Discovery of a p53 variant that controls metastasis

The tumor suppressor p53 encoded by the tumor protein 53 (TP53) gene has been extensively studied since its discovery in 1979 (1, 2). The last four decades have seen the publication of an enormous number of reports on the role of p53 in suppressing cancer formation. These studies have demonstrated that TP53 mutations are the most common genetic alterations in human tumors (3).

In PNAS, an article by Senturk et al. in the laboratory of R. Sordella describes the discovery of a new p53 isoform, named p53ψ, that may contribute to, rather than prevent, tumorigenesis (4). The new p53 isoform described by Senturk et al. derives from the use of an alternative 3′ splice acceptor site within TP53 intron 6. p53ψ encodes a protein of 27 kDa that lacks the critical residues necessary for p53 oligomerization, DNA binding, and nuclear localization. As a result, p53ψ does not have transcriptional activity, does not form oligomers, and is localized principally in the cytoplasm.

Before the report of Senturk et al., 12 distinct p53 isoforms were known to exist in humans (5). Alternative splicing of TP53 intron 9 and use of the alternative promoter in intron 4 generates p53FL, p53β, p53γ, Δ133p53, Δ133p53β, and Δ133p53γ. Similarly, Δ40p53, Δ40p53β, and Δ40p53γ originate from the alternative splicing of TP53 intron 9 and the initiation of translation from a promoter in intron 2. Examination of mRNA levels of p53 variants has revealed that their generation is tissue specific, suggesting that internal promoter use and alternative splicing of TP53 are differentially regulated. Moreover, protein levels of p53 variants can be affected by external stimuli such as DNA damage induced by UV or ionizing radiation or changes in pH or oxygen concentration. With respect to function, p53 isoforms differ in their subcellular localization patterns and biological activities. Some p53 variants have been shown to interfere with the activity of p53 full length (FL) in vitro (6). In vivo, abnormal expression of p53 isoforms has been associated with disruption of the p53 FL response and, consequently, with accelerated aging, shorter life span, and promotion of tumor formation (7, 8).

The discovery of p53ψ raises the number of known p53 isoforms to 13. As is true for other p53 isoforms, this variant differs from p53 FL in its subcellular localization and putative function. However, unlike p53 isoforms that act as dominant-negative inhibitors of p53 FL activity, p53ψ does not interfere with p53 FL’s mode of action. In fact, the authors show that the down-regulation of p53 FL’s transcriptional targets in p53ψ-expressing cells is due solely to the generation of p53ψ transcripts at the expense of p53 FL mRNA.

Senturk et al. initially discovered p53ψ as a variant expressed in a specific cell subset isolated from the lungs of mice that had been treated with naphthalene, which rapidly induces injury and necrosis in lung tissue. This cell subset was characterized by high expression of the glycoprotein CD44 and low expression of the heat stable antigen CD24 (CD44^hiCD24^low). A subsequent experiment showed that p53ψ was also highly expressed in lung adenocarcinoma cancer cells with a CD44^hiCD24^low phenotype. Senturk et al. then demonstrated that p53ψ expression correlated with increased cell motility as well as enhanced expression of markers of epithelial-mesenchymal transition (EMT), including E-cadherin, vimentin, slug, twist, and zeb1. In human tumors, increased p53ψ levels correlated with a highly aggressive prometastatic phenotype and a decrease in the average disease-free and overall survival of patients bearing these malignancies. These results are in line with previous work demonstrating that the CD44^hiCD24^low cell surface signature is associated not only with stem cell/progenitor properties but also with tumor invasiveness and metastasis (9). On the basis of their results, Senturk et al. propose a mechanism to link p53ψ expression and EMT: (i) p53ψ localizes in the mitochondrial matrix via a mechanism dependent on the chaperone protein Tid1; (ii) mitochondrial p53ψ interacts with cyclophilin D (CypD) to activate the mitochondrial permeability transition pore, increasing intracellular reactive oxygen species (ROS); and (iii) increased intracellular ROS induces expression of EMT markers and bolsters invasive capacity. Rescue experiments conducted to explore this model confirmed that both ROS and CypD were required for the EMT marker expression and invasive capacity of p53ψ-expressing tumor cells.

This report by Senturk et al. further exposes the complexity of WT p53 regulation in cancer cells and highlights two crucial questions that remain to be addressed: what are the biological and physiological functions of p53ψ, and how is the alternative splicing leading to the generation of p53ψ regulated? Based on the results of this study, we can engage in some wide-ranging speculation on these issues.

The first fascinating aspect of this study was the finding that p53ψ levels are elevated in cells that express markers of “stemness” (such as CD44). These cells occur both in normal mouse tissues that sustain injury and in human tumors. These data imply that stemness may determine the generation of the p53ψ variant. Stemness is indeed a property associated with the expression of different splicing variants of critical factors (10), and p53ψ seems to be the p53 variant preferentially present in undifferentiated cells. Interestingly, loss of p53 promotes symmetric cell division in both normal stem cells and cancer stem cells, suppressing cancer formation in the latter case (11). This observation suggests the following potential scenario: (i) stemness-regulated expression of an unidentified splicing factor may promote the abnormal transcription of p53ψ at the expense of p53 FL; and (ii) this up-regulated p53ψ may then suppress asymmetric stem cell division and promote the expansion of cancer stem cells, favoring tumor growth. Senturk et al.’s finding that p53ψ was up-regulated on naphthalene-induced lung injury may be related to this point. Naphthalene causes double-strand DNA damage as measured by persistent H2AX staining, oxidative stress, and inflammation (12, 13). It is possible that one of these three factors is the key inducer of the alternative splicing generating p53ψ. Indeed, DNA damage and oxidative stress are potent activators of p53 signaling pathway, and the generation of p53ψ by the same stimuli may be a mechanism designed to balance p53 FL activity. Genotoxic damage has been found to trigger alternative splicing by affecting the expression or posttranslational modification of splicing factors in a way that alters their intracellular localization (14). It is therefore conceivable that p53ψ could counteract the apoptotic and cell cycle inhibitory functions of p53 FL and contribute to tissue remodeling after DNA damage. In this scenario, p53ψ would function as a driver of stem cell expansion in both cancers and nontransformed tissues that sustain injury.

The second provocative result of Senturk et al. is the link observed between p53ψ and EMT. p53 FL inhibits EMT (15–17), whereas p53ψ supports it. In addition to promoting the expression of EMT-associated markers such as E-cadherin, slug, snail, and twist, p53ψ may also influence the metabolism of metastatic cells. p53 FL suppresses glycolysis and promotes oxidative phosphorylation (18), and Senturk et al. observe that p53ψ is produced at the expense of p53 FL. Such a bias in favor of p53ψ could allow metastatic cells to use the Warburg effect at secondary sites of colonization, where glucose diffusion may be limited. In this situation, p53ψ generation would flip a metabolic switch that favors catabolic over anabolic reactions and thereby sustains the survival of metastatic cancer cells. This type of mechanism resembles the regulation of the M2 isoform of pyruvate kinase (PKM2), which is preferentially expressed in cancer cells and controls their metabolism. It is well established that PKM2 expression and activity are regulated by multiple factors, including the oncogene c-Myc, the hypoxia-inducible factor HIF-1a, the MAPK kinase ERK, the peroxisome proliferator-activated receptor gamma (PPRAG), and the serine-threonine kinases Akt and the mTOR (19). All these proteins are involved in signaling pathways that determine tumor survival and metabolism and are positively or negatively linked to p53. It will be interesting to investigate whether one or more of these factors regulates p53 activity by controlling p53ψ generation.

The third very interesting aspect of the study by Senturk et al. is that p53ψ is associated with ROS generation via the opening of the mitochondrial permeability transition pore. Both p53 FL and p53ψ regulate the activity of this pore by interacting with CypD (20) but the consequences for the cell are very different. When p53 activity opens the pore, the cell succumbs to oxidative stress-induced necrosis; when p53ψ activity opens the pore, the cell undergoes EMT. Thus, p53ψ expression once again counteracts the tumor-suppressive effects of p53 FL on the mitochondria. A puzzle arises here, however: if CypD can interact both with p53 FL and p53ψ, how is a cell’s fate determined? It may be that the ratio of p53 FL vs. p53ψ levels governs the formation of the p53-CypD complex and that this complex determines the amount of ROS produced. A high level of ROS would kill the cell, whereas a lower level would promote motility.

In conclusion, the work presented by Senturk et al. sheds light on a new p53 isoform that may be regulated by major signaling pathways and serves as a counterweight to p53 FL (Fig. 1). In doing so, p53ψ promotes tissue remodeling, the expansion and survival of malignant stem cells, and their metabolic adaptation and EMT. The ultimate proof establishing p53ψ’s physiological roles will necessarily emerge from the generation of two types of mouse models: a knock-in mouse strain that specifically expresses the p53ψ isoform (and not p53 FL) and a KO mouse strain that lacks p53ψ due to inactivation of the required specific splicing site. Once bred to the appropriate mouse tumor models, studies of these mutants should clearly reveal the biological roles of p53ψ in both the normal and cancer settings. On the basis of the results of Senturk et al., it can be speculated that increased levels of p53ψ will enhance tumor development and aggression by sustaining the expansion and survival of putative cancer-initiating cells. Whatever their outcome, such in vivo studies will be of great value in peeling back yet another layer of complexity in p53 regulation.

Fig. 1.

Schematic view of a model of p53ψ isoform regulation. The signaling pathways triggered by DNA damage, hypoxia, metabolic alterations, oncogene activation, and stress signals may regulate the activity, localization, or expression of an as yet unidentified splicing factor that controls the generation of p53ψ transcripts at the expense of full-length p53 FL transcription. Unlike p53 FL, p53ψ promotes cell survival, stem cell expansion, cell motility, and glycolysis.