TP53 Family Members and Human Cancers

p53 REVIEW ARTICLE TP53 Family Members and Human Cancers Jean Be´nard,1,2n Setha Douc-Rasy,2 and Jean-Charles Ahomadegbe1,3,4 1Unite´ de Ge´ne´tique Tumorale, Service de Ge´ne´tique, De´partement de Biologie Clinique, Institut Gustave Roussy, Villejuif, France; 2Centre National de la Recherche Scientifique UMR8126, Institut Gustave Roussy, Villejuif, France; 3UPRESS Pharmacologie et Nouveaux Traitements des Cancers, Institut Gustave Roussy, Villejuif, France; 4Faculte´ de Pharmacie, UPJV, Amiens, France For the p53 Special Issue Based on gene sequence homologies, a p53 (TP53) gene family become apparent with the addition of the most recently identified p63 (TP73L; formerly TP63) and p73 (TP73) genes to the already known p53. The p53 gene encodes for a unique protein eliciting well-known tumor suppressor gene (TSG) properties that mediate cellular response to DNA damage, e.g., cell cycle arrest or apoptosis. In contrast, both homologues specify an array of isoforms different in their N- and C-terminal domains. Transactivating isoforms, such as TAp63/p73, show TSG properties similar to p53, while isoforms lacking N-terminal transactivating domain such as DNp63/p73, induce a functional block against p53 as well as TAp63/p73 activities. Both p63/p73 types of isoforms are involved in development: p63 is critical for epithelial stem cell renewal and epithelial homeostasis, and p73 is involved in neurogenesis and natural immune response. These facts support interdependent functions for the p53 family members, which appear linked together in a complex and tight regulation network to fulfill cellular functions related to DNA damage and tissue homeostasis maintenance. The lack of p63/p73 mutations in human cancers rule out a typical TSG role for either of the p53 homologues. Nonetheless, p63 and p73 genes seem strongly involved in malignancy acquisition and maintenance process because of: 1) their tissue identities, and 2) their close interplay activities within the p53 family members, and primarily through the negative regulatory role played by DNp63/p73 isoforms for cell death control and differentiation. Hum Mutat 21:182–191, 2003. r 2003 Wiley-Liss, Inc. KEY WORDS: p53; TP53; p63; TP73L; p73; TP73; tumor suppressor; splicing; transactivation; apoptosis; differentiation; DNA damage; development; tumorigenesis; cancers DATABASES: TP53 – OMIM: 191170; GenBank: NM_000546 (mRNA) http://p53.curie.fr/ (p53 Web Site at Institut Curie) www.iarc.fr/P53/index.html (IARC p53 Mutation Database) TP73L – OMIM: 603273; GenBank: NM_003722 (mRNA) TP73 – OMIM: 601990; GenBank: NM_005427 (mRNA) INTRODUCTION The discovery in 1997 of the p73 gene (TP73; MIM# 601990) at the 1p36 locus, a location known to be subject to recurrent loss of heterozygosity in various human cancers [Kaghad et al., 1997], led to the critical issue of whether this first-identified p53 homologue functions as the archetypal tumor sup- pressor gene (TSG) p53 (TP53; MIM# 191170). One year later, the p53 gene family expanded with the cloning of a second homologue, the p63 gene (TP73L; MIM# 603273) [Yang et al., 1998], located at 3q27- ter. This chromosomal region is not subject to loss but to gain in cancers, and therefore suggests some oncogenic functions for p63. These apparently para- doxical findings caused head scratching and skepti- cism in an oncology research community in search of a role for p53 homologues in tumorigenesis. Moreover, in stark contrast to p53, which is mutated in about 50% of human cancers, it was soon revealed that p73 and p63 genes were not inactivated by mutations in a large number of cancers from various tissues. All together, these findings raised a hot question: Do p53 homologues elicit functions similar to p53? These issues have been in part elucidated for each p53 homologue from the evidence of an array of isoforms nCorrespondence to: Jean Be¤ nard, UniteŁ de GeŁ neŁ tiqueTumor- ale, Service de GeŁ neŁ tique, Departement de Biologie Clinique, Institut Gustave Roussy, Villejuif, France. DOI10.1002/humu.10172 Published online in Wiley InterScience (www.interscience.wiley. com). rr2003 WILEY-LISS, INC. HUMANMUTATION 21:182^191 (2003) expressed in tissues, some harboring transactivation domain (TAD), a hallmark for transcription factors with TSG activity, others lacking TAD, e.g., DN- isoforms, with antagonistic biological properties. There- fore, unlike their namesake, which stands as a canonical TSG, each p53 homologue can play either a TSG or an oncogenic role with respect to its specific isoforms. In this review we will first present what is currently known about the two genes and their role during developmental and adult states, both at the cell and tissue levels. Thereafter, we will examine their possible roles in pre-malignancies and cancers. COMPARATIVEORGANIZATION AND TRANSCRIPTION REGULATION IN THE TP53 FAMILY Akin to their p53 leader, p63 and p73 exhibit the three typical domains of a transcription factor across various species: namely an acidic, amino-terminal transactivation (TA) domain; a central core DNA- binding domain (DBD); and a carboxy-terminal oligomerization domain (OD) [Kaghad et al., 1997]. Both p53 ‘‘siblings’’ exert p53-like activities: They can bind to p53 DNA consensus target sites, transactivate p53 responsive genes, and induce cell growth arrest or apoptosis [Kaghad et al., 1997; Jost et al., 1997; Yang and McKeon, 2000]. The 20kb p53 is organized in 11 exons, whereas p63 and p73 genes are both over 60 kb and comprise 15 and 14 exons respectively. Unlike the unique and alternative, but infrequently used, splicing of the p53 gene, both p63 and p73 consistently give rise to an array of multiple protein isoforms due to differential mRNA splicing [Kaghad et al., 1997; De Laurenzi et al., 1998; Fillippovich et al., 2001] and to alternative promoter usage [Yang and McKeon, 2000; Pozniak et al., 2000]. Another specific feature of the p63 and p73 C-termini is the occurrence of a sterile alpha motif domain (SAM), which is involved in protein–protein interactions. Figure 1 depicts a comparison of the domain structure of p53 protein and major isoforms encoded by human p63 and p73. Historically, the C-terminal complexity was decrypted before the N-terminal 641 PXXP OD SAM BasicTAD DBD 1 59 142 321 353 397 p63α 1 63654 131 310 345 390 PXXP SAM BasicTAD DBD ODp73α PXXP NLS OD BasicDBD 1 39345 113 300 325 363 p53 22% 60% 37% 63%29% 38% 1 2 4 5 7 9 12 13 143 δ β ζ γ ε 6 8 10 11 TAD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 TAD DBD OD 3bis P1 P2 p73α ∆Np73α ∆exon2 (A) (B) (C) FIGURE 1. Comparative gene structures and functional organizationof the p53 familymembers. A: Foreach gene, the transac- tivation (TAD), proline-rich (PXXP), DNA binding (DBD), oligomerization (OD), sterile a motif (SAM), and post SAM basic domains are represented.B:Genomicorganizationof thep73 gene.Theproximal promoter (P1) yields b,c,d,e,f TAp73 isoforms while the distal promoter (P2) located in intron 3 gives rise to DN isoforms. C: In addition to the a isoform, COOH terminal splicing leads to b,c,d,e,fp73 isoforms for bothTAand DNp73molecules species. [Color ¢gurecanbeviewed in theonline issue at www.interscience.wiley.com.] TP53 FAMILY MEMBERS 183 complexity of the 2 gene loci. Most of alternative splicing occurs at the 30 end and involves more specifically exons 10 to 13, hence yielding transcripts that encode protein isoforms with various C-terminal structures. So far, among the high number of spliced p63 and p73 isoforms found at the RNA level, three of them are consistently found at the protein level, namely a (full structure of both genes), b (splicing of exon 13), and g (splicing of exons 10 to 12 for p73 and 15 for p63) [Yang et al., 2002]. In contrast, the p53 gene encodes one major transcript yielding a unique protein with transcriptional activity. Another striking finding regarding p63 and p73 gene expression regulation has been the occurrence of two different promoters, P1 and P2, that yield two distinct classes of proteins [Yang et al., 1998, 2000]. The P1 promoter leads isoforms showing TA domain with p53 homology. The P2 promoter, located within intron 3 and over 30-kb downstream, gives rise to N- terminal-truncated (DN) isoforms with biological properties opposite to those of p63/p73 TA isoforms, and lacking TA domain. The (DN) isoforms were identified first in mouse, then in human [Ishimoto et al., 2001]. Specific sequences of P1 and P2 bring about distinct biological properties. E2F-1 binding sites are present in both promoters, but p53-binding sites are present only in the P2 promoter [Yang et al., 2000]. Very recently, two crucial studies have indicated that: 1) TAp73 directly activates the transcription of endogenous DNp73 by binding to the two p73-specific target elements located on P2 [Nakagawa et al., 2002]; and 2) p53 induces DNp73 both at the mRNA and protein levels, as a result of a p53 direct activation of the P2 promoter [Kartasheva et al., 2002]. DNp73a-isoform readily associates with TAp73a/b and p53, as assessed by immunoprecipita- tion assay, and inhibits their transactivation activities. On the contrary, TAp63 shows only a marginal DN p73 transcription. Importantly, the negative feedback regulation of TAp73 and p53 by their DNp73 target provides for a novel autoregulatory system modulating cell survival and death. At the post-translational level, DN-isoforms are consistently identified in various tissues expressing p53-homologues, suggesting that DN-isoform elicits a protein of higher stability than the TA isoforms transiently expressed and subject to rapid degradation. Yet another remarkable distinctive property of the p53 homologues concerns their degradation process by MDM2. The turnover of p53, a short-lived protein, is regulated by ubiquitination through MDM2 binding, leading to p53 degradation by proteasome, and thereby limiting p53 accumulation [Haupt et al., 1997]. Similar to p53, the p73a and b proteins bind to MDM2 through their N-terminal but this interaction leads to transcription and apoptosis inactivation, and does not result in p73 degradation by proteasome [Balint et al., 1999; Zeng et al., 1999]. So far, p73 ubiquitination has not been demonstrated, but a p73a modified by a covalent linking with SUMO-1 (small ubiquitin-related modifier) was found to be more rapidly degraded by the proteasome than the un- modified p73 [Minty et al., 2000]. Unlike p53 and p73, p63 does not bind to MDM2 [Dohn et al., 2001], revealing yet another difference between p53 family members. Therefore one can distinguish two negative feed- back loops controlling p53 and TAp73, namely the mdm2 and DNp73 loops, keeping the cell death trigger under tight control. A RANGEOF P63 AND P73 ISOFORMSWITH DISTINCT BIOLOGICAL ACTIVITIES The combination of C-terminal diversity and acidic N-terminal regulation, including two distinct promo- ters, produces at least six major transcripts and subsequent protein isoforms (a,b,g) for each gene [De Laurenzi et al., 1998; Schmale and Bamberger, 1997]: TAp63/p73 p53-like isoforms and DNp63/p73 p53-antagonist isoforms (Fig. 1). This unusual gene regulation suggests a ‘‘two genes in one’’ concept for p63/p73 since TA and DN-isoforms act respectively as tumor suppressors and oncogenes [Stiewe et al., 2002; Stiewe and Pu¨tzer, 2002; Grob et al., 2002]. It is well known that p53 TSG properties arise from its binding to specific DNA sequences and from transactivation of target genes that specify cell cycle control proteins (p21, MDM2, GADD 45, 14-3-3-s, BTG2) and apoptosis (PIG3, Bax). Active p53 was also shown to play a key role in B lymphocyte, muscle, spermatogenesis, and neuronal differentiation [Rotter et al., 1994; Sidell and Koeffler, 1998] as well as in retinoic acid-mediated terminal differentiation of embryonic carcinoma cells [Curtin et al., 2001]. p53 is also involved in keratinocytic differentiation, although it does not determine this process because, in p53 knock-out mice, keratinocytic differentiation may be activated by Ca++ treatment. Another important difference between p53 family members concerns their interactions with viral onco- proteins. Indeed if E6 protein, SV40 large T antigen, and AdE1B 55 kd protein bind to and inactivate p53 during cell transformation, they do not interact with either p63 or p73 [Irwin and Kaelin, 2001]. When over-expressed in human cells, TA p63/p73 proteins also bind to p53 DNA target sequences, transactivate p53-responsive genes, and thereby induce cell cycle arrest, differentiation, and apoptosis in a p53-like manner [Kaghad et al., 1997; Jost et al., 1997; Yang et al., 1998] running through p53 most characteristic biological effects. However, transcrip- tional activity fluctuates with p63/p73 splice variants and must be considered. Indeed, TAp63/p73a iso- forms display a less active transcriptional activity than TAp63/p73b isoforms in p53 assays [Yang et al., 1998; 184 BEŁ NARD ETAL. Schmale and Bamberger, 1997] indicative of a negative regulatory effect of the a isoforms from the sterile a motif (SAM) (Fig. 1) [Thanos and Bowie, 1999]. Similarly the COOH-terminal region located in the post SAM domain can have an auto-inhibitory effect on the TAp63/p73a transactivation [Ozaki et al., 1999]. Noteworthy, the above-described biological activa- tion has been demonstrated using forced expression of TA isoforms of the p53 homologues, likely to mask differences between the various p53 -family members. Not to mention that, so far, no binding of endogenous p63/p73 to their putative target gene promoters has been demonstrated in vivo by chromatin precipitation. Nonetheless, differences may exist between human p53 family members regarding their potency to induce activation of p53 target genes. Indeed, if p53 leads to strong p21 and MDM2 induction, p73 was shown to induce a major and strong 14-3-3 s up-regulation [Zhu et al., 1998]. In response to various stresses resulting in DNA damage, p53 is activated by phosphorylation to perform its cell-cycle arrest and apoptotic tasks. Similarly, in response to DNA-damaging agents such as cis-platin and g-irradiation, p73 up-regulation results in apoptotic response albeit using a pathway distinct from that of p53. p73 is phosphorylated by c- abl, the non-tyrosine kinase receptor, which is itself phosphorylated by the ATM (ataxia telangiectasia- mutated) protein [Gong et al., 1999; Shaul, 2000]. Therefore, a p73 repair pathway for DNA damage exists, which is p53-independent (Fig. 2). So far regarding p63 response to DNA damage, important information has been gained from UV- treated human keratinocytes: TAp63 isoforms are up- regulated while DNp63 isoforms are dramatically down-regulated in these p63 expressing cells [Liefer et al., 2000]. As a matter of fact, the DNp63 down regulation parallels p53 stabilization, which, in turn, acts as the apoptotic executor of UV-damaged keratinocytes. These findings provide evidence for a full interplay between the two main antagonistic p63 isoforms and p53 for DNA-damage response (Fig. 2). In p53-deficient SaoS-2 cells, oncogenes such as E2F1, c-myc, and E1A can induce and activate endogenous p73 protein expression and drive cells to apoptosis, pointing to a p73 upstream pathway that is p53-independent [Zaika et al., 1999]. Importantly, a very recent work shows that, in response to various stresses (drug, g-irradiation), p53 requires TAp73, as well as p63, to activate promoters of apoptotic genes such as NOXA, PERP, and BAX [Flores et al., 2002], supporting a crucial role for the two homologues in p53-mediated apoptotic activity. To sum up, a large body of evidence favors specific upstream and downstream target gene regulation for DNA damage ATM c-Abl ∆Np73 ∆Np63 ATR, Casein KII p53 mutant Adenovirus E4orfE6 ATM DNA-PK DNA-PKCisplatin Growth arrest Apoptosis Neuronal development/ differentiation Epithelial development/ differentiation ? P300 PCAF Mismatch repair Morphogenesis, Neurogenesis, p63 p73 p53 UV IR Stress UV ? FIGURE 2. Schematic p53 family members pathways. Besides speci¢c developmental and physiological functions, p63 and p73 participate to p53 genomic guardian function. Upon genotoxic stresses induced by ultraviolet (UV), g irradiation (IR) or cisplatin, the two homologues interplay with p53 to achieve growth arrest and apoptosis functions. It is now established that p53 as well asTAp73 induce a direct activation of DN-p73 creating thus a feedback loop to control negatively these functions. [Color ¢gure can be viewed in the online issue at www.interscience.wiley.com.] TP53 FAMILY MEMBERS 185 each p53 family member, and for their respective isoforms, playing a role in non-redundant functions. Undoubtedly, RNA interference strategies specifically designed against these isoforms will provide definitive clues with regard to biological activities of each p53- family member. THE TP53 HOMOLOGUESASDEVELOPMENT PROTAGONISTS The respective deficient mice have provided further insights concerning the physiological role of p53, p73, and p63. p53 null-mice develop spontaneous tumors at high frequency without exhibiting a specific phenotype, in particular during their development [Donehower et al., 1992]. If active p53 does not appear to be involved in physiological apoptosis occurring throughout embryogenesis, it becomes crucial for DNA-damage-induced apoptosis [Lowe et al., 1993]. In stark contrast, both p73 and p63 null- mice show specific developmental defects but no spontaneous tumors at all [Yang et al., 1999, 2000]. Studies with p63-null mice have revealed the occurrence of TA and DN isoforms in the stem cell compartment of stratified epithelia, thereby also offering a very useful guide for our understanding of p53-homologues regulation and functions [Yang et al., 1998, 1999]. p63 knockout-mice are born alive but display severe deformations of limbs as well as of epithelia including skin, breast, urothelia, and prostate [Mills et al., 1999; Yang et al., 1999]. The p63 –/– mouse skin does not undergo normal development because it is lacking stratification and differentiation markers. These mice also lack mammary glands, hair follicles, and teeth. Such murine phenotypic disorders were confirmed by p63 mutations in various human syndromes involving limb and ectodermal develop- ment. Indeed p63 mutations have been related to EEC syndrome, a set of autosomal dominant disorders (ectrodactyly, ectodermal dysplasia, and cleft lip/ palate), as well as to sporadic split, hand-split foot malformations [Yang et al., 1999; Celli et al., 1999; van Bokhoven et al., 2001]. In fact, epithelial loss in p63-deficient mice reflects the inability of the multi- layered regenerative epithelia stem cells to undergo asymmetric division. Indeed, in contrast to the physiological asymmetrical division (one stem cell providing one stem cell to repopulate the stem compartment, and one basal cell for differentiation), all p63 –/– basal cells undergo terminal differentiation, leading to a complete depletion of the stem cell compartment. In mice [Yang et al., 1998] and human [Faridoni-Laurens et al., 2001] epithelia, p63 is highly expressed in stem cells. Neuronal defects have been the first phenotypic traits observed in p73-deficient mice, including high intracranial pressure and hippocampus dysgenesis [Yang et al., 2000]. Importantly, it was shown that DNp73 isoform is predominantly expressed in murine developing brain and sympathetic neurons, hence explaining neuronal defects from lack of DNp73. As a matter of fact, DNp73 isoform was shown to inhibit neuronal apoptosis from NGF withdrawal by blocking the p53 pro-apoptotic function [Pozniak et al., 2000]. These seminal findings designate p73 and DNp73 isoforms as determinants of cellular differentiation and apoptosis in neuronal tissues. Moreover, in devel