The p53 family of transcription factors is made up of p53, p63 and p73, which share significant structural homology. In particular, transcriptional complexity and the expression of multiple protein isoforms are an emergent trait of all family members. p63 is the evolutionarily eldest member of the p53 family and the various isoforms have critical roles in the development of stratifying epithelia. Recent results have uncovered additional splice variants, adding to the complexity of the transcriptional architecture of p63. These observations and the emerging extensive interplay between p63 and p53 in development, proliferation and differentiation underline the importance of considering all isoforms and family members in studies of the function of p53 family members.
p63 and p73 were identified 10 years ago in a search of p53 homologues with tumour-suppressive functions. p63 was subsequently shown to be the oldest evolutionary member of the newly coined p53 family of transcription factors. Initial expectations were that p63 and p73 would constitute novel tumour suppressors; however, despite overlapping functions particularly in control of apoptosis , differences were rapidly apparent. Perhaps most significantly, the p63- and p73-knockout mice [2–5] exhibited more severe developmental defects in contrast with the p53-knockout mouse, which is developmentally normal, but prone to spontaneous or induced tumours . In particular two independently derived p63-knockout mice [2,3] were perinatally lethal, as the mice had severely compromised skin and died shortly after birth due to dehydration. This was shown to be a result of impaired development of stratified epithelia, which also results in dramatic craniofacial and limb abnormalities, mirrored in phenotypic spectrum of human syndromes caused by mutations of the p63 gene . However, despite the similarities in gross phenotype, conflicting results were generated with regard to the role of p63 in the epidermis and neoplasia. Extensive studies indicate that p63 is critical in the development, maintenance and differentiation of all stratifying epithelia [2,3,7]; however, its precise role is still unclear, probably due the increasing complexity of transcriptional regulation within the p53 family and the influences the family members and their isoforms exert on one another.
Transcriptional complexity is characteristic of p63 and the p53 family
The initial characterization of p63 [2,3] and p73 [4,5] indicated that multiple splice variants resulted in several protein isoforms, in contrast with the intensively investigated isoform of p53. This discrepancy has recently been reconciled by the identification of ΔN (N-terminal truncated isoform) p53 variants and revisiting of p53 isoforms that had been described previously, but largely ignored, confirming that all family members have extremely complex transcriptional architecture (reviewed in ).
The basic structure of the p53 family of transcription factors comprises a highly conserved core DBD (DNA-binding domain) present in all isoforms, which recognizes and binds to overlapping sets of transcription factor-binding sites in regulatory elements. There is an oligomerization domain immediately C-terminal to the DBD, which facilitates tetramerization and finally all three family members encode an N-terminal TA domain (transactivating domain). However, diversity is derived at the 5′-end, by utilization of alternative promoters, and at the 3′-end with splice variants .
For p63, six isoforms were initially described as the result of the combination of two alternative promoters and three 3′ splicing events at the 3′-end of the gene (α, β, γ) (Figure 1A). The TA isoforms contain an N-terminal TA domain (Figure 1B) with a high degree of homology to the TA domain of p53 and can bind to p53-responsive elements via their DBD to activate p53 target genes such as bax, p21 and mdm2 (murine double minute 2) . In addition, p63 can bind to its own discrete consensus DNA binding sequence to modulate transcription of a unique subset of genes [10,11]. The TA domain is omitted in the ΔN variants  (Figure 1A) and these isoforms can act as dominant-negative inhibitors of transcriptional activation by TA family members and full-length p53 and/or compete for binding sites [9,12]. Furthermore, the ΔN isoforms, including ΔNp63α, are capable of inducing a subset of target genes via a second TA domain present in the C-terminal (TA2) extended isoforms (Figure 1B) , which may also be influenced by the ΔN N-terminus . Transcriptional activating and repressive effects exerted by individual p63 isoforms are influenced by 3′ splicing, which in the longest α isoforms results in the inclusion of a SAM domain (sterile α motif domain) thought to be involved in protein–protein interactions  and a TID (transcriptional inhibitory domain) (Figure 1B). This latter domain can autoinhibit transcriptional activity of TA and ΔNp63α or inhibit heterocomplexes with other transactivating p63 or family member isoforms . The β isoforms are generated by exon-12 skipping, resulting in a truncated isoform, which omits the SAM domain and TID. The γ isoforms are truncated further due to splicing of an alternative final exon 10′, resulting in a short C-terminal peptide sequence of as yet unknown function. Intriguingly, two additional C-terminal variants, δ and ε, have recently been described, both of which result in truncation because of premature transcriptional termination in intron-10 in the former or exon-12/13 skipping in the latter  (Figure 1A). This is of particular interest as the ε transcript, which is the result of retention of part of an intron adjacent to exon 10, is identical with one of the two transcripts recently reported  to be abundantly expressed in the Brdm2 p63-knockout mouse . These two mRNAs were assumed to be artefacts of the knockout strategy, particularly since the second transcript was the result of exon-10 splicing into the non-functional end of the hprt mini gene, which was used to create the knockouts. Abundant expression of these transcripts was detected at the mRNA level and at protein level by an antibody directed to the DBD , indicating that the Brdm2 p63-knockout mouse expresses the recently described ε transcript  and effectively constitutes a knockout of the C-terminally extended isoforms only. In fact, the expression of these truncated transcripts may help to explain the subtly different epidermal phenotype observed between the Brdm2-knockout and the p63-knockout mouse [2,3], which led to the proposal of two contrasting models for p63 in the maintenance of proliferative capacity [2,18] or cell fate determination and differentiation [3,19]. Furthermore, the starkly different observations in studies to elucidate whether p63 had tumour-suppressive functions in compound p63+/− and p53+/− mice may also be explained in part by the expression of the ε isoforms in the Brdm2-derived mice [20,21].
Schematic representation of p63 splice variants and resulting protein isoforms
The role of p63 in epidermal morphogenesis
Intriguingly, the two knockout mice exhibited distinct skin phenotypes, in addition to gross developmental abnormalities, which led to the differing interpretations of the role of p63 in epidermal morphogenesis. In the p63-knockout mouse, a primordial ectodermal layer and infrequent patches of cells expressing markers of differentiation, loricrin, fillagrin and involucrin, were observed . They hypothesized that p63 was required for the maintenance of proliferative capacity of the basal stem cell compartment, suggesting that the absence of p63 led to proliferative rundown and a lack of mature epidermis and that p63 was not essential for expression of markers of differentiation. In contrast, in the Brdm2 p63-knockout mouse, a simple K8/18 (keratin 8/18) expressing epithelium was described that did not express markers of differentiation . The authors therefore deemed that p63 expression was necessary for commitment to a stratifying lineage. The recognition that the Brdm2 mouse expresses the p63ε isoforms may help reconcile these apparently disparate interpretations, particularly in the light of other recent studies. In fact, not only did Wolff et al.  find expression of the truncated p63ε isoforms in the epithelia of the Brdm2 mouse, but also they found rare patches of keratinizing epidermis overlying hair follicles and in anatomically protected niches in E18 (embryonic day 18) embryos. At E15, when commitment to terminal differentiation is occurring , an intact multilayered epidermis was observed and shown to express the p63 targets K14  and perp ; however, this is transient and fails to develop into a mature epidermis. It may be that expression of the TA and ΔNp63ε isoforms in the Brdm2 mouse is sufficient to allow development of a squamous stratifying epithelium, but, in the absence of the C-terminally extended isoforms, is unable to transition into a mature stratifying epithelia. In the total p63 knockout, in the absence of any p63 transcriptional activity or dominant-negative effects of ΔNp63, the proliferative capacity and development of stratifying epithelia appear to be more dramatically altered. What is clear, however, is that the ectodermal developmental abnormalities can be attributed to the C-terminal extended isoforms (α, β and now δ), which is unsurprising since a cluster of human mutations in the SAM domain causes similar phenotypes .
The complexity of the role of p63 and its isoforms in epidermal morphogenesis is still emerging. The ΔNp63 isoforms are expressed at log-fold higher levels than the TAp63 isoforms in development and differentiation, and expression of the TA isoforms at least in the mature epidermis is still a matter of debate. The predominant isoform of p63 expressed during development, the adult skin and the basal cells of all stratifying epithelia is ΔNp63α. Expression of ΔNp63α is down-regulated in suprabasal skin cells during the differentiation process by proteolysis and by targeting of the 3′-UTR (3′-untranslated region) by the microRNA mir-203 [26,27]. TAp63 isoforms are not detectable in keratinocytes by Western blotting, but have been shown, by a non-commercially available antibody, to be expressed in the upper layers of mature epidermis . Depletion of the TA isoforms in organotypic culture (as measured by real-time PCR) leads to a modest decrease in cornification in organotypic culture, without effecting proliferation .
In complementation mouse experiments, the expression of ΔNp63α in the p63-knockout mouse  under the control of the K5 promoter was able to induce expression of basal keratins K5/14 and markers of differentiation K1/loricrin . However, expression of TAp63α had no effect alone, although, in combination with ΔNp63α, resulted in a greater number of patches of skin with more complete expression of markers of differentiation. Despite this enhancement of differentiation, in both cases this was still insufficient for the mice to survive, possibly due to the use of K5 promoter to drive transgene expression as its activity would be low in p63-knockout cells.
Recent results indicate that ΔNp63α is both necessary and sufficient to induce commitment of K8/18-positive simple epithelia to K5/14-expressing stratifying lineage in in vitro differentiation of ES (embryonic stem) cells . ΔNp63α or ΔNp63β exogenous expression in vivo in murine lung epithelia also results in pleuristratified epithelia . Furthermore, both the K14  and the K5 promoters  were shown to be p63 responsive. Conversely, expression of K8/18 was observed in the Brdm2 mouse , and p63 depletion in organotypic cultures have been shown to induce expression of K8/18 , although K5/14 expression was maintained. The recent report indicating that expression of the ΔNp63β isoform can at least recapitulate some of the phenotypes observed on expression of ΔNp63α suggests that this is a result of the transcriptional activity of the ΔN isoforms and not the SAM domain or TID function of ΔNp63α .
In the mature epidermis, it is clear that p63 is critical for the maintenance of proliferative capacity and induction of differentiation. Hyperproliferation is due to aberrant activation of p53 target genes, such as p21, in the absence of repression mediated by ΔNp63α, and as such, this can be alleviated by simultaneous p53 knockdown . However, this is insufficient to rescue differentiation, indicating that p63 plays a positive role in induction of differentiation.
The ΔNp63α isoform has also been shown to directly repress and induce expression of transcriptional targets that are important for differentiation. In proliferating keratinocytes, it can bind to and suppress promoters such as p21 and 14-3-3σ, induction of which are important for differentiation to proceed . In contrast, it has also been shown to transactivate expression of the notch ligand Jag-1, IKK-α (inhibitory κB kinase-α) and perp [24,34,35] all of which are important for induction of normal epidermal differentiation. Taken together, these results indicate that ΔNp63α plays a dual role in the mature epidermis maintaining adhesion and proliferation in basal cells and inducing expression of genes required for differentiation to proceed. Furthermore, the balance of the p63 isoform expression and that of other family members is critical for normal epidermal morphogenesis. This finely tuned regulation is underlined by the complexity of control of p63 levels in the epidermis by proteolytic degradation and microRNAs.
Interplay between p63 and p53 in development and neoplasia
The high degree of homology of p63 to p53 and the resulting overlap in target gene selectively at least by the TAp63 isoforms make p63 an attractive target as a potential tumour suppressor. However, this is complicated by the expression of ΔN isoforms, which can oppose the action of TAp63, p73 and p53 in a dominant-negative manner or by transcriptional repression mediated by the TID of ΔNp63α and therefore would constitute an oncogene. These opposing actions of p63 N-terminal isoforms are supported by current data and underline the importance of considering specific p63 isoform expression in addition to that of p53 and p73 [29,39].
Initial characterization indicated that the p63 locus is infrequently mutated in cancer  and that increased expression of p63 frequently occurred in head and neck and other types of SCC (squamous cell carcinomas) (reviewed in ). These studies have been limited in their efforts to define specific isoform expression, but analyses in SCC indicate that ΔNp63 levels are increased (presumably α) , which may oppose TA activity of family members  and act in an oncogenic manner. In the bladder the TAp63 isoforms predominate over the ΔNp63; loss or reduction of TAp63 expression and/or overexpression of ΔNp63 has been detected in bladder carcinoma . This is consistent with TAp63 playing the role of a tumour suppressor and as such p63 loss has been shown to be associated with progression in bladder  and mammary carcinoma .
Further confusion was generated by the conflicting results obtained in experiments in mice, designed to investigate whether p63 has tumour-suppressive properties. Two laboratories compared the effect of p63 haploinsufficiency on tumorigenesis alone or in combination with p53 haploinsufficiency or knockout. In both laboratories, p63+/− mice exhibited squamous cell metaplasia, a premature aging phenotype and shortened life span [20,43]; however, their observations with regard to tumorigenesis were completely opposite. Flores et al.  reported that p63+/− haploinsufficient mice were predisposed to benign and malignant tumours, particularly those of epithelial origin, suggesting that p63 behaves as a tumour suppressor. In a separate study using the Brdm2 p63+/− mouse, haploinsufficiency was shown to, in fact, have a protective effect ; however, despite decreasing the number of spontaneous tumours in these mice, those that were observed were of stratifying epithelial origin. The two laboratories also crossed their p63+/− with p53+/−− mice to generate compound p63+/−;p53+/− mice. Strikingly, in both studies, p63 haploinsufficiency led to a marked shift in the pathological spectrum of tumours from lymphomas and sarcomas in p53+/− mice, toward tumours of epithelial origin. However, as with the p63+/− mice, Flores et al.  observed increased tumour burden and a dramatic increase in metastasis. In fact 50% of the tumours in these p63+/−;p53+/− mice were SCC, all of which exhibited loss of the wild-type p63 allele in LOH (loss of heterozygosity) studies, a hallmark of a tumour suppressor. In stark contrast, the study using the Brdm2 for their compound mutant mice reported that loss of one p63 allele had a protective effect when combined with p53 haploinsufficiency or loss and therefore p63 behaves in this instance as an oncogene . The contradictory results with regard to tumour suppression were initially attributed to the different genetic backgrounds used. However, these disparate observations may be reconciled by the recent discovery of the p63ε isoforms. The Brdm2 mouse used by Keyes et al.  expresses the C-terminal truncated TA and ΔNp63ε isoforms , which may be sufficient for p63 to perform its tumour-suppressive function, resulting in a net protective effect in the absence of the pro-proliferative and potentially oncogenic α isoforms. In this model, the shift in pathological spectrum and premature aging phenotypes could be attributed to loss of the ΔNp63α isoform, which is highly expressed in stratifying epithelia. If this is indeed the case, this may render these results much more informative and underlies the potential for particular isoforms of p63 to act as a tumour suppressor and an oncogene.
There is increasing evidence of the importance of the interplay between p63 and p53 in growth control development, differentiation and regulation of epithelia–mesenchymal interactions. ΔNp63 has been shown to inhibit mesoderm induction in Xenopus by opposing p53 activity, resulting in an ectodermal cell fate . Furthermore, p63 has been implicated in establishment of epithelial–mesenchymal interactions  and p63 knockdown increases motility and the mesenchymal phenotype . The role of p63 in maintaining normal epidermal differentiation is also highlighted by the observation that gain-of-function mutant p53 sequesters p63, leading to increased epithelia-to-mesenchymal transition and metastasis . Intriguingly, the phenotype of mouse models of the Li–Fraumeni syndrome, where gain-of-function p53 hot-spot mutants have been introduced into p53-knockout mice, mimics that of the p53+/−;p63+/− mice, exhibiting a shift in tumour type towards carcinomas and increased metastasis [48,49]. Furthermore, similarly to previous cell culture experiments , Lang et al.  demonstrated that this mutant p53 could bind to and inhibit the transcriptional activity of p63 in vivo. Taken together, these results indicate that there is extensive interplay between p63 and p53; in particular there is accumulating evidence that this may be crucial for control of the development and maintenance of the epithelial–mesenchymal boundary, disruption of which occurs frequently in cancer. Therefore careful investigation, taking into account potential effects of other isoforms and family members, is critical to expand our understanding of the full role of the p53 family members in neoplastic progression.
Given the complexity described above, it is not surprising that the roles of p63 in epidermal morphogenesis and neoplasia remain the subject of debate. It is, however, clear that p63 is a key regulator of epidermal development, proliferation and differentiation. In the light of recent evidence, some of the contradictions within the field may be reconciled, underlining the importance of careful consideration of not only the p63 isoforms, but also other family members, when undertaking studies of these complex genes. The interplay between p53 and p63 may also be very important and the gain of function observed with some p53 mutants may be due to changes in their interactions.
Gene Expression in Development and Disease: 13th Tenovus-Scotland Symposium, an Independent Meeting held at University of Glasgow, Glasgow, U.K., 16–17 April 2009. Organized and Edited by Sheila Graham (Glasgow, U.K.).
The work on p63 in our laboratory is funded by the Medical Research Council.