomes of vertebrates and in mollusks, but not in insects or worms.41, 42 The DNA-binding domain, reflecting its activity as a transcription factor, is the most highly conserved region. Since no p53 ortholog occurs in yeast or plants, the evolution of p53 may have perhaps accompanied the appearance of metazoan animals. Attempts to reconstruct a p53 lineage infer different histories, depending on whether domain architecture (e.g. the presence or absence of SAM domain) is prioritized or whether primary sequence in the DNA-binding domain is used as the primary characteristic. Applying the former criteria together with knowledge of a p73 gene in clams, we can deduce that along with a p53 gene, the most recent common ancestor of vertebrates and invertebrates probably also encoded a SAM domain-containing paralog. However, if similarity scores and primary sequences in the DNA-binding domain are used, it is clear that invertebrate p53 genes share slightly more similarity to the DNA-binding domains of human p63/7343, 44 prompting suggestions that the ancestral origin of the p53 family were p63/73-like sequences preserved in mollusca (Figure 2). Despite this discrepancy, both methods deduce phyletic scenarios that include a p53-like gene (without a SAM domain) and a p63/p73-like gene (with a SAM domain) in the common ancestor of vertebrates and invertebrates.

What can we reasonably surmise about the primordial p53 gene and its regulatory network from these cross-species comparisons? Given what we know about p53 mutations in distinct models, the gene is probably nonessential, regulated by the ATM/CHK2 kinases and intimately coupled to proapoptotic target genes. Consistent with this, it is safe to assume that the p21-mediated checkpoint was a recent acquisition, specific perhaps to the vertebrate lineage as a p53-controlled effector pathway. If correct, a corollary to this scenario argues that the primordial elements of the p53 death program do not require a subroutine that first impacts cell cycle checkpoints, as some have proposed. From these hypothetical considerations, perhaps the most compelling and perplexing questions relate to the adaptive pressures that presumably selected for this gene in the first place. Although radiation is commonly used as a stimulant, the p53 regulatory network was obviously not ‘invented’ on the chance that tissues would find themselves exposed to acute radiation doses produced by medical imaging devices or nuclear weapons. And clearly, when we impose genotoxic stress in a laboratory setting, we may be ‘overchallenging’ cells in a way that is analogous to ‘hitting nails with a sledge hammer’. So, although it is clear that the network is engaged by DNA strand breaks and stalled replication forks, the proximal evolutionary pressures that shaped this network are still an enigma. In this regard, a possible contender could be exogenous or endogenous reactive oxygen species, which in mammals has been intimately linked to p5345, 46 and, more recently, implicated in the Drosophila network through a novel ‘mitochondrial checkpoint’.26

Future Challenges

p53 research in non-mammalian models will continue to illuminate important biological questions relevant to human health. One emerging question relates to recently discovered isoforms of p53 that arise from an internal promoter. This newly appreciated complexity embodies fundamental properties of p53 gene structure that are conserved from flies to man and are highly correlated with tumorogenesis.9 However, since it is not understood how (or whether) these variant isoforms encode distinct activities in vivo, we can expect that functional lessons drawn from simpler non-mammalian models should inform our knowledge of human p53. A second challenge relates to the growing appreciation for the nonautonomous properties that can impact p53 function, particularly as they relate to tumor microenvironments in situ. Understanding the p53 network beyond the single cell level is an essential task for describing the full dimension of tumor suppression by this oncoprotein. This deduction follows from the notorious fact that stress-induced behaviors of cells grown in cultured dishes can differ markedly from behaviors seen for cells in their native tissues or ‘niches’. Here, lessons drawn from simpler non-mammalian models should provide important insights as to how, for example, p53 action in one cell might influence the behavior of its neighbors, leading to emergent ‘tissue level properties’ that could otherwise not have been predicted. A third challenge relates to recently described mitochondrial pathways of p53 action, which involve apoptogenic mechanisms that are independent of transcription (reviewed in Erster and Moll47). Whether these noncanonical activities are conserved in non-mammalian models remains an open but important question. Finally, the simpler non-mammalian models should help us to resolve how p53 can be a fundamental determinant of so many pathogenic states – from tumor progression to neurodegeneration to aging and lifespan. Are these seemingly disparate pathologies intertwined through the same functional activity? Or might they result from distinct functional modalities propagated by this single factor? One way to rephrase this question at the molecular level might ask whether p53 status governs gene expression only in a stimulus-dependent manner. A possible alternative, perhaps, is that the p53 regulatory network might also impact basal gene expression in a way that explains these pathogenic phenotypes. We note here that recent array studies are certainly consistent with this latter possibility.27 Finally, considering that DNA damage response/checkpoint genes are well conserved from yeast to humans, p53 must have been endowed with additional properties specific to metazoan organisms that should ultimately be illuminated from future studies in these non-mammalian model systems.