Rescue of key features of the p63-null epithelial phenotype by inactivation of Ink4a and Arf

Loss of Arf or Ink4a rescues features of the phenotype of p63^−/− mice

Mice lacking functional p63 are born with defects in the epidermis, have cleft lip and palate, and craniofacial abnormalities (Mills et al, 1999; Yang et al, 1999). These mice die within hours after birth because of desiccation resulting from the lack of a stratified epidermis covering the mouse. We hypothesized that loss of Arf or Ink4a could lead to the formation of a complete epithelium in the p63^−/− mice; therefore, Arf^GFP/GFP (Zindy et al, 2003) and Ink4a^−/− (Sharpless et al, 2001) mice were intercrossed with the p63^+/− mice (Yang et al, 1999) to first obtain double heterozygous mice (Arf^GFP/+;p63^+/− and Ink4a^+/−;p63^+/−). Subsequent matings by intercrossing compound heterozygous mice gave rise to Arf^GFP/GFP;p63^−/− and Ink4a^−/−; p63^−/− mice, respectively. The Arf^GFP/GFP knock-in allele is functionally null but yields green fluorescent cells when the Arf promoter is activated (Zindy et al, 2003). Macroscopic examination of 10 neonates of each genotype showed the presence of skin covering 80–95% of the body, but these animals did not survive after birth because of maternal neglect. Day 18.5 embryos (E18.5) of Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− mice had also developed a thin layer of skin covering 85–90% of the body (Table I). Twenty Arf^GFP/GFP;p63^−/− and 20 Ink4a^−/−;p63^−/− E18.5 embryos had dull and opaque skin with partially sloughed stratified epithelium (Figure 1C and D), compared with the shiny and translucent appearance of the p63^−/− embryos devoid of such epithelium (Figure 1B). In addition, both double mutant embryos exhibited rudimentary limbs and a long tail (Figure 1C and D). Strikingly, 16 of 20 Arf^GFP/GFP;p63^−/− and 18 of 20 Ink4a^−/−;p63^−/− mice had normal facial features including a complete palate, lip, and fully developed ears (Figure 1C and D) (Table I).

A more rigorous analysis of the skeletal structures of 10 embryos of each genotype was therefore undertaken (Figure 1E–P) (Table II). In the Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− embryos, the scapula, humerus, and ulna were well developed and in most cases comparable in size to wild-type embryos (Figure 1I–L) (Table II). In addition, rudimentary digits were apparent in both Arf^GFP/GFP; p63^−/− and Ink4a^−/−;p63^−/− embryos (Figure 1K and L), whereas the forelimbs were significantly smaller in the p63^−/− embryos (Figure 1J) (Table II). Similarly to the p63^−/− embryos, the Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− embryos did not develop hindlimbs; however, their pubic bones were comparable in size to those of wild-type mice and were fused (compare Figure 1O and P with Figure 1M) (Table II), a process that did not occur in the p63^−/− mice (Figure 1N). Thus, inactivation of Ink4a and Arf contributes independently to partially reverse cardinal features of the p63-null phenotype.

Loss of Arf or Ink4a leads to formation of a uniform layer of skin in the absence of p63

From macroscopic examination of double mutant embryos, we reasoned that loss of Arf or Ink4a alone had rescued epidermal defects detected in the p63^−/− embryos. The epidermis of the embryos was characterized by microscopic analysis using haematoxylin and eosin (H&E) staining and immunofluorescence (IF) for the epidermal markers, keratin 5 (K5), keratin 14 (K14), keratin 10 (K10), keratin 15 (K15), and filaggrin (Figure 2; Table I). The H&E-stained sections of E18.5 Arf^GFP/GFP;p63^−/− embryos showed an intact basal layer of epithelium surrounding the mice (Figure 2C). Twenty Arf^GFP/GFP;p63^−/− embryos had an average coverage of 85% as determined by H&E staining. The extent of rescue in the 20 Ink4a^−/−;p63^−/− embryos examined was even more striking with the appearance of structures that resembled hair follicles (Figure 2D; Table I); 10±2 rudimentary hair follicles per centimetre were apparent in all 20 of the Ink4a^−/−;p63^−/− embryos as compared with 64±4 cm in the wild type and none in the Arf^GFP/GFP;p63^−/− and p63^−/− mice (Table I). These data highlight the fact that inactivation of Ink4a and Arf exerts differential phenotypic effects in the p63-null background.

The structure of the epithelium was further analysed by IF using antibodies to K5 and K14, markers of the basal layer, K10, a marker of the spinous layer, and filaggrin, a marker of the granular layer of the epithelium. Immunofuorescence was also performed using markers for the basement membrane, collagen IV (Figure 2I–L), and for cellular adhesion, E-cadherin (Figure 2M–P). All of these markers are completely absent or are only weakly detected in the p63^−/− mice (Mills et al, 1999; Yang et al, 1999). Consistent with earlier published data (Yang et al, 1999), 3 of the 20 p63^−/− embryos had an average of 2% K5-positive or 5% K14-positive cells in patches above the dermis (Figure 2F, J, and N) (Table I). K5-positive and K14-positive cells in the Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− mice encircled the embryos (Figure 2G, H, K, L, O and P) (Table I) and formed a multi-layer stratum basale. Double IF for K14 and collagen IV further showed a multi-layer stratum basale above the basement membrane marked by collagen IV in the Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− embryos (Figure 2K and L). Immunostaining for K14 and E-cadherin showed cellular disorganization in the stratum basale and somewhat disrupted cell–cell adhesion in the double mutant embryos (Figure 2O and P). In addition, K10 and filaggrin staining were apparent, consistent with further differentiation of suprabasal cells (Figure 2S, T, W and X) (Table I). In the Ink4a^−/−;p63^−/− mice, the appearance of a multi-layer stratified epithelium, which stained positively for K5, K14, K10, and filaggrin (Figure 2H, L, P, T and X), was even more apparent, and hair follicle-like structures that stained positively for K15, a marker of epithelial stem cells residing in the hair follicle bulge (Liu et al, 2003), were also visible (Figure 2Y, AA, and BB). K15 staining was apparent in the stratum basale of the epidermis of Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− embryos (Figure 2AA and BB) indicating the presence of epidermal precursor cells, as was evident in the developing epidermis of wild-type embryos (Figure 2Y).

To determine whether loss of p53 could likewise rescue the p63^−/− phenotype, p53^−/−;p63^−/− embryos were also analysed at E18.5. Fourteen of 20 p53^−/−;p63^−/− embryos died in utero at day E15.5. Some had neural tube closure defects as has been reported earlier for p53^−/− mice (Sah et al, 1995). However, 6 of 20 embryos were viable at day E18.5 and had a similar phenotype to p63^−/− embryos (Table I), raising the possibility that rescue of the p63-null phenotype by Arf inactivation might be mediated, at least in part, through a p53-independent pathway.

Proliferation of p63^−/− keratinocytes is restored in the absence of Arf or Ink4a

Owing to grossly abnormal epithelialization in p63^−/− mice, it is difficult to explant and culture keratinocytes from these embryos. Indeed, when we attempted to culture keratinocytes on J2-3T3 feeder cells under conditions that normally allow expansion of clones from epidermal precursors (Barrandon and Green, 1987; Flores et al, 2000), cells could be extracted from the surface of p63^−/− embryos, but they did not proliferate (Koster et al, 2004). In contrast, when epidermis was isolated from Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− embryos at E18.5, recovered keratinocytes formed colonies much similar to those formed by wild-type keratinocytes (Figure 3A–E), and these stained positively with antibodies to the stratum basale marker, K5 (Figure 3F–J). However, the double null colonies were less compact than wild-type colonies (Figure 3C, E, H, and J), and the cells seemed to be less adherent to one another, based on their propensity to be more easily detached from the plate and dispersed by trypsinization. This feature is not surprising, given the discontinous staining of E-cadherin in Arf^GFP/GFP;p63^−/− and Ink4a^−/−; p63^−/− embryos and that p63 is known to regulate genes involved in cell adhesion (Ihrie et al, 2005; Carroll et al, 2006).

Although p63^−/− keratinocytes at passage-2 did not proliferate further, cells of all the other genotypes continued to accumulate over an 8-day time course (Figure 3K). The most rapidly expanding keratinocyte populations were those derived from the Ink4a^−/− and Arf^GFP/GFP embryos, but the Arf^GFP/GFP;p63^−/−, and Ink4a^−/−;p63^−/− keratinocytes divided as rapidly as wild-type keratinocytes. In addition to measuring their proliferative rates, we passaged the cultured keratinocytes every 5 days, and assessed their cumulative population doublings for seven passages (Figure 3L). The p63^−/− keratinocytes senesced at passage-1, whereas cells of all other genotypes continued to proliferate on further passage. The wild-type keratinocytes senesced at passage-3, but the Ink4a^−/−, Arf^GFP/GFP, Ink4a^−/−;p63^−/−, and Arf^GFP/GFP; p63^−/− keratinocytes reached 12–14 accumulated population doublings by passage-7 (Figure 3L). Notably, in both assays, the p53^−/−;p63^−/− keratinocytes behaved similarly to p63^−/− keratinocytes (Figure 3K and L) providing further evidence that loss of p53 does not rescue the ability of p63^−/− keratinocytes to proliferate. Taken together, these results show that loss of either Arf or Ink4a can rescue the proliferative defect of p63-deficient keratinocytes and that this rescue is not phenocopied by p53 inactivation.

The ability of rescued epidermal cells to proliferate in vivo was determined by the incorporation of the thymidine analogue, bromodeoxyuridine (BrdU), into the DNA of the developing epidermis of wild type, p63^−/−, Arf^GFP/GFP;p63^−/−, and Ink4a^−/−;p63^−/− embryos at day E18.5 (Figure 4A–H). BrdU-positive cells could be detected in multiple layers of the epidermis in Arf^GFP/GFP;p63^−/− and Ink4a^−/−;p63^−/− embryos (Figure 4E–H). Consistent with the findings made with cultured keratinocytes, inactivation of Ink4a or Arf significantly rescued the proliferation of p63-null cells, with elimination of Ink4a having the greater effect and restoring the proliferative capacity of the cells to wild-type levels (Figure 4I).

p63 downregulates p19^Arf and p16^Ink4a to rescue the proliferation and differentiation defects of cells lacking p63

The failure of p63^−/− mice to generate proliferative keratinocytes together with the rescue of this process in Arf-null and Ink4a-null backgrounds suggests that p63 negatively regulates p19^Arf and p16^Ink4a expression. Presumably, in the absence of p63, the untimely expression of p16^Ink4a and p19^Arf inhibits the outgrowth of keratinocytes, leading to the profound defects in epidermal development. Indeed, upregulation of p16^Ink4a in cultured primary keratinocytes was observed earlier after conditional p63 ablation, but in this setting, shRNA-mediated knockdown of Ink4a expression did not bypass cellular senescence; p19^Arf expression was not studied (Keyes et al, 2005). We made use of the GFP reporter built into the Arf^GFP/GFP knock-in mouse. Arf expression is not detectable during normal mouse epithelial development but is induced when cells are explanted into culture (Zindy et al, 2003). Direct measurements of GFP fluorescence indicated that Arf^GFP/GFP;p63^−/− keratinocytes expressed significantly higher GFP levels than those detected in age-matched Arf^GFP/GFP keratinocyte cultures (Figure 5A–C). Most pertinent, GFP levels were significantly elevated in the epithelium of Arf^GFP/GFP;p63^−/− embryos when compared with those in Arf^GFP/GFP embryos at E18.5 (Figure 5D–F). Immunoblotting of lysates from cultured keratinocytes documented a two-fold increase in GFP protein expression in Arf^GFP/GFP;p63^−/− cells versus that in their Arf^GFP/GFP counterparts (Figure 5G, lanes 3 and 4). Strikingly, the difference in the level of p19^Arf expressed in wild type versus p63^−/− cells was much greater (Figure 5G, lanes 1 and 2) indicating that p63 suppressed p19^Arf expression. The p19^Arf protein levels were also elevated in MEFs lacking p53 (Figure 5G, lane 5) as reported earlier (Kamijo et al, 1998; Stott et al, 1998). Likewise, similar increases in the levels of p16^Ink4a were detected in the absence of p63 also indicating that p63 downregulates p16^Ink4a expression (Figure 5H).

Quantitative real-time (qRT) PCR was performed on RNA extracted from embryonic skin containing the dermis and attached epidermal cells of embryos of the indicated genotypes (Figure 5I–K). Notably, the analysis of RNA expressed in dermal/epidermal tissues taken directly from embryos bypasses any inductive effects of ‘culture shock' seen with explanted cells propagated in vitro. Indeed, we found that levels of GFP RNA were 18-fold higher in Arf^GFP/GFP;p63^−/− versus Arf^GFP/GFP embryos (Figure 5I). Similarly, Arf levels were found to be 15-fold higher in p63^−/− compared with wild-type embryos (Figure 5J). Likewise, Ink4a RNA levels were increased in the absence of p63, although to a lesser extent (Figure 5K).

p63 directly represses Ink4a and Arf expression

Both the Ink4a and Arf promoters were found to have elements that match the p53 consensus binding site (Figure 6A and C). Chromatin immunoprecipitation (ChIP) analysis using an antibody for p63 and primers to amplify the discovered sites by qRT–PCR identified p63-binding sites upstream of the Ink4a and Arf promoters (Figure 6). Two sites were identified in both promoters (Figure 6B and D). The sites bound by p63 on the Ink4a promoter were located between nucleotides -4835 and -4621 (Figure 6A and B[I]) and -1850 and -1539 (Figure 6A and B[III]). Other sites analysed within the Ink4a promoter from nucleotides -3996 and -3845 (Figure 6A and B [II-UP]), between -3632 and -3362 (Supplementary Figure 1A[II-DOWN]), and -1519 and -1157 (Supplementary Figure 1B[IV]) did not show p63 binding. The p63-binding sites identified on the Arf promoter reside between -1533 and -1331 (Figure 6C and D[I]) and between -1276 and -943 (Figure 6C and D[IIA and IIB]). Other amplicons analysed upstream, -1811 and -1622 (Figure 6C and D[UP]), or in intervening sequences, -1417 and -1303 (Supplementary Figure 1C[INT]), were negative for p63 binding. As a positive control, ChIP analysis was performed using an antibody for p63 and primers to amplify known transcriptional targets of p63, p21, and Perp, by qRT–PCR (Supplementary Figure 2).

Transfection of the p63 isoforms in Arf^GFP/GFP;p63^−/− MEFs showed a marked repression of GFP protein and RNA levels (Figure 7A, C, and E) with ΔNp63α and -γ having the greatest effect. Similarly, introduction of the p63 isoforms into wild type and p63^−/− MEFs repressed expression of Arf protein and mRNA (Figure 7B, D, and F). p63 can also repress Ink4a. p63^−/− MEFs express Ink4a mRNA and protein at high levels. Transfection of the p63 isoforms resulted in a reduction in Ink4a mRNA (Figure 7G). Although all isoforms of p63 repressed Ink4a, the ΔNp63 isoforms repressed to the greatest extent (Figure 7G).

These data, taken together, indicate that p63 directly represses Ink4a and Arf gene expression and that the ectopic expression of these genes when p63 is absent results in a failure to form stratified epithelium.

Mouse husbandry

p63^+/− mice on (C57BL/6 × 129SvJae) background enriched for C57BL/6 (95%) (Yang et al, 1999; Flores et al, 2005) were intercrossed with Ink4a^−/− (Sharpless et al, 2001), Arf^GFP/GFP (Zindy et al, 2003) or p53^−/− (Jacks et al, 1994) mice to obtain compound mutant mice and embryos of the following genotypes: Ink4a^−/−; p63^−/−, Arf^GFP/GFP;p63^−/−, and p53^−/−;p63^−/−. Wild type and p63^−/− mice were also generated from each cross and analysed. All animal studies were approved by the Institutional Animal Care and Use Committee.

Histology and immunostaining of E18.5 embryos

Pregnant females at day 18.5 of gestation were injected three times at 1-h intervals with 100 μg of BrdU per gram of total body weight. Embryos were extracted 1 h later and fixed in 10% formalin at room temperature overnight, sectioned, and stained with H&E for microscopic analysis. For IF, paraffin-embedded sections were rehydrated in xylene and decreasing concentrations of ethanol. Sections were subjected to antigen unmasking in citrate buffer unmasking solution (Vector Laboratory) followed by a 1-h incubation with blocking solution, and overnight incubation at 4°C with primary antibodies to K5 (1:1000) (Abcam), K14 (1:500) (LifeSpan BioSciences), collagen IV (1:80) (Abcam), E-cadherin (1:200) (Abcam), K10 (1:1000) (Covance), K15 (1:500) (Covance), and filaggrin (1:1000) (Abcam). For staining with collagen IV antibody, unmasking was performed with Protease XXV (Fisher, AP-9006-002) at 37°C for 10 min. Visualization was performed using an anti-rabbit secondary antibody conjugated to Texas-red (1:5000) (Jackson ImmunoResearch Laboratories), an anti-guinea pig secondary antibody conjugated to FITC (1:1000, Jackson ImmunoResearch), or an anti-chicken secondary antibody conjugated to Alexa 488 (1:1000) (Molecular Probes) followed by counterstaining with DAPI (Vector Laboratory). Incorporation of BrdU was quantified using the BrdU detection kit II (Roche). For BrdU analysis by IF, an anti-rabbit secondary antibody conjugated to Texas Red was used with the BrdU detection kit II (Roche). GFP immunofluorescence was performed on frozen sections prepared from E18.5 embryos. Briefly, the epidermis was peeled from day 18.5 embryos and frozen in OCT compound (Tissue Tek). Frozen sections were fixed in 2% formaldehyde/0.2% glutaraldehyde solution for 2 min before incubation with anti-GFP (1:1000) (Invitrogen) at 4°C overnight followed by incubation with anti-rabbit secondary antibody conjugated to Texas-red (1:2000) (Jackson ImmunoResearch Laboratories) and counterstaining with DAPI (Vector Laboratory).

Skeleton preparation and staining

Embryos were collected at E18.5. Skin and internal organs were discarded, and the remaining embryo was fixed overnight in 95% ethanol at room temperature. The following day, the cartilage was stained in 0.05% alcian blue (Sigma) in 95% ethanol in acetic acid for 6 h and then destained in 1% KOH overnight. Bone staining was performed in 0.015% alizarin red (Sigma) in 1% KOH. The embryonic skeleton was washed in 20% glycerol/1% KOH for 5 days and changed twice per day until the solution was clear.

MEF and keratinocyte isolation and culture

Wild type, p53^−/−, p63^−/−, Arf^GFP/GFP, and Arf^GFP/GFP;p63^−/− MEFs were isolated from embryos at day 13.5 and cultured in DMEM with 10% FBS as described earlier (Flores et al, 2002). Wild type, p63^−/−, p53^−/−;p63^−/−, Arf^GFP/GFP, Arf^GFP/GFP;p63^−/−, Ink4a^−/−, and Ink4a^−/−; p63^−/− keratinocytes were isolated from E18.5 embryos. Briefly, the skin was peeled off and incubated in dispase II (Roche) for 18 h at 4°C to separate the epidermis from the dermis. Epidermal sheets were cut into small pieces and incubated in 0.25% trypsin–EDTA (Gibco-BRL) for 20 min at 37°C. Cells were plated on J2-3T3 feeder cells and cultured in F media (Sigma) supplemented with 0.4 μg/ml hydrocortisone, 24 ng/ml adenine, 8.4 ng/ml cholera toxin, 5 μg/ml insulin, 13 ng/ml 3,3,5-triiodo-L-thyronine, and 10 ng/ml EGF as described earlier (Barrandon and Green, 1987; Flores et al, 2000).

Immunostaining of keratinocytes

Keratinocytes seeded at a density of 4 × 10⁴ cells/6 cm diameter dish were fixed after 8 days in culture in 10% formalin for 30 min. Immunostaining was performed by blocking and permeabilization of cells using 8% bovine serum albumin in 0.3% Triton X-100 for 30 min at room temperature. Keratinocytes were incubated with anti-K5 (1:1000) (Abcam) or anti-GFP (1:200) (Invitrogen) overnight at 4^oC in a humidified chamber. Visualization of K5 was performed using an anti-rabbit secondary antibody conjugated to Texas-red (1:500) (Jackson ImmunoResearch laboratories) and GFP using an anti-rabbit secondary antibody conjugated to FITC (1:500) followed by counterstaining with DAPI (Vector Laboratory).

Keratinocyte proliferation assays

Passage-2 keratinocytes of 1 × 10⁴ were plated in triplicate on J2-3T3 feeder cells. Keratinocytes were counted every day for 8 days. Keratinocytes were also subjected to seven serial passages at 5-day intervals. After isolation, 1 × 10⁴ keratinocytes were plated in triplicate, passaged, and the cumulative number of population doublings was calculated.

Immunoblotting

MEFs and keratinocytes were lysed in buffer (100 mM Tris–HCl pH 8.0, 100 mM NaCl, 1% NP40, protease inhibitor cocktail tablet (Roche), and 0.2 M phenylmethylsulfonyl fluoride) and 100 μg of lysate were resolved by electrophoresis on a 12.5% SDS–PAGE gel, transferred to a PVDF membrane and probed with antibodies against p19^Arf (5C3-1) (Bertwistle et al, 2004), GFP (1:1000) (Invitrogen), p16^Ink4a (1:200) (Santa Cruz), p63-4A4 (1:500) (Santa Cruz), or Actin (1:5000) (Sigma) followed by goat anti-rat immunoglobulin G (IgG) or goat anti-rabbit IgG secondary antibodies conjugated to HRP (1:5000) (Jackson ImmunoResearch)

RNA extraction and real–time PCR

RNA was prepared from the dermis/epidermis peeled from wild type, p63^−/−, Arf^GFP/GFP, and Arf^GFP/GFP;p63^−/− 18.5 day embryos using an RNeasy plus kit (Qiagen Ltd.). cDNA was synthesized from 800 ng of total RNA using the SuperScript^® III First-Strand Synthesis kit (Invitrogen) according to the manufacturer's protocol. The RNA levels of Ink4a, Arf, and GFP encoded by the cellular Arf promoter were determined by qRT–PCR and performed in triplicate using an ABI 7900 machine. The following primers were used for performing real-time PCR: Ink4a sense 5′-CGTACCCCGATTCAGGTGAT-3′, Ink4a antisense 5′-TTGAGCAGAAGAGCTGCTACGT-3′ (Bruggeman et al, 2005), GFP sense 5′-GTCCGCCCTGAGCAAAGA-3′, GFP antisense 5′-TCCAGCAGGACCATGTGATC, and p19^ARF sense 5′-GCCGCACCGGAATCCT-3′, p19^ARF antisense 5′-TGGAGCAGAAGAGCTGCTACGT-3′ (Bruggeman et al, 2005). Primers for GAPDH were used as an internal control.

Chromatin immunoprecipitation

Wild type and Arf^GFP/GFP;p63^−/− keratinocytes were grown to near confluence on J2-3T3m feeder cells in F media as described above. Twenty-four hours before collecting chromatin, the feeder cells were removed with 0.02% EDTA. Cellular proteins were crosslinked to DNA using 1% formaldehyde and chromatin was prepared as described earlier (Flores et al, 2002). p63–DNA complexes were diluted 10-fold in ChIP dilution buffer and incubated overnight at 4°C with 2 μg of pan-p63 antibody (4A4, Abcam) or 2 μg IgG (no antibody). Resulting chromatin was resuspended in 300 μl of double-distilled H₂O. Each ChIP was performed in triplicate using keratinocytes from three embryos from each genotype. qRT–PCR was performed by using primers specific for the indicated regions of the Ink4a and Arf promoters (primer sequences below), p21 (Flores et al, 2002), and Perp promoters (Ihrie et al, 2005) using a 96 well plate ABI Step One Plus real-time PCR machine and 1 × SYBR green PCR master mix (Applied Biosystems). The percentage DNA bound was calculated using the following formula:

Percentage total bound=(2^(ΔCT) × 2.5)_p63antibody −(2^(ΔCT) × 2.5)_{no antibody}

where ΔCT=CT_{input chromatin}−CT_sample and CT is cycle threshold.

Primers for qRT–PCR are as follows:

A. Ink4a promoter

(1) Site I

-4835F-5′-ACCTCTCCAGGGTGGTCACAGGC-3′

-4621R-5′-AAAGGCCGAGTCAGATGATCTGGCC-3′

(2) Site II-upstream (II-UP)

-3996F-5′-ACACTGCACTGGAAGAGGACATCA-3′

-3845R-5′-GTTAAAAATGTGTAGACATGGGCC-3′

(3) Site II-downstream (II-DOWN)

-3632F-5′-CCACAGTTTGAACAGCGTGAAGCA-3′

-3362R-5′-AGGAAACGGATGTGATCCCAGGAA-3′

(4) Site III

-1850F-5′-AACTAGTAGAAGGGAGATTACTTATTG-3′

-1539R-5′-ATAAACTATGTAAATTTTCTGAGGC-3′

(5) Site IV

-1519F-5′-TGGTAGCAGCTTCTAACCCAGCA-3′

-1157R-5′-TTTGTATGTGTGTGGTGAGAATGTC-3′

B. Arf promoter

(1) Upstream regions (UP)

-1811F-5′-AGACATCAGGCAACAGGGAATGGA-3′

-1622R-5′-TTCGGGCTACCCTACCTGGAATTT-3′

(2) Site I

-1533F-5′-ATACTCAGGGAGCCCTCAAAACTC-3′

-1331R-5′-TAGACATAGGAAGATGGCAAGGCC-3′

(3) Intervening sequences (INT)

-1417F-5′-ACCTGCCCTGCGGCTCTCAGTTGC-3′

-1303R-5′-TCTTATAGAGCGGATTCCAAATGC-3′

(4) Sites IIA and IIB

-1276F-5′-ATTGCTACTTTACTGCAGCCAGAC-3′

-943R-5′-TTGATCGGAGCGCCGGCCGGGGAC-3′

Transfection of MEFs

MEFs were plated on a six-well dish at a density of 4 × 10⁵ cells/well. Twenty-four hours later, 2 μg of pcDNA3-p63 expression vectors (TAp63α, TAp63γ, ΔNp63α, or ΔNp63γ) were individually transfected using FuGENE HD Transfection Reagent (Roche) following the manufacturer's protocol. Cells were collected 24 h after transfection for analysis by western blotting and qRT–PCR.