p63/p51-induced Onset of Keratinocyte Differentiation via the c-Jun N-terminal Kinase Pathway Is Counteracted by Keratinocyte Growth Factor*

p63/p51, a homolog of the tumor suppressor protein p53, is chiefly expressed in epithelial tissues, including the epidermis. p63 affects cell death similar to p53, and also plays important roles in the development of epithelial tissues and the maintenance of epithelial stem cells. Because it remains unclear how p63 regulates epithelial cell differentiation, we examined the function(s) of p63 in keratinocyte differentiation through the use of a keratinocyte culture system. ΔNp63α (ΔNp51B), a p63 isoform specifically expressed in basal keratinocytes, suppressed the differentiation of specific late-stage proteins, such as filaggrin and loricrin. In contrast, ΔNp63α induced keratin 1 (K1), which is expressed at the start of differentiation, via c-Jun N-terminal kinase (JNK)/AP-1 activation. However, p63 did not induce K1 expression in the basal layer in vivo, although basal keratinocytes had high levels of p63. This discrepancy was explained by the suppression of K1 expression by dermis-secreted keratinocyte growth factor. This suppression occurred via extracellular signal-related kinase (ERK) signaling, and counteracted the p63-mediated induction of K1. Thus, a precise balance between p63 and keratinocyte growth factor mediates the onset of epithelial cell differentiation, through JNK and ERK signaling. These data may provide mechanistic explanations for the pathological features of skin diseases, including psoriasis.

p63, also known as p51, is a homologue of the tumor suppressor protein p53. The p53 family of transcription factors consists of three members, p53, p63, and p73 (1). The p63 gene possesses two transcriptional start sites that generate transcripts encoding proteins with (termed TAp63) or without (termed ⌬Np63) an N-terminal transactivation (TA) 3 domain. In addi-tion, alternative splicing primarily results in 3 isoforms, p63␣, p63␤, and p63␥, each of which is unique at its C terminus. Despite a high similarity in their structures, members of the p53 family seem to play mostly distinct functions in tumor suppression and development. Whereas mutations of p53 are found in more than half of all human cancers, the overexpression of p63, and particularly of the ⌬Np63␣ isoform, has been reported in a variety of squamous cell cancers (2)(3)(4). Furthermore, p63 mutations result in various human hereditary diseases involving ectodermal dysplasia, a variety of abnormalities characterized by defects in the skin and its associated structures (5). p63 Ϫ/Ϫ mice also display severe abnormalities of the skin, limbs, and mammary glands (6,7). This is in contrast to p53 Ϫ/Ϫ mice that develop apparently normal (8). The skin of p63 Ϫ/Ϫ mice has an epidermis that is thin, lacks stratification, and does not express markers of keratinocyte differentiation. These features indicate a critical role for p63 in regulating various facets of epithelial cell biology.
The epidermis is organized in several distinct overlying layers designated, from the bottom to the surface, as the basal, spinous, granular, and cornified layers (9,10). Actively proliferating keratinocytes are limited to the basal layer, and are characterized by their attachment to the underlying basement membrane. As keratinocytes begin to differentiate, they lose contact with the basement membrane and migrate to the suprabasal layer, which is the bottom layer of the spinous layer. Suprabasal layer cells express the differentiation-specific keratins 1 (K1) and 10 (K10) (11). In the granular layer, keratinocytes synthesize filaggrin (Fil) and loricrin (Lor), which contribute to cornified envelope formation in the outermost layer. Thus, the cells undergo complicated changes during differentiation, which requires the precise coordination of various events. A large body of work has highlighted important roles for many molecules in this differentiation process, including myc, Notch, and NFB (12)(13)(14).
p63 also plays an important role in keratinocyte regulation (15,16). Based on the severe epidermal defects found in p63 Ϫ/Ϫ mice, p63 has been implicated in the commitment of the ectoderm to stratified epithelia, as well as in the proliferative poten-tial of epidermal stem cells (6,7). However, the function(s) of p63 in a post-developmental context remain unclear. Analyses using skin organotypic culture have shown that p63 up-regulates cell growth in a p53-dependent manner (17). In addition, p63 induces K1 expression, which suggests that p63 helps initiate the differentiation process (18). Recent studies suggest that p63 function relies on key signaling molecules, such as Notch and IKK␣ (18,19). Thus, increasing data reveal the importance of p63 function in multiple keratinocyte biological processes.
We recently analyzed how p63 controls epidermal homeostasis in the post-developmental stage. Our study revealed that p63 maintains cell immaturity by inhibiting Notch activity (19). In addition, p63 inhibits UV-induced apoptosis via Akt activation (20). In the present study, we focused on the role of p63 in keratinocyte differentiation. We found that whereas p63 induced the initiation of differentiation, it also suppressed progression to late-stage differentiation. Because basal keratinocytes maintain an undifferentiated state in vivo despite high levels of p63, we hypothesized the existence of suppressive fac-tor(s) that counteract the induction of differentiation by p63. We found that differentiation was inhibited by keratinocyte growth factor (KGF) secreted from the underlying dermis. We therefore postulated the existence of a network for controlling the initiation of keratinocyte differentiation.
Cell Cultures, Transfection, and Adenoviral Infection-Primary keratinocytes were prepared from the newborn ICR mouse epidermis as described (24). In brief, the epidermis was separated from the dermis with 0.25% trypsin (Invitrogen, Tokyo) overnight at 4°C, plated in dishes precoated with colla-FIGURE 1. Induction of K1 and suppression of late differentiation markers by ⌬Np63␣. A, ⌬Np63␣ modulates the expression of differentiation markers in keratinocytes. Primary keratinocytes were infected with the LacZ (control) or ⌬Np63␣ adenoviruses (Ad-LacZ or Ad-⌬Np63␣), and were cultivated for the time period indicated (in hours) after replacement into a high calcium medium (2 mM). Total cell extracts were collected, and the proteins were analyzed by 7.5% SDS-PAGE and immunoblotting with antibodies against the various differentiation markers. Filaggrin was synthesized as a high molecular weight precursor, profilaggrin, which was subsequently processed. The diffuse bands correspond to the multiple products of this processing. The same blots were stripped and reprobed with an anti-K5 antibody, as a control for equal loading conditions. B, ⌬Np63␣ expression dose-dependently increases K1 expression. Keratinocytes were infected with increasing multiplicity of infections of Ad-⌬Np63␣. Virus was compensated for a total of 50 multiplicity of infection (MOI) by LacZ virus. K5 was used as a loading control. C, ⌬Np63␣ expression dose-dependently inhibits Fil expression. Keratinocytes, infected similarly to those in B, were cultivated in low (Ϫ) or high (ϩ) calcium concentrations. D, K1 transcription is increased by ⌬Np63␣. K1 mRNA levels were determined by real time RT-PCR with primers specific for the K1 gene. Values were normalized for glyceraldehyde-3-phosphate dehydrogenase mRNA levels and expressed as arbitrary units. E, immunoblots of keratinocytes transfected with p63-specific siRNAs. K1 levels decreased with p63 down-regulation by p63 sequence-specific (P-1 and -2), but not with green fluorescent protein sequence-specific (C), siRNAs. Arrow indicates ⌬Np63␣. K5 and tubulin-␣ were used as loading controls. The above results are representatives of at least three independent experiments.

p63-induced Keratinocyte Differentiation by JNK Activation
gen type I (Nitta Gelatin, Osaka, Japan), and cultured in minimum essential medium supplemented with 4% Chelex-treated fetal calf serum, epidermal growth factor (10 ng/ml; Invitrogen), and 0.05 mM CaCl 2 . Under these conditions, keratinocytes are maintained in an immature state, and differentiation was induced by the addition of CaCl 2 to a final concentration of 2 mM. The reagents added to the media were KGF, fibroblast growth factor-10 (FGF-10; R&D Systems), SP600125 (Merck, Tokyo), and PD98059 (Cell Signaling Technology). Transfections were carried out by use of Lipofectamine 2000 (Invitrogen, Tokyo).
To examine promoter activity, the transfected keratinocytes were subjected to a dual luciferase reporter assay (Promega) 72 h after transfection. Small interfering RNAs (siRNAs) targeting nucleotides 576 -596 (P-1) and 1728 -1748 (P-2) of ⌬Np63␣, where nucleotide 1 corresponds to the A base of the ⌬Np63␣ cDNA start codon, were synthesized commercially (Thermo, Ulm, Germany). P-1 and P-2 siRNAs recognize the DNA-binding domains of all isoforms and the carboxyl-terminal domains specific to ⌬Np63␣ and TAp63␣, respectively. An siRNA targeting green fluorescent protein was used as a control. Recombinant adenoviruses expressing either ⌬Np63␣ or LacZ were generated as described and used at a multiplicity of infection of 50 or as indicated (19,20). Adenovirus with LacZ was used as a control.
Immunoblotting and Immunostaining-Proteins were extracted from cultured keratinocytes and subjected to immunoblotting as described previously (24). In brief, electroblotted membranes were blocked with TBST (50 mM Tris, pH 7.5, 0.5% Tween 20) containing 5% nonfat dried milk. The membranes were then incubated with primary antibodies, rinsed with TBST, and incubated with peroxidase-conjugated secondary antibodies. After additional rinses, the blots were exposed to LumiGLO reagent (Cell Signaling Technology) and then to x-ray film. Equal loading was achieved by normalization of the protein concentration using a BCA protein assay (Pierce). Frozen skin sections (6 m thick) were fixed with 2% paraformaldehyde. Nonspecific binding was blocked by using 5% serum, and the skin sections were incubated with primary antibodies followed by incubation with isotype-specific secondary antibodies. Nuclei were stained with propidium iodide after the immunostaining. Stained preparations were photographed with a TCS 4D scanner (Leica, Heerbrugg, Switzerland) connected to an inverted LEITZ DM IRB microscope (Oberkonen, Germany).
Real-time PCR-The total RNA was prepared from primary keratinocytes using the RNeasy mini kit (Qiagen, Tokyo) according to the manufacturer's instructions. The cDNAs were synthesized by SuperScript II reverse transcriptase (Invitrogen), as recommended by the manufacturer, using 3 g of the total RNA as a template. Real-time PCR analyses were performed with a Quantitect SYBR Green PCR kit (Qiagen) and an iCycler (Bio-Rad, Tokyo), using the following primers: for K1, 5Ј-GAC CAG TCA CGG ATG GAT TC-3Ј and 5Ј-CGA ACT CAT TCT CTG CGT TG-3Ј; for FGF receptor 2 III␤ (FGFR2-III␤), 5Ј-AGT TTA AGC AGG AGC ATC GC-3Ј and 5Ј-TCA CAT TGA ACA GAG CCA GC-3Ј.
Statistical Analyses-The data were analyzed using unpaired Student's t tests. Differences were considered significant for p Ͻ 0.05.

RESULTS
⌬Np63␣ Suppresses Late Differentiation Proteins but Increases K1 Expression-We first examined the effects of p63 on the induction of differentiation markers in keratinocytes.

FIGURE 2. ⌬Np63␣-induced K1 expression is mediated by the JNK/AP-1 pathway.
A, the K1 regulatory region is activated by calcium addition. Keratinocytes were transfected with a luciferase reporter plasmid carrying the K1 regulatory region. B, ⌬Np63␣ expression increases the activity of the K1 regulatory region in a dose-dependent manner. Keratinocytes were transfected with either the ⌬Np63␣ expression plasmid or with an empty vector control, along with a luciferase reporter plasmid carrying the K1 regulatory region. C, ⌬Np63␣ expression dose-dependently increases the activity of the AP-1 response element. D, reduction of ⌬Np63␣ expression inhibits AP-1 activity. Keratinocytes were transfected with the AP-1 response vector together with p63-specific siRNAs (P-1 and -2). Green fluorescent protein-specific siRNA was used as a control (C). E, JNK phosphorylation in ⌬Np63␣-overexpressing cells. Primary keratinocytes were infected with Ad-LacZ or Ad-⌬Np63␣, and cell extracts were immunoblotted with phosphorylated JNK (p-JNK), JNK, and p63. JNK was used as a loading control. F, JNK inhibitor suppresses ⌬Np63␣-dependent K1 expression. Keratinocytes, infected with Ad-LacZ or Ad-⌬Np63␣, were incubated in various concentrations of SP600125. Tubulin-␣ was used as a loading control. The above results are representatives of at least three independent experiments. DECEMBER

p63-induced Keratinocyte Differentiation by JNK Activation
The keratinocyte is an excellent tool for analyzing epithelial tissues, because its features can be modulated in vitro through changes in extracellular calcium concentration. Cells cultivated in a low calcium environment mimic undifferentiated, proliferating keratinocytes. Upon replacement with a high calcium medium, the cells mimic in vivo differentiation processes, which include induction of differentiation-specific proteins such as K1, K10, Fil, and Lor. These proteins can be categorized as early markers (K1 and K10), which are initially expressed in the spinous layer, and late markers (Fil and Lor), whose expression is initially in the granular layer and beyond (11).
⌬Np63␣ is the predominant p63 isoform expressed in keratinocytes; it is down-regulated at the start of differentiation (25). We overexpressed ⌬Np63␣ in keratinocytes using an adenovirus transduction system. We observed that ⌬Np63␣ overexpression decreased levels of the late differentiation markers Fil and Lor, but did not alter K10 levels (Fig. 1A). However, ⌬Np63␣ increased K1 expression (Fig. 1A). This was unexpected, because ⌬Np63␣ is abundantly expressed in the basal layer where K1 is not expressed. K1 expression in the spinous layers was initiated during the differentiation process at time points when ⌬Np63␣ levels were gradually decreasing.
To confirm that ⌬Np63␣ was responsible for these changes in differentiation marker levels, we added various amounts of the ⌬Np63␣ vector to our culture system. K1 levels increased with increasing levels of ⌬Np63␣, in a dose-dependent manner. In contrast, Fil levels decreased dose-dependently (Fig. 1, B and  C). K1 up-regulation was also verified by real time RT-PCR (Fig.  1D). Conversely, ⌬Np63␣ reduction by two independent p63 sequence-specific siRNAs decreased K1 expression (Fig. 1E). Taken together, these results indicate that ⌬Np63␣ induces the expression of K1, which suggests that p63 induces the initiation of keratinocyte differentiation.
⌬Np63␣ Induction of K1 Is Mediated by JNK Signaling Activation-To investigate the mechanism of the ⌬Np63␣-dependent induction of K1, we examined the effects of p63 on K1 transcription. Human K1 induction is controlled by a regulatory region located in the 3Ј flanking area that includes the AP-1 response element (22,23). We isolated the K1 regulatory region and monitored luciferase reporter activity to examine the effects of p63 on K1 gene expression. A shift from low to high calcium concentration transactivated the K1 regulatory region ( Fig. 2A), which mimicked the endogenous response of the K1 gene. Transfection of ⌬Np63␣ transactivated the K1 regulatory region (Fig. 2B) and the AP-1 response element (Fig. 2C). Conversely, the activity of the AP-1 response element was decreased following siRNA reduction of p63 (Fig. 2D).
AP-1 activity is activated by JNK via Jun phosphorylation. Therefore, we used Western blot analysis to examine the effects of p63 on JNK activity by detecting JNK phosphorylation at threonine 183 and tyrosine 185. ⌬Np63␣ induced JNK phosphorylation in a dose-dependent manner (Fig. 2E). Furthermore, the JNK inhibitor SP600125 abrogated the ⌬Np63␣-dependent expression of K1 (Fig. 2F). These results indicate that ⌬Np63␣ induces K1 expression by activating the JNK/AP-1 pathway.
KGF Blocks the p63 Induction of K1 Expression-Despite K1 induction by ⌬Np63␣ in pure keratinocyte cultures in vitro, K1 is not induced in basal keratinocytes in vivo, even with high amounts of ⌬Np63␣. To investigate the reasons for this discrepancy, we sought to determine whether K1 induction was suppressed by the paracrine field effect that is mediated by skin cells except keratinocytes. We focused on dermal fibroblasts, which are thought to be responsible for preserving keratinocyte homeostasis. We collected conditioned medium (CM) from a culture of dermal fibroblasts, and added this CM to a keratinocyte culture (Fig. 3A). Cultured medium partially suppressed the ⌬Np63␣-induced expression of K1, suggesting that K1 was suppressed by some soluble factor(s) secreted from dermal fibroblasts.
We examined the roles of KGF (also known as FGF-7) and granulocyte macrophage-colony stimulating factor (GM-CSF) in the suppression of K1, because both KGF and GM-CSF regulate keratinocyte homeostasis in a paracrine manner (26). ⌬Np63␣-dependent K1 induction was blocked by KGF, but not by GM-CSF (Fig. 3B). KGF down-regulated the expression of the K1 protein and its mRNA in a dose-dependent manner (Fig.  3, C and D). Furthermore, a KGF neutralization antibody clearly enhanced K1 expression, albeit weakly, in LacZ-expressing or ⌬Np63␣-overexpressing keratinocytes (Fig. 3E). Because our cultivated dermal fibroblasts produced KGF mRNA (Fig. 3F), these results indicated that KGF secreted from dermal fibroblasts was responsible for K1 down-regulation. However, epidermal homeostasis is controlled by several soluble factors in addition to KGF, including interleukin-1, interleukin-6, tumor growth factor-␤, and platelet-derived growth factor (27). The ⌬Np63␣ induction of K1 was inhibited by FGF10, an FGF family member that closely resembles KGF (Fig. 3G). Thus, the FIGURE 3. KGF inhibits ⌬Np63␣-induced K1 expression. A, conditioned medium (CM) from dermal fibroblasts alleviates the p63-mediated increase in K1 expression. Keratinocytes were infected with Ad-LacZ or Ad-⌬Np63␣, and were cultivated with or without CM derived from dermal fibroblast culture. Cell extracts were prepared from keratinocytes, in which differentiation was induced by exposure to elevated extracellular calcium levels. The number of hours indicates the amount of time the cells were exposed to high calcium levels. The cell extracts were then immunoblotted with K1, and the same blot was stripped and reprobed with an anti-K5 antibody for equal loading. The densitometric analysis of the blot is also shown (right panel, cells without CM, white box; cells with CM, filled box). Values were normalized for K5 signals and expressed as arbitrary units. B, KGF suppresses ⌬Np63␣-induced K1 expression. Keratinocytes infected with Ad-LacZ or Ad-⌬Np63␣ were incubated with KGF (K), GM-CSF (G), or both (KG). K5 was used as a loading control. C, KGF suppresses ⌬Np63␣induced K1 expression in a dose-dependent manner. D, KGF suppresses K1 mRNA expression in a dose-dependent manner. Keratinocytes, transfected with either ⌬Np63␣ expression plasmid (filled box) or empty vector control (white box), were cultivated with various amounts of KGF. K1 mRNA levels were determined by real time RT-PCR with primers specific for the K1 gene. Values were normalized for glyceraldehyde-3-phosphate dehydrogenase mRNA levels and expressed as arbitrary units. E, addition of a KGF antibody enhances K1 expression in keratinocytes. Keratinocytes infected with Ad-LacZ or Ad-⌬Np63␣ were incubated with KGF neutralizing (␣-K) or control IgG (IgG) antibodies. The densitometric analysis of the blot is also shown (right panel: no treatment, white box; IgG, gray box; ␣-K, filled box). F, KGF mRNA expression in keratinocytes and dermal fibroblasts. Primary keratinocytes (K) and dermal fibroblasts (DF) were cultured for several days, followed by KGF mRNA quantification by real time RT-PCR. Values were normalized for glyceraldehyde-3-phosphate dehydrogenase mRNA levels and were expressed as arbitrary units. G, FGF10 suppresses ⌬Np63␣-induced K1 expression. Keratinocytes infected with Ad-LacZ or Ad-⌬Np63␣ were incubated with or without FGF10, followed by the induction of differentiation by elevated extracellular calcium levels. K5 was used as a loading control. H, KGF suppresses the expression of K10 and Lor, but not that of Fil. Keratinocytes infected with Ad-LacZ or Ad-⌬Np63␣ were incubated with or without KGF in low (Ϫ) or high (ϩ) calcium conditions. K5 was used as a loading control. The above results are representatives of at least three independent experiments. DECEMBER 5, 2008 • VOLUME 283 • NUMBER 49 induction of K1 expression by ⌬Np63␣ was inhibited by dermal fibroblast-expressed KGF. In addition, KGF blocked the expression of K10 and Lor, but did not alter the expression of Fil (Fig. 3H).

p63-induced Keratinocyte Differentiation by JNK Activation
The KGF-activating ERK Signal Competes with p63 in Inducing K1 Expression-To assess the mechanism of KGF-dependent K1 inhibition, we first tested the possibility that KGF inhibits the activity of the transcriptional regulatory element. ⌬Np63␣ increased K1 mRNA levels (Fig. 1D) and the activity of the K1 regulatory element (Fig. 2B), but KGF inhibited the ⌬Np63␣-dependent increase of K1 mRNA (Fig. 3D). However, KGF did not inhibit luciferase activity in the K1 regulatory element or in the AP-1 response element (Fig. 4, A and B). These results suggest that KGF controls K1 expression through a mechanism other than the AP-1 response element.
We next analyzed ERK signaling, which is thought to mediate not only cell proliferation but also cell differentiation. KGF induced MEK-ERK phosphorylation, and hence MEK-ERK activation (Fig. 4C). Use of the MEK inhibitor PD98059 revealed that K1 expression was up-regulated by the reduction of ERK activity (Fig. 4D). These results suggested the possibility that KGF inhibits K1 expression via ERK activation. However, despite its role in inducing K1 expression, ⌬Np63␣ activated the ERK pathway, similar to KGF (Fig. 4E). To clarify the role of ERK signaling in p63/KGF-mediated K1 expression, we inhibited ERK signaling in ⌬Np63␣-overexpressing and/or KGF-supplemented keratinocytes. The inhibition of ERK signaling in ⌬Np63␣-overexpressing keratinocytes enhanced K1 induction (Fig. 4F), indicating that ⌬Np63␣-mediated ERK activation inhibits K1 induction. Furthermore, ERK signaling inhibition counteracted the KGF-mediated inhibition of p63-induced K1 expression (Fig. 4G). Therefore, we concluded that ⌬Np63␣ suppresses K1 induction by ERK signaling, but that this suppression can be overwhelmed by JNK/AP-1-mediated K1 induction.
KGF Inhibits K1 Induction in Vivo-We examined the expression of the KGF receptor, FGFR2-III␤, by real time RT-PCR, which revealed an abundance of FGFR2-III␤ in undifferentiated keratinocytes under low calcium conditions (Fig. 5A). FGFR2-III␤ expression was minimal in differentiated keratinocytes and in dermal fibroblasts, consistent with previous data showing that FGFR2-III␤ localizes to the basal layer in vivo (28). Thus, KGF chiefly affects undifferentiated keratinocytes, such as basal keratinocytes. We also examined the effect of p63 on FGFR2-III␤ expression, and found that FGFR2-III␤ transcription was increased by ⌬Np63␣ expression, but not by TAp63␣ expression (Fig. 5B).
To assess the effects of KGF in vivo, we injected a KGF neutralizing antibody into the subcutaneous space of mouse skin.

. KGF-mediated suppression of K1 expression is independent of AP-1 activity and is dependent on ERK activation.
A, KGF does not modulate the increased activity of the K1 regulatory region by ⌬Np63␣. Keratinocytes, transfected with either the ⌬Np63␣ expression plasmid or an empty vector control, along with luciferase carrying the K1 regulatory region, were incubated with various amounts of KGF. B, KGF does not modulate the ⌬Np63␣-mediated increase in AP-1 activity. Keratinocytes were transfected with either the ⌬Np63␣ expression plasmid or an empty vector control, along with luciferase carrying the AP-1 response element. Similar to A, cells were incubated with KGF. C, KGF induces ERK and MEK phosphorylation. Protein extracts prepared from primary keratinocytes incubated with KGF at the concentrations indicated were subjected to immunoblotting with phospho-specific ERK and MEK antibodies (p-ERK and p-MEK). Antibodies against ERK and MEK were used as loading controls. D, MEK inhibition increases K1 expression. Keratinocytes were incubated with the MEK inhibitor PD98059 in low (Ϫ) or high (ϩ) calcium conditions. Anti-ERK and tubulin-␣ were used as loading controls. E, ⌬Np63␣ enhances ERK phosphorylation. Primary keratinocytes were infected with Ad-LacZ or Ad-⌬Np63␣, and cell extracts were immunoblotted with p-ERK, ERK, and p63. ERK was used as a loading control. F, MEK inhibition enhances p63-induced K1 expression. Keratinocytes infected with Ad-LacZ or Ad-⌬Np63␣ were incubated in various concentrations of PD98059. Anti-ERK and K5 were used as loading controls. G, MEK inhibition blocks the KGF-mediated suppression of K1 expression in p63-overexpressing keratinocytes. Keratinocytes were infected as in F and incubated in various concentrations of PD98059 with or without KGF. K5 was used as a loading control. The above results are representatives of at least three independent experiments.

p63-induced Keratinocyte Differentiation by JNK Activation
Although K1 is usually absent from basal keratinocytes, the blockade of KGF-induced K1 expression (Fig. 5, C and D). This suggests that KGF competes with p63 for K1 expression in undifferentiated keratinocytes, both in vitro and in vivo.

DISCUSSION
Previous studies have established specific requirements for p63 in epithelial homeostasis. Here, we show that p63 initiates keratinocyte differentiation via JNK activation, but suppresses keratinocyte progression to the late stages of differentiation. By signaling through ERK, dermis-secreted KGF can counteract the p63-mediated initiation of differentiation. Thus, the balance of p63 and KGF levels regulates the onset of cellular differentiation.
p63 Induces Keratinocyte Differentiation Onset via JNK Activation-Defects in the squamous epithelia of p63 Ϫ/Ϫ mice indicate that p63 is necessary for epithelial development. However, p63 is expressed abundantly in the squamous epithelia even after birth and also in some types of cancers, suggesting that p63 can still play an important role in the post-developmental stage. Previously, we reported that p63 supports cell proliferation and integrin expression by inhibiting Notch activity, causing growth-capable keratinocytes to remain in the basal layer (19). In the present study, we showed that ⌬Np63␣ induced K1, the earliest marker that is normally observed in the suprabasal layer, and blocked the induction of the late markers Fil and Lor. Although King et al. (29) have reported that ⌬Np63␣ suppresses K1 expression, which is contrary to our findings, Nguyen et al. (18) have reported K1 induction in primary mouse keratinocytes, similar to our results. The reason for the discrepancy between our study and that of King et al. (29) remains to be elucidated.
Recently, analyses using skin organotypic culture have suggested that p63 plays an important role in initiating keratinocyte differentiation, as well as in maintaining cell growth in developmentally mature keratinocytes (17). In the organotypic epidermis, ⌬Np63␣ knockdown leads to the impaired expression of squamous epithelium-specific proteins, including K1, K10, and Lor. ⌬Np63␣ knockdown also leads to the re-induction of the simple epitheliumspecific proteins K8 and K18. In the absence of ⌬Np63␣, this differentiation should be blocked at an early point and forbidden to proceed further. Therefore, p63 supports the continuance of cell differentiation, rather than the maintenance of cells in an undifferentiated state.
We showed here that p63 activates JNK, a classic stress-activated protein kinase. A number of stress stimuli, including UV radiation and inflammatory cytokines, induce JNK activation. Activated JNK can phosphorylate Jun family transcription factors, such as c-Jun that participates in the activation/formation of the AP-1 complex. The AP-1 complex is a heterogenous set of dimeric proteins, consisting of members of the Jun, Fos, and ATF families. In keratinocytes, AP-1 activation induces K1 transcription (22,23). However, many other genes have AP-1 binding sites in their promoters, including the late differentiation markers Fil and Lor (30). A detailed analysis of whether p63 controls late differentiation markers through the JNK/AP-1 pathway has yet to be performed. However, based on our data that ⌬Np63␣ differentially regulates the expressions of K1 and Fil/Lor, ⌬Np63␣ would be required for the initial commitment to the differentiated phenotype (K1 induction). Furthermore, the subsequent down-regulation of ⌬Np63␣ would be required for differentiation progression (Fil/Lor induction). In addition, because AP-1 binding sites are also located in the epidermal growth factor receptor, JNK activation in keratinocytes results in hyperproliferation and tumor formation via epidermal growth factor signaling (31,32). Therefore, p63 induces JNK activation, which can modulate not only differentiation but also proliferation, cell motility, and apoptosis.
We primarily analyzed the predominant ⌬Np63␣ isoform of p63, and not the TAp63 isoforms. The ⌬Np63 isoforms were originally proposed to act as repressor molecules against TAp63 and p53, but recent studies have indicated that ⌬Np63 proteins have unique and independent biological functions. Similar to ⌬Np63␣, TAp63␣ could induce K1 expression, but its effect was weaker than that of ⌬Np63␣ (data not shown). Given that the ⌬Np63 isoform shares a number of downstream genes with TAp63 (17), the weak induction by TAp63 may be caused by unstable characteristics of the isoform. Furthermore, the skin phenotype of p63-null mice can be partially rescued by the expression of ⌬Np63, but not by the expression of TAp63 (33). Thus, differentiation can be chiefly governed by ⌬Np63␣, although we cannot neglect the contribution of the TAp63 isoforms.
KGF Counteracts p63-induced K1 Expression-In contrast with the predominantly held view of p63, our data reveal that this protein promotes the expression of K1, a factor expressed at the start of keratinocyte differentiation. Despite its role in inducing K1 expression in vitro, p63 is highly expressed in basal keratinocytes, which do not induce K1 expression in vivo. This discrepancy suggests that the processes regulating the start of differentiation must involve other factors presumably derived from non-keratinocytes, because our cell culture consisted of almost pure keratinocytes. KGF meets this criterion.
KGF is a paracrine-acting growth factor produced by mesenchymal cells to stimulate epithelial cell proliferation via FGFR2-III␤ (34). It stimulates keratinocyte proliferation, consistent with its rapid induction in the healing wound (35). However, the effects of KGF on differentiation have not been well examined. In organotypic culture, KGF induces a delay in keratinocyte differentiation (36), suggesting that it inhibits the differentiation process. We show here that KGF counteracts differentiation progression via ERK phosphorylation, consistent with the function of ERK in delaying the differentiation process (37). In addition, the effect of KGF is chiefly localized to the basal layer, because FGFR2-III␤ is restricted to this layer (28). Therefore, basal keratinocytes can maintain their undifferentiated state through p63 and KGF signaling, whereas suprabasal keratinocytes undergo differentiation due to the function of p63 and the loss of FGFR2-III␤. Although not thought to be the sole dermis signaling molecule, KGF regulates keratinocyte homeostasis cooperatively with p63.
p63 can induce FGFR2 splicing toward the III␤ type by binding with ABBP1, a member of the RNA processing machinery (38). The p63 mutations found in Hay-Wells syndrome abolish p63/ABBP1 complex formation, which disturbs splicing to FGFR2-III␤. The altered skin phenotype observed in Hay-Wells syndrome is attributed to abnormalities in FGFR2 splicing. We observed an increase in FGFR2-III␤ expression when ⌬Np63␣ was overexpressed by keratinocytes, but not when TAp63 was overexpressed. This result suggests the possibility that p63 suppresses differentiation via KGF receptor expression while inducing differentiation via the JNK/AP-1 pathway. Furthermore, p63 dysregulation is also observed in psoriasis, a common skin disease marked by the hyperproliferation and delayed differentiation of epidermal cells (39). p63 is expressed throughout almost the whole epidermis, except for a few layers located at or near the skin surface. Interestingly, and at the same time as p63 expressed, FGFR2-III␤ is expressed basally, and in psoriasis is also expressed by differentiated keratinocytes (28). In addition, KGF production is increased in the mesenchymal cells subjacent to the epidermis (28). Whereas p63 induces the initiation of keratinocyte differentiation through JNK activation, the cooperative increase in KGF signaling may enhance cell proliferation and delay differentiation by activating both the JNK and ERK signaling pathways. This model is consistent with the activation of both signals in psoriasis (37,40,41). Our data on the role of p63 in keratinocyte differentiation may serve as a basis for the dissection of disease mechanisms, and may shed new light on the molecular mechanisms underlying skin physiology and pathology.