ΔNp63 Is Essential for Epidermal Commitment of Embryonic Stem Cells

In vivo studies have demonstrated that p63 plays complex and pivotal roles in pluristratified squamous epithelial development, but its precise function and the nature of the isoform involved remain controversial. Here, we investigate the role of p63 in epithelial differentiation, using an in vitro ES cell model that mimics the early embryonic steps of epidermal development. We show that the ΔNp63 isoform is activated soon after treatment with BMP-4, a morphogen required to commit differentiating ES cells from a neuroectodermal to an ectodermal cell fate. ΔNp63 gene expression remains high during epithelial development. P63 loss of function drastically prevents ectodermal cells to commit to the K5/K14-positive stratified epithelial pathway while gain of function experiments show that ΔNp63 allows this commitment. Interestingly, other epithelial cell fates are not affected, allowing the production of K5/K18-positive epithelial cells. Therefore, our results demonstrate that ΔNp63 may be dispensable for some epithelial differentiation, but is necessary for the commitment of ES cells into K5/K14-positive squamous stratified epithelial cells.


Introduction
During embryonic development, the body surface is first covered by a single layer of basal epithelial cells. As development proceeds, this primitive epithelium switches on a differentiation program leading to stratification: the ectodermal cytokeratins K8/ K18-positive cells of the primitive epithelium are replaced by basal epidermal K5/K14-positive cells, and progressively, several layers of suprabasal differentiated cells appear offering an efficient resistance against environmental stress. In the adult skin epidermis, the stratified squamous epithelium maintains a basal layer of proliferating cells that gives rise to multiple layers of terminally differentiating cells that continually reach the body surface from where they are shed. Epidermal stem cells located interfollicularly and within the bulge region of the hair follicles retain a selfrenewing capacity, critical to preserve the tissue integrity during the life of the organism. The molecular mechanisms inducing the stratification program and those regulating the self-renewing process still remain to be clearly defined.
The p63 transcription factor, a homologue of the tumour suppressor gene p53 [1], is characterized by six different isoforms under the control of two distinct promoters and three different alternative splicings leading to the production of DNor TA-p63, and alpha, beta or gamma isoforms, respectively. Recently p63 has been implicated in both the development and the maintenance of stratified epithelia [2][3][4]. In loss-of-function models generated by two different groups, mice lacking p63 have been shown to be severely compromised in their ability to generate an epidermis, many other stratified epithelia as well as the limbs [2,3]. Recent reports convincingly demonstrated the implication of p63 in epithelial stem cell renewal potential, as the absence of p63 was shown to induce a loss of proliferative potential of epidermal [5,6], corneal [7] or thymic epithelial cells [6,8]. On the other hand, experiments to complement the absence of p63 during development were designed by crossing p63-null mice with transgenic mice expressing different isoforms of p63 under the control of a K5-promoter [9]. Those mice expressing TAp63 and DNp63 developed patches of epidermal differentiated cells, but the overall rescue of the epidermal phenotype remained partial, suggesting an earlier role of p63 during epidermal morphogenesis, before the K5-promoter is activated. Accordingly, Koster et. al showed that the exogenous expression of p63 in single-layered lung K8/K18positive epithelial cells was able to initiate a stratification program, with formation of K14-positive cells [10], suggesting that p63 is required for the initiation of epidermal stratification, a process mediated in part by the direct induction of AP-2c expression [11]. However, the precise role of p63 in the morphogenesis of epidermal cells remains to be defined.
Embryonic stem (ES) cells, derived from blastocyst-stage early mammalian embryos, are pluripotent and can initiate lineagespecific differentiation programs of many tissues and cell types in vitro. ES cells are thus an effective tool in recapitulating in vitro the main steps of embryonic development. We have recently reported the efficient derivation of ectodermal and K5/K14-positive epidermal cells from murine ES cells [12] and clarified the molecular mechanism of BMP-4 in the neuroectodermal fate [13]. In the present study, we show that the DNp63 isoform is expressed soon after BMP-4 treatment, along with the simultaneous expression of direct or indirect p63-target genes associated with epidermal morphogenesis. Furthermore, p63 knockdown drastically inhibits the progression of ES cells along the pluristratified epithelial fate, while de novo expression of p63 is compulsory for pluristratified epithelial commitment once the ectodermal stage has been reached. Finally, ectopic expression of DNp63 in ESderived ectodermal cells is able to induce their commiment. Therefore, our results demonstrate that DNp63 is absolutely necessary at the ectodermal stage for the proper commitment of ES cells into K5/K14-positive epithelial cells, and therefore suggest that DNp63 function is required for the embryonic development of pluristratified squamous epithelia.

Results
DNp63 is expressed during epithelial differentiation of mouse ES cells We have previously established a cellular model that recapitulates in vitro, the main steps of embryonic neuroectodermal fate (Fig. S1). ES cells seeded on a layer of fixed-NIH-3T3 cells in a serum free medium differentiate into neuroectodermal precursors, while early mesodermal and endodermal commitments are inhibited [13,14]. These neuroectodermal precursors have been shown to further differentiate in 7 to 10 days into neural cell populations [14,15]. When treated with BMP-4 from day 3 to 5, the neural differentiation process is efficiently inhibited, partially through the apoptosis of Sox-1 + neural precursor cells, as we recently demonstrated [13], allowing the remaining committed cells to differentiate into K18-positive ectodermal cells, which are the putative precursors of epithelial cells. Addition of serum at day 9 allowed keratinocyte differentiation to proceed further [13]. Epidermal differentiation proceeds through successive steps that lead proliferative keratinocytes, positive for cytokeratins K5 and K14, to undergo terminal differentiation with the appearance of specific markers like K1, K10 and involucrin.
During ES cell differentiation, K5/K14-positive epithelial cells were detected by day 10 /11 and further proliferated into large patches by day 14 (Fig. 1-A panels a-c). FACS analysis indicated that, among the differentiated cells, about 10% became epidermal cells at day 14 ( Fig. 1-B). The terminal differentiation events occured as K10-positive cells as well as involucrin-positive cells were also detected by immunofluorescence analysis (Fig. 1-A, panels d,e). These cells were in close contact to patches of K5/K14-positive cells ( Fig. 1-A, panels d,e) suggesting that the stratification program was efficiently initiated during ES differentiation. Confocal analysis clearly confirmed that the differentiated K10-positive cells were different from the proliferating K14-positive cells ( Fig. 1-A, panel f). Real-time Q-PCR (qRT-PCR) analysis was performed at different time points: day 5 (48 h after BMP-4 treatment), day 8 (24 h before serum addition) and day 14 (end of the culture) ( Fig. 1-C). Transcription of the K18, K14 and K5 genes was induced after BMP-4 treatment (days 5 and 8). Following serum addition (day 14), K18 gene expression was reduced while transcription of K5 and K14 genes were enhanced ( Fig. 1-C, panel a). A detailed kinetic experiment for K18 and K14 gene expression confirmed the progression along the epithelial developmental process (Fig. S2-A). The transcription of the three genes encoding laminin-5, the major adhesion ligand of keratinocytes, was also clearly induced during this epithelial differentiation process ( Fig. 1-C, panel b). At day 14, specific markers of terminal differentiation of stratified epithelia (K1, K10, involucrin and filaggrin) were also detected ( Fig. 1-C, panel c). These results indicate that this ES cell model recapitulates in vitro the stepwise appearance of stratified epidermal cells and is therefore suitable to study the function of p63 isoforms during embryonic epithelial cell differentiation.
DNp63 transcripts were detected after BMP-4 addition at day 5 and remained high during ES differentiation (Fig. 2 Fig. S2-B). TAp63 was never detected during epidermal commitment. By Western blot, DNp63 was detectable at days 9, 11 and 14 ( Fig. 2-B and Fig. S2-C). In accordance with the qRT-PCR analysis, TAp63 protein was never detected during ES differentiation. Several non specific bands, which are characteristic of the pan-p63 antibody used in this study, were observed but their migration rates did not correspond to any p63 isoform (Fig. 2-B and Fig. S2-C).

-A and
Of note, DNp63 isoform was detected by Western blot analysis much later (day 9) than its transcript (day 5), probably due to the heterogeneity of the cell population and the difference of sensitivity of the two methods. Therefore, we evaluated p63 detection at the cellular level by immunofluorescence staining. In the absence of BMP-4, no DNp63 expression could be observed at any time point.
In the presence of BMP-4, nuclear-stained DNp63-positive cells became detectable among K18-positive ectodermal cells by immunofluorescence staining at day 7 ( were not yet committed into the keratinocyte lineage, or that part of them were already committed to alternative epithelial fates. Finally, we were not able to detect TAp63-positive cells using a specific antibody, confirming the absence of TAp63 transcripts (data not shown).

DNp63 expression is required for the commitment of mouse ES cells into keratinocytes
To evaluate whether DNp63 was required for stratified epithelial commitment, stable ES cell clones that express a small hairpin (sh) RNA inhibiting p63 expression were produced. As shown in Fig. 3A, two independent ES cell clones (sh-p63-cl1 and sh-p63-cl2) displayed, by Western-blot analysis, a strong inhibition of exogenously expressed DNp63, which was not observed with the sh-control clone. Upon differentiation with BMP-4 and serum, the sh-p63 clones produced, at day 14, a number of ectodermal K18positive cells similar to the wild type control cells ( Fig. S3-A, B). It confirmed that the ectodermal differentiation of ES cells was independent of the expression of DNp63. However, while large patches of K14-positive cells were observed within the sh-control clone ( Fig. 3-B), very few isolated K14-positive cells were detected with sh-p63-cl1 and sh-p63-cl2 clones at day 14. Analysis by FACS confirmed that the absence of DNp63 resulted in a strong reduction of K14 and K5-expressing cells at day 14 ( Fig. 3-C). At the transcriptional level, we further confirmed the strong inhibition of the K14 mRNA level in differentiating sh-p63-ES clones ( Fig. 3-D). Accordingly, lama3 gene expression, as well as terminal differentiation markers such as K1, K10, involucrin and filaggrin were completely repressed when p63 was knocked down (Fig. 3E). The two other genes encoding for laminin-5 (lamb3 and lamc2) were repressed as well in sh-p63 ES cells (data not shown). Interestingly, while K14 expression was drastically reduced, the expression of K5 mRNA was not modulated by the absence of DNp63 and a significant percentage of K5-positive cells was still observed by FACS analysis (Fig. 3-C) with the two sh-p63-ES clones (5.6% and 5.7%, respectively), as compared to sh-control and wt-ES cells     (Fig. 3-B). Actually, this particular structure was also detectable within wt and sh-control ES cell differentiation along with the large K14/K5-positive patches. Since they were detected as early as day 11 and still present at the end of the culture (up to day 17), they may represent alternative epithelial cells.
Activation of p63-regulated or-dependent genes during epidermal differentiation Recently, several genes dependent on p63 expression and associated with the murine early embryonic epidermal morpho-genesis have been identified [16][17][18]. Among them, BMP-7, FGFR2b, Notch1 and Jagged1 transcripts have been shown to be absent from the developing epidermis of p63-deficient mice [18]. IKKa and GATA3 are two other genes regulated by p63 and associated with epidermal development [16,19,20]. Finally, Perp, a protein promoting the stable assembly of desmosomal adhesive complexes, was one of the first described p63 regulated proteins implicated in epidermal development [17]. We thus monitored their expression during ES epidermal differentiation as compared to undifferentiated ES cells. Upon BMP-4 treatment, in parallel to the induction of DNp63 gene expression, all the transcripts were detected by qRT-PCR (Fig. 4A). The relatively low activation of some of them (2-5 fold) as compared to the very pronounced induction of DNp63 (100 fold) reflected that DNp63 was not expressed at all before BMP-4 treatment (.35 cycles of qPCR) while mRNA of these genes were already detected before (22-25 cycles). To demonstrate that p63 is directly responsible for at least part of this modulation, the expression of potential p63-target genes were tested at day 14 in sh-63-cl2 ES cells as compared to control ES cells. Absence of p63 drastically reduced the expression of each gene during epidermal commitment, with the exception of BMP-7, which could be activated by an alternative mechanism, independent of the epidermal commitment (Fig. 4B). Similar results were obtained with the second sh-p63-cl1 (data not shown). We can conclude that the present model of ES-derived stratified epithelium commitment faithfully recapitulates molecular events known to be essential in epidermal morphogenesis.
Exogenous expression of DNp63 does not allow the production of K14-positive cells in absence of BMP-4 and serum Since DNp63 expression was detected at day 7, before the appearance of K14-positive cells, we evaluated whether exogenous expression of DNp63 could, in the absence of a BMP-4 dependent signalling pathway, induce keratinocyte commitment. ES cells were transduced, at day 4 of differentiation, with a lentiviral vector expressing DNp63. At days 8 and 14, expression of keratinocyte-specific gene was barely detected after infection with DNp63lentivirus, as compared to cells cultivated in the presence of BMP-4 and serum (Fig. 5A). Careful systematic counting of large areas under the microscope detected very few K14 + cells (data not shown). Thus, DNp63 is not able to induce keratinocyte differentiation in absence of BMP-4. To define if DNp63 can synergize the effect of BMP-4, a DNp63-expressing construct was stably inserted into ES cells which were induced to differentiate with and without BMP-4. As illustrated in Fig. 5B, no enhancement of K14 + or K5 + cell production was detected by qRT-PCR when the ES cells were treated with both BMP-4 and DNp63 as compared to BMP-4 alone. Furthermore, infection of ES cells with a lentivirus expressing DNp63 at day 4 of differentiation did not enhance the effect of BMP-4 on keratinocyte production (data not shown). Since we previously have shown that BMP-4 efficiently induces ES cell to differentiate into ectodermal K8 + /K18 + cells ( [13] and Fig. S2-A), it suggested that DNp63 can induce epidermal commitment only in ES-derived cells that already have received and integrated ectodermal signals delivered by BMP-4.

DNp63 expression promotes committed ectodermal cells to differentiate into keratinocytes
To further demonstrate that DNp63 is potent to induce ectodermal precursors to become keratinocytes, we isolated by serial dilution of BMP-4-treated ES cells at day 7, a homogenous ectodermal cell line positive for K8/K18 cytokeratins [25]. Exogenous expression of DNp63 in this ectodermal cell line indeed induced the differentiation of DNp63-transduced cells into K14positive cells (Fig. 6-A and B). None of the control-transduced cells expressed the keratinocyte-specific genes (Fig. 6-B). Since not all (about 25-40%) of the ectodermal cells expressing de novo DNp63 became K14 + -cells, we hypothesized that the ES-derived ectodermal K18 + cell population could be heterogeneous in its response to p63. To test if p63 still remains able to transactivate in these unresponsive cells, the ES-derived ectodermal cell line was cotransfected with DNp63 and a p63-responsive reporter construct [26]. 48 h after transfection, cells were analyzed by immunofluorescence staining for tagged-p63, luciferase and K14 gene expression (Fig. 6C). Interestingly, although not all the p63transfected cells became K14 + keratinocytes, they all expressed luciferase, strongly suggesting that the K18 + ectodermal cells unable to become K14 + cells are locked by an unknown mechanism. Further studies will be necessary to identify the molecular differences between cells belonging to this apparent homogenous ectodermal cell population. Finally, p63-target genes were also induced upon ectopic expression of DNp63 (not shown, Rostagno et al, in preparation). Altogether, these results indicate that DNp63 is necessary to switch-on the differentiation program of pre-committed ectodermal cells into K14-positive epithelial cells.

Discussion
The ability of ES cells to remain pluripotent and to differentiate in vitro in directed or selected cell fates, depending on culture conditions, make them a powerful tool to investigate in vitro molecular events involved in embryonic development. Here, we demonstrated that DNp63 is a key factor necessary for embryonic epidermal differentiation. In absence of p63, committed ectodermal cells are not able to produce neither proliferating (K14 + /K5 + ) nor terminally differentiated (K1 + /K10 + ) keratinocytes. The implication of p63 in pluristratified epithelia morphogenesis/ maintenance in mice was revealed by the severe phenotype of p63deficient mice, born with a defect in all pluristratified epithelia [2,3]. Nevertheless, a key question remained regarding the precise role of p63 in such epithelial developments. Indeed, the presence of uncommitted, single layered, primitive ectodermal cells covering the body surface of p63-deficient mice suggested that the p63-null phenotype leads to an absence of progression along the epidermal lineage with an early block in the epithelial stratification program [2,10]. In that case, the expression of p63 must occur very early during epidermal development before the stratification program will commit ectodermal cells into epidermal cells, as already suggested by Koster and Roop [10]. Accordingly, in our cell model, the expression of the DNp63 isoform occurred very early along the development pathway, soon after the ectodermal cell fate was induced by BMP-4 treatment and prior to the formation of K14 + /K5 + cells. Interestingly, modulation of DNp63 expression in zebrafish embryos has demonstrated its role in early dorsal/ventral ectodermal patterning as well as neural/ epidermal specification: in vivo inhibition of p63 was shown to prevent skin formation, while its exogenous expression promoted epidermal specification [21,22]. Furthermore, it has been recently reported that in an organotypic reconstitution assay, p63knockdown keratinocytes dedifferentiated toward an ectodermal fate, as they expressed de novo K8/K18 genes, and became unable to stratify [23]. Moreover, ectopic expression of p63 was shown to be able to initiate a stratification program in single-lung epithelia, as demonstrated by histological and molecular analyses [10]. All this strengthens the notion that K14 + /K5 + epithelial cells can only be produced when DNp63 is expressed, as clearly demonstrated in the present study.
On the other hand, clumps of terminal-differentiated cells in epidermis of p63-deficient mice were observed by McKeon et al. [3,6,9]. Thus, the authors suggested that the phenotype of these mice was not due to a default in embryonic epidermal differentiation but resulted from the inability of the epithelial stem cells to self renew. Accordingly, they recently demonstrated that DNp63 is required for adult epithelial stem cells to maintain their proliferative potential [6]. We believe that, instead of a contradiction, our data along with those described by Roop's team, strongly support a dual function for DNp63 during embryonic epidermal fate and in epithelial homeostasis. Indeed, self renewal deficiency in absence of p63 cannot explain that embryonic commitment of thymic epithelial cells is not affected in p63-deficient mice [6,8]. Furthermore, these thymic p63-deficient epithelial cells do not express K14 + /K5 + but K5 + /K18 + and cannot therefore be considered as equivalent of epidermal stem cells. It suggests that mechanisms, in addition to the proliferative capability, are affected by the absence of p63 in order to explain the lack of K5 + /K14 + epidermal cells in our in vitro model, as well as in vivo in p63-null mice.
Finally, we observed in absence of p63, the presence of structures composed of epithelial cells positive for K5 and K18. Interestingly, as for thymic K5 + /K8 + epithelial cell commitment shown in p63-deficient mice [6], the differentiation of these ESderived K18/K5-positive cells was not inhibited by the absence of p63. Altogether, these and our results suggest that epithelial lineages are differently affected by the absence of p63 during their morphogenesis. Distinct functions that might be activated sequentially during embryogenesis and the adult life span can then be attributed to p63. It is clear that p63 will have a major contribution in epithelial cell development through its regulatory impact on stem cell proliferative potential [6]. But it also appears that p63 may contribute to the required molecular steps implicated in other epithelial cell commitments, a role which is in accordance with the severe phenotype described for the p63-deficient mice. In this regard, we propose that, during embryogenesis, the ectodermal inducing phase (triggered by BMP-4 in our system) must include, as a necessary step, the expression of DNp63 to allow pluristratified epithelial cell fate to occur.

Cell culture and neuro-ectodermal differentiation
The mouse ES cell line CGR8 and derived clones were maintained as previously described [13]. Briefly, the ES cells were routinely cultured in flasks coated with PBS 0.1% gelatin, in GMEM medium (Gibco BRL, Invitrogen, Cercy-Pontoise, France) supplemented with 10% FCS (FCIII, Perbio-Hyclone, Bezons, France), 1 mM non-essential amino acids, 1 mM Sodium Pyruvate, 0.1 mM ß-mercaptoethanol and 10 3 U/ml LIF (Leukemia Inhibitory Factor, our own production), and passaged every 2 days. The ES-derived ectodermal cell line was routinely cultured in DMEM medium (Gibco BRL) supplemented with 10% FCS (FCII, Hyclone) and 0.1 mM bME, and passaged every 2 days. For the differentiation protocol, NIH-3T3 cells were grown to confluency in DMEM 10% FCS, they were fixed with PBS 3% formaldehyde (Sigma-Aldrich, France) for 15 min, washed with PBS and incubated with glycine (1 mM) (ICN Pharmaceuticals, France) for 15 min to saturate free formaldehyde sites. Dishes or coverslips were stored in PBS at 4uC until used for ES cell differentiation. For neuro-ectodermal differentiation, ES cells were cultured on fixed NIH-3T3 cells for 9 days in ES cell culture medium in which serum was replaced with 10% Knock-out Serum Replacement (KSR) (Gibco BRL). At day 9 and till day 14, the differentiation medium (10% KSR) was replaced by the ES cell culture medium (10% FCS, without LIF addition). When indicated, 0.5 nM human recombinant BMP-4 (R&D Systems Europe, UK) was added to the differentiation medium each day from day 3 to day 5.

Real-time quantitative RT-PCR
RNAs were extracted using Trizol reagent (Invitrogen) and incubated with DNase I (Ambion, France), according to the manufacturer's instructions to minimize genomic DNA contamination. cDNAs were synthesized using the Superscript II Reverse Transcriptase (Invitrogen). The relative expression level of transcripts was quantified by q-real time-PCR using the Power SYBR-Green PCR Master Mix (Perkin Elmer Applied Biosystems, France) on an ABI PRISM 7500 Sequence Detection System, according to the manufacturer's instructions. The expression of each gene was determined relative to 36B4 as an internal control and the fold stimulation was calculated by using the equation (2) 2DDCT , where DDCT = CT gene -CT 36B4 and DDCT = DCT stimulated condition -DCT unstimulated condition. Each gene was amplified using the appropriate specific primers.

Plasmid DNA Transfection: Generation of stable ES Shp63 clones
Plasmid pSuper vectors containing short-hairpin (sh) p63 or control sequence were kindly provided by C. Missero (CEINGE Biotecnologie Avanzate Scarl, Napoli). A plasmid c-myc-tagged-DNp63 vector was kindly provided by H. Van Bohkoven (Nijmegen, Netherlands). Undifferentiated ES cells (passage 25) were transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer instructions. Twenty-four hours later, transfected ES cells were passaged and treated 12 hours later with puromycin for 15 days. Resistant ES stable clones were isolated and cultured separately in the presence of puromycin (1 mg/ml). Isolated clones were tested by immunoblotting with anti-DNp63, either for expression of DNp63 (ES cell stably expressing DNp63) or for expression of the sh-p63 after transfection with the c-myc-tagged-DNp63 expression vector.

FACS analysis
FACS analysis was performed as previously described [13] with anti-mouse K14 (NE12 8EW) or K5 (PRB-160P) antibodies. Briefly, ES cells cultured on fixed NIH-3T3 cells were detached with trypsin 0.5 mM EDTA. They were fixed in 2% formaldehyde in PBS, permeabilized in PBS-0.5% saponin (Sigma-Aldrich)-0.5% BSA for 10 min and saturated with PBS-0.5% saponin-0.5% BSA-0.5% rabbit or swine serum depending on secondary antibodies for 30 min. The ES cells were stained with primary mAbs for 45 min. Mouse isotype control mAb (PharMingen) and control rabbit serum was used to set the background level of fluorescence. Rabbit anti-mouse and swine anti-rabbit coupled FITC antibodies (Dakko A/S, country) were added for 45 min. Cells were monitored by flow cytofluorimetry on a FACScan system using CellQuest software (BD Biosciences).