Treasure or artifact: a decade of p63 research speaks for itself

Dear Editor,

The p63 gene encodes six transcription factors, which are generated by the use of two promoters, giving rise to TA and ΔN N-termini, and alternative splicing, giving rise to three C-termini, termed α, β, and γ.¹ p63 is expressed primarily in stratified epithelia, including the epidermis, as well as in epithelial appendages.² To investigate the role of p63 in these tissues, several p63 knockout mice have been generated.^{3, 4, 5, 6} Of these, p63^Brdm2/Brdm2 mice, generated by the Bradley laboratory,⁶ have been widely used by numerous research groups. These groups have consistently reported that p63^Brdm2/Brdm2 mice fail to develop an epidermis, internal epithelia, and epithelial appendages.^{7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17} Wolff et al.¹⁸ have recently re-evaluated the p63^Brdm2/Brdm2 mice and reported that early stages of epidermal and hair follicle morphogenesis occur in these mice. Further, Wolff et al. assert that one or more truncated p63 proteins are expressed from the p63^Brdm2 allele, leading them to conclude that these truncated p63 proteins are sufficient to initiate the early stages of epidermal morphogenesis.

The work by Wolff et al. contradicts a large body of literature by several independent research groups in which the developmental phenotype of p63^Brdm2/Brdm2 mice has been extensively characterized. Most strikingly, it has been well-documented that the epidermis fails to develop in p63^Brdm2/Brdm2 mice.⁶ This failure to develop an epidermis was found to result from an inability of the surface ectoderm, the single-layered epithelium that initially covers the developing embryo, to commit to an epidermal lineage.¹⁰ Thus, the surface epithelium of p63^Brdm2/Brdm2 mice remains single-layered throughout gestation. Consistent with these findings, the p63^Brdm2/Brdm2 surface epithelium expresses keratins K8 and K18, structural proteins that are normally expressed in the surface ectoderm before the commitment to the epidermal lineage¹⁰ (Figure 1). In contrast, the surface epithelium of p63^Brdm2/Brdm2 mice does not express markers of epidermal development and differentiation, including K14 and Perp, at any developmental stage^{9, 10} (Figure 1). As a result of the failure to develop an epidermis, p63^Brdm2/Brdm2 mice do not develop an epidermal barrier and die shortly after birth owing to excessive water loss.⁶ In addition to the epidermis, structures of which the development relies on reciprocal signaling between the epithelium and the underlying mesenchyme, such as teeth and hair follicles, fail to develop in p63^Brdm2/Brdm2 mice.^{6, 7, 11} Moreover, the finding that hair follicle and dental placodes do not form in p63^Brdm2/Brdm2 mice shows that appendage development does not initiate in p63^Brdm2/Brdm2 mice.⁷ Finally, internal epithelia, including the bladder,¹⁵ prostate,¹³ cervicovaginal epithelia,^{8, 12} esophagus,¹⁴ and testis,¹⁷ also fail to develop normally in p63^Brdm2/Brdm2 mice.

Figure 1

Skin phenotype of p63^Brdm2/Brdm2 mice. Immunofluorescence analysis using antibodies against K14 (green), a marker for epidermal keratinocytes, and K8/K18 (red), markers for single-layered epithelia. Both at E13.5 (a) and E16.5 (b), the p63^Brdm2/Brdm2 epidermis expresses K8/K18, but not K14, indicating that the surface epithelium has not adopted an epidermal fate. In contrast, the epidermis from control littermates expresses K14, but not K8/K18. The K8/K18-expressing cells on the surface of the E16.5 control epidermis represent cells of the periderm, which is a transient layer of cells that may protect the underlying epidermis

Using the same p63^Brdm2/Brdm2 mice as in the above-described literature, Wolff et al. describe strikingly different phenotypes.¹⁸ In sharp contrast to previous studies, Wolff et al. report that, except for limb morphogenesis, embryonic development proceeds essentially normally in p63^Brdm2/Brdm2 mice until E15. At this developmental stage, the authors did not observe a marked difference between p63^Brdm2/Brdm2 skin and wild-type skin. Instead, they observed that, like in control skin, p63^Brdm2/Brdm2 epidermis was multilayered and that hair follicle buds were present. In addition, they found that p63^Brdm2/Brdm2 epidermis expressed K14 and Perp, further suggesting that the epidermis is normal. Even though Wolff et al. report that p63^Brdm2/Brdm2 skin is normal at E15, only patches of normal skin were observed in E18 p63^Brdm2/Brdm2 embryos. Unfortunately, intermediate developmental stages were not evaluated, and thus the reason for the apparent disintegration of the skin remains unclear. The authors attribute the normal development of the epidermis, hair follicles, and internal epithelia until E15 to their finding that one, or perhaps two, truncated p63 proteins are expressed from the p63^Brdm2 allele. The western blot analysis performed by Wolff et al. fails to convincingly demonstrate that such truncated proteins are actually expressed in p63^Brdm2/Brdm2 mice. Further, we have performed extensive western blot analyses on embryonic p63^Brdm2/Brdm2 skin samples and have never observed a band corresponding to a truncated p63 protein (Figure 2).⁷ However, even if truncated p63 proteins are expressed from the p63^Brdm2 allele, they would not correspond to endogenous p63 isoforms. Whereas the N-termini of the presumed truncated proteins are identical to those of endogenously expressed p63 proteins, the C-termini lack the unique exons for α, β, or γ isoforms. Although the authors argue that these truncated proteins functionally resemble TAp63γ and ΔNp63γ, this is not convincingly demonstrated. Thus, the conclusion that these truncated p63 proteins, if they exist, can faithfully regulate epidermal and hair follicle morphogenesis is not supported by the data.

Figure 2

Embryonic p63^Brdm2/Brdm2 skin does not express any p63γ-like proteins. Western blot analyses on protein extracts isolated from the skin of E13 and E14 p63^Brdm2/Brdm2 and control littermates. Note the absence of a fast-migrating band corresponding to truncated p63 proteins in the p63^Brdm2/Brdm2 samples. The molecular weight of the p63γ-like protein described by Wolff et al. is between 36.5 and 42.7 kDa. NS, non-specific bands

To reconcile the differences in observed phenotypes reported by Wolff et al. and other groups, it is important to bear in mind that the p63^Brdm2 allele was generated by insertional mutagenesis, resulting in duplication of a segment of the p63 gene.⁶ Follow-up studies have consistently shown that p63^Brdm2/Brdm2 mice do not express detectable levels of p63 protein, thus demonstrating that the observed phenotypes are caused by a complete loss of p63 expression.^{7, 8, 13, 14} However, because of the partial duplication of the p63 gene, reversion events in which the wild-type p63 allele is re-created through spontaneous homologous recombination occur sporadically in these mice (Figure 3).¹⁹ In fact, we routinely observe such reversion events in p63^Brdm2/Brdm2 embryos of all developmental stages. Although these patches are generally rare and small in size, on some occasions, they are larger and easily discernable by eye (Figure 4a). As expected, cells within these patches show normal epidermal differentiation, as demonstrated by histological analysis as well as by the analysis of expression of markers of epidermal differentiation (Figure 4b–d and data not shown).

Figure 3

Reversion of the p63^Brdm2 allele into a wild-type p63 allele. (a) Structure of the wild-type p63 allele. Light gray boxes indicate exons. (b) Structure of the p63^Brdm2 allele. The p63^Brdm2 allele was generated by insertional mutagenesis, resulting in a duplication of exons 5–10 (gray) and insertion of a selection cassette (dark gray). (c) Spontaneous homologous recombination at the p63^Brdm2 locus can occur, resulting in (d) restoration of a wild-type p63 allele. The restored p63 allele can contain exons 5–10 from the original p63 allele (light gray), or exons 5–10 derived from the targeting vector (gray)

Figure 4

Reversion events in p63^Brdm2/Brdm2 mice. (a) Patches of normal-looking skin on an E18.5 p63^Brdm2/Brdm2 embryo. Asterisks indicate reversion events, where re-expression of p63 has resulted in normal epidermal development. (b) Histological analysis of a reversion event that occurred in an E18.5 p63^Brdm2/Brdm2 embryo. The epidermis on the left side is single-layered, and represents epidermis where p63 is not expressed (c). The epidermis on the right side is stratified, and represents a reversion event where p63 is re-expressed (d). Arrow in (b) indicates transition between single-layered and stratified epidermis. The images in (c) and (d) were taken from sections of the same embryo stained with antibodies against p63 (green) and either K8K18 (c, red) or K14 (d, red). Images in (a) and (b) provided by Dr. Alea A Mills

Wolff et al. attempt to exclude the possibility that the normal-appearing skin they observe in p63^Brdm2/Brdm2 mice is a result of spontaneous reversion events by analyzing p63 transcripts in p63^Brdm2/Brdm2 embryos. Although they were unable to detect transcripts representing the α, β, and γ C-termini of p63, the analysis was performed on mRNA isolated from whole embryos, rather than on mRNA isolated from microdissected areas of normal-appearing skin. Therefore, any wild-type p63 transcripts, expressed from a reverted allele, would have been easy to miss in this analysis. In fact, this seems to be the most likely explanation for these observations, especially considering that reversion events are known to occur in p63^Brdm2/Brdm2 mice (Figure 3). In addition to reversion events, other types of novel genetic changes could have occurred in the p63^Brdm2/Brdm2 mice, which may account for the phenotypic differences that were observed by Wolff et al. (reviewed by Aberdam and Mantovani²⁰).

In summary, the phenotypic analysis of p63^Brdm2/Brdm2 mice presented by Wolff et al.¹⁸ is inconsistent with the extensive documentation of the p63^Brdm2/Brdm2 phenotype by several independent research groups.^{7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17} Whether this is caused by an increase in reversion events in the p63^Brdm2/Brdm2 mice used by Wolff et al. remains to be determined. However, as Wolff et al. report extended areas of the normal epidermis in the p63^Brdm2/Brdm2 mice they used, it is most likely that the mice analyzed by Wolff et al. are genetically not identical to those generated by Mills et al. Thus, the suggestion by Wolff et al. that all previous work involving p63^Brdm2/Brdm2 mice needs to be re-interpreted is not warranted.

Conflict of interest

The authors declare no conflict of interest.