ΔNp63α Protein Triggers Epithelial-Mesenchymal Transition and Confers Stem Cell Properties in Normal Human Keratinocytes

Background: ΔNp63α is an isoform of p63 that is predominantly expressed in normal epidermis.

Results: Retroviral transduction of ΔNp63α into rapidly proliferating primary human epidermal keratinocytes led to epithelial-mesenchymal transition (EMT) and acquisition of stemlike properties.

Conclusion: ΔNp63α regulates EMT in primary human keratinocytes in a TGF-β-dependent manner.

Significance: Altering p63 level in NHEK may be a novel method to generate “induced mesenchymal stem cells” with multipotent capacity.

Abstract

p63 is a p53 family protein required for morphogenesis and postnatal regeneration of epithelial tissues. Here we demonstrate that ΔNp63α, a p63 isoform lacking the N-terminal transactivation domain, induces epithelial-mesenchymal transition (EMT) in primary human keratinocytes in a TGF-β-dependent manner. Rapidly proliferating normal human epidermal keratinocytes (NHEK) were infected with retroviral vector expressing ΔNp63α or empty vector and serially subcultured until replicative senescence. No phenotypic changes were observed until the culture reached senescence. Then the ΔNp63α-transduced cells underwent morphological changes resembling mesenchymal cells and acquired the EMT phenotype. Treatment with exogenous TGF-β accelerated EMT in presenescent ΔNp63α-transduced cells, whereas the inhibition of TGF-β signaling reversed the EMT phenotype. TGF-β treatment alone led to growth arrest in control NHEK with no evidence of EMT, indicating that ΔNp63α altered the cellular response to TGF-β treatment. ΔNp63α-transduced cells acquiring EMT gained the ability to be differentiated to osteo-/odontogenic and adipogenic pathways, resembling mesenchymal stem cells. Furthermore, these cells expressed enhanced levels of Nanog and Lin28, which are transcription factors associated with pluripotency. These data indicate that EMT required ΔNp63α transduction and intact TGF-β signaling in NHEK.

Introduction

p63 is a member of the p53 gene family, along with p73, that shares sequence and functional homologies (1). p63 is transcribed from two distinct promoters, resulting in TAp63 and ΔNp63 isoforms, based on alternative transcription start sites, and additional α, β, and γ isoforms, resulting from varied C-terminal lengths. p63 is highly expressed in basal cell nuclei of stratified epithelia in adult skin and absent in the underlying connective tissues. TAp63 is expressed in the early stage of skin morphogenesis and is responsible for initiation of epithelial stratification (2). On the other hand, mature epithelium primarily expresses ΔNp63, which is required for keratinocyte terminal differentiation and basement membrane integrity (3). Embryonic knock-out of the p63 gene in mice resulted in impaired craniofacial, skin, and limb development and embryonic lethality (4, 5). The most striking feature of these knock-out mice was the complete absence of stratified epithelium of skin, oral mucosa, and gut as well as other epithelial tissues. When p63 was deleted postnatally, mice exhibited shortened life span, features of accelerated aging, and cellular senescence (6). Because p63 is involved in epithelial stem cell maintenance (7), lack of epithelial structure in p63 knock-out mice may result from premature senescence of epithelial stem cells. These studies underscore the importance of p63 in morphogenesis and homeostasis of epithelial tissues, especially for keratinocytes, although the detailed mechanisms remain to be elucidated.

Epithelial-mesenchymal transition (EMT)² is a process by which epithelial cells lose their characteristics to gain mesenchymal phenotype, and this process is known to play crucial roles in embryogenesis, wound healing, cancer invasion, and metastasis. Through EMT, the cells lose epithelium-specific protein expressions, such as those of cytokeratins and E-cadherin (E-Cad), and enhance the expression of fibronectin (FN), vimentin, and N-Cad, resulting in increased cell motility (8, 9). EMT may occur as part of embryogenesis or wound healing and in disease processes, such as organ fibrosis and cancer metastasis. Although EMT may be induced by a plethora of signaling pathways, activation of TGF-β signaling plays critical roles in EMT either through its downstream Smad signaling or through a Smad-independent pathway (10). Smad signaling is triggered by activation of the TGF-β type II receptor (TβRII) upon binding of extracellular TGF-β ligand, which then recruits TGF-β type I receptor (TβRI), causing subsequent phosphorylation and nuclear translocation of Smad2/3 (11, 12). Independent of Smad, TGF-β may lead to EMT through activation of ERK MAPK, Rho GTPases, and the PI3K/Akt pathways (10). In addition, TGF-β up-regulates Snail and ZEB family proteins, which are the key transcription factors responsible for EMT (13, 14). Some studies show cooperative effects of TGF-β and other factors, such as activated Ras, in causing EMT, suggesting that TGF-β alone may not be sufficient in certain cell types (15).

EMT has been demonstrated in human cells using various methods. Overexpression of TGF-β, Snail, or Twist induced EMT in HMLE, human mammary epithelial cells immortalized with SV40 large T antigen (16). EMT was also observed in MCF10A, immortalized human breast epithelial cell line, by transduction of ΔNp63γ, through enhanced TGF-β production (17). Primary human mammary epithelial cells underwent EMT when exposed to exogenous TGF-β protein alone without any additional factors (18). Interestingly, some of these studies show a gain of stemlike properties in cells after conversion through EMT. Both HMLE and human mammary epithelial cells exhibited CD24^low/CD44^high surface marker profiles, which are linked with normal mammary and cancer stem cells, as well as mesenchymal stem cells (MSCs) (19, 20, 21). Another study showed multilineage differentiation capacity of HMLE into osteogenic, chondrogenic, and adipogenic pathways through EMT (22). These studies raise a hypothesis that EMT may be a novel approach to reprogram normal human epithelial cells to acquire multipotency and transdifferentiation. However, a great deal of research is needed to understand the underlying mechanisms, cell type specificity, functional resemblance or dissimilarity with MSCs, and tumorigenic potential. Most prior studies demonstrate EMT in human mammary epithelial cells, and no report has shown EMT in primary human keratinocytes. In this report, we demonstrate EMT in normal human epidermal keratinocytes (NHEK) by transduction of ΔNp63α in a TGF-β-dependent manner. Our data show the transdifferentiation ability of ΔNp63α in NHEK into osteo-/odontogenic and adipogenic pathways and phenotypic and functional resemblance to MSCs. Hence, altering the p63 level in NHEK may be a novel approach in generating “induced MSCs” with multipotent capacity.

EXPERIMENTAL PROCEDURES

Cells and Cell Culture

Primary NHEK were prepared from foreskin epidermal tissues according to the methods described elsewhere (23). The epithelial tissues were obtained from discarded surgical specimens according to the approval and guidelines of the UCLA Institutional Review Board. The primary cells were maintained in EpiLife medium supplemented with the growth factors (Invitrogen). These cells were serially subcultured, being passaged at 60–70% confluence levels until they reached senescence, and the replication kinetics were determined as described previously (23). For some experiments, keratinocyte differentiation was induced by Ca²⁺ exposure for 3 days at 1.5 mm. Normal human oral fibroblasts (NHOF) were obtained from gingival connective tissue explants and cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) supplemented with 10% fetal bovine serum (FBS) (Invitrogen). SCC9 was cultured according to the methods described previously (24). DPSC and SCAP were cultured in α-minimum Eagle's medium (Invitrogen) supplemented with 15% FBS (Invitrogen). Primary human sebocyte culture was obtained from Celprogen (San Pedro, CA) and maintained in Sebocyte culture medium from the same vendor supplemented with 10% FBS.

Retroviral Vector Construction and Transduction of Cells

We constructed retroviral vectors expressing wild-type ΔNp63α or TAp63α. Full-length cDNAs of ΔNp63α and TAp63α were cloned from the cDNA library obtained from primary human keratinocytes by PCR amplification using the following primers: ΔNp63α-Forward, 5′-CGGCTCGAGATGTTGTACCTGGAAAACAAT-3′; TAp63α-Forward, 5′-CGGCTCGAGATGAATTTTGAAACTTCA-3′; Reverse (common for both ΔNp63α and TAp63α), 5′-CGGGGATCCTCACTCCCCCTCCTCTTTG-3′. ΔNp63α and TAp63α cDNAs were then subcloned into pLXSN retroviral expression vector (Clontech) at XhoI/BamHI restriction sites (underlined sequences). The resulting clones were sequenced to confirm their isoform. Retroviral vector construction and infection were performed according to the methods described elsewhere (25). The infected cells were selected with 200 μg/ml G418 (Sigma), and drug-resistant cells were maintained in serial subculture as described above.

RT-qPCR

Total RNA was isolated from the cultured cells using the High Pure RNA Isolation Kit (Roche Applied Science). DNA-free total RNA (5 μg) was dissolved in 15 μl of H₂O, and the RT reaction was performed in the first strand buffer containing 300 units of Superscript II (Invitrogen), 10 mm dithiothreitol, 0.5 μg of random hexamer (Promega, Madison, WI), and 125 μm dNTP mixture (Promega). qPCR was performed in triplicates for each sample with LC480 SYBR Green I master (Roche Applied Science) using universal cycling conditions on a LightCycler 480 (Roche Applied Science). A total of 50 cycles were executed, and the second derivative Cq value determination method was used to compare -fold differences. The primer sequences for RT-qPCR are shown in supplemental Table S1.

Quantitative Measurement of Telomeric DNA and Telomerase Activity

Genomic DNA was extracted from cultured cells using the DNeasy blood and tissue kit (Qiagen). Relative telomere length was determined by using the approach as previously described (26). We also assayed for telomerase activity in cells using the quantitative telomeric repeat amplification protocol (Q-TRAP) assay (27). Complete methods are described in the supplemental material.

Western Blotting

Whole cell extract (WCE) from cultured cells was fractionated by SDS-PAGE and transferred to Immobilon protein membrane (Millipore, Billerica, MA) and probed with the following antibodies: Inv and FN (Sigma); Snail-1 (Abcam, Cambridge, MA); and p63/4A4, ΔNp63/N16, E-Cad, K14, p16^INK4A, Twist, GAPDH, and β-actin (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Horseradish peroxidase (HRP)-conjugated secondary antibodies were obtained from Santa Cruz Biotechnology, Inc. Chemiluminescence signal was detected using the HyGLO Chemiluminescent HRP antibody detection reagent (Denville Scientific Inc., South Plainfield, NJ).

Migration Assay

Scratch wound assay was performed as described previously (28). Briefly, the cells were seeded at 1.5 × 10⁵ cells/well in 6-well tissue culture plates. After cells reached 100% confluence level, a wound was made in the center of each well. Photographs were taken immediately or after 12 or 24 h of wounding. Cell migration was quantitated by measuring the area remaining from the original wounding, using Scion Image software (Scion Corp., Frederick, MD).

Cell migration was measured using transwell chambers with polycarbonate membranes (Corning Glass), according to the methods described elsewhere (29). Cells were seeded in the upper chamber of a transwell in EpiLife (Invitrogen) and were allowed to migrate for 24 h. The transwell was then washed with 1× phosphate-buffered saline (PBS), fixed in 10% formalin for 10 min, and stained with 1% crystal violet dissolved in 10% formalin for 1 h. After removing the non-migratory cells from the chamber, the transwells were photographed under phase-contrast microscopy. Migrated cells were measured by counting the number of cells that had crossed the membrane.

TGF-β1 Enzyme-linked Immunosorbent Assay (ELISA)

A TGF-β1 ELISA was performed according to the protocol (BD OptEIA Set Human TGF-β1; BD Biosciences). To measure secreted TGF-β1 level by NHEK, supernatants at low population doublings (PDs) with or without ΔNp63α transduction, senescent NHEK (PD 23), and NHEK/ΔNp63α-M (PD 25) were collected and centrifuged to remove any particulate materials. Briefly, 96-well microplates were coated with purified anti-human TGF-β1 antibody overnight at 4 °C. The plates were then blocked with an assay diluent for 1 h and incubated with the collected samples for 2 h. The plates were incubated with the substrate solution for 30 min, and the substrate reaction was stopped using the stop solution. The samples were read at 450 nm by a microplate reader within 20 min. The standard curve was established according to the manufacturer's protocol, and the reading value was normalized against cellular protein amount.

Luciferase Reporter Assay

TGF-β-responsive promoter activities were measured by p3TP-Luc plasmid, which is one of the standard reporters for assessing TGF-β activity. Prior to transfection, a 6-well plate with ∼5 × 10⁴ cells/well was inoculated and cultured for 24 h. p3TP-Luc plasmid (3 μg/well) was introduced into cells using Lipofectamine reagent (Invitrogen). Transfection efficiencies were controlled by pRL-SV40 plasmid (2 ng/well) containing the Renilla luciferase cDNA under the control of SV40 enhancer/promoter. Cells were treated with TGF-β1 (5 ng/ml) after 24 h post-transfection for 24 h and then collected, and the lysates were prepared using the Dual Luciferase reporter assay system (Promega, Madison, WI). Luciferase activity was measured using a luminometer (Turner Designs, Sunnyvale, CA). p3TP-Luc activity was determined as the mean of at least triplicates per experiment.

Analysis of Stem Cell Phenotype

Following trypsinization, the cells were fixed with 2% formaldehyde in PBS for 10 min at 4 °C. The cells were suspended in 2% FBS for a density of 1 × 10⁶ cells/ml for 10 min at 4 °C. Monoclonal antibodies against human CD24 (conjugated with FITC) and CD44 (conjugated with allophycocyanin) were added to the cell suspension and incubated at 4 °C in the dark for 1 h. The labeled cells were resuspended in PBS containing 3% FBS after washing with the incubation buffer and then analyzed using a BD FACScan flow cytometer (BD Biosciences). The fluorophore-conjugated antibodies were obtained from BD Biosciences.

For the Hoechst dye exclusion assay, the cells were resuspended at 1 × 10⁶ cells/ml in Hanks' balanced salt solution containing 2% FBS and then incubated with 10 μg/ml Hoechst 33342 (Sigma) for 90 min at 37 °C. A parallel aliquot of cells was stained with Hoechst 33342 after pretreatment with 40 nm reserpine (Sigma) for 10 min at 37 °C. After washing the cells with PBS, the cells were kept in 500 μl of PBS containing 3% FBS in ice and analyzed for Hoechst 33342 dye efflux using a BD FACSAria II cell sorter (BD Biosciences). Hoechst 33342 dye was excited at 350 nm using a UV laser. Side population (SP) was defined as reported previously (30).

Immunofluorescence Staining

Cells were fixed in 3.7% formaldehyde for 15 min and then permeabilized with 0.25% Triton X-100 for 10 min. Blocking was carried out in 10% normal goat serum for 1 h. Rabbit polyclonal anti-E-Cad (Santa Cruz Biotechnology, Inc.) and Alexa Fluor®488 goat anti-rabbit IgG (Invitrogen) were used as primary and secondary antibodies, respectively. Cells were counterstained with DAPI. Images were obtained using an Olympus BH2-RFCA fluorescence microscope under appropriate filters.

Assay of Alkaline Phosphatase (ALP) Activity and in Vitro Mineralization

Cells were placed at 1.5 × 10⁵ cells/cm² in 6-well plates. When the culture reached confluence, osteo-/odontogenic differentiation was induced in calcifying conditions (100 μm l-ascorbic acid 2-phosphate, 10 mm β-glycerophosphate, 10 nm dexamethasone, 0.18 mm KH₂PO₄, 5 μg/ml gentamicin sulfate) in α-minimum Eagle's medium containing 15% FBS. After 7 days of induction, the cells were stained for ALP activity in situ using an ALP staining kit (Sigma). To determine formation of mineralized nodules in vitro, cells were cultured to confluence and exposed to calcifying conditions for up to 3 weeks. The cultures were fixed in 70% ethanol for 1 h at 4 °C, rinsed in 1× PBS, and stained for 10 min with 40 mm Alizarin Red solution (pH 4.2) at room temperature. The cells were then rinsed five times in H₂O, followed by a 15-min wash in 1× PBS to eliminate nonspecific staining.

Adipogenic Differentiation

Rapidly proliferating NHOF or NHEK/ΔNp63α-M were placed at 1 × 10⁵ cells/cm² in 12-well plates. When the culture reached confluence, adipogenic differentiation was induced in induction medium (DMEM, 10% FBS, 60 μm indomethacin, 0.5 mm 3-isobutyl-1-methylxanthine, 1 μm dexamethasone, 5 μg/ml insulin, and 5 μg/ml gentamicin sulfate), as described previously (31). Basal, non-inducing medium consisted of DMEM, 10% FBS, and 5 μg/ml gentamicin sulfate. After 3 weeks of induction, cells were washed once with PBS and fixed in 4% formaldehyde for 15 min at room temperature. Fixed cells were washed in 60% isopropyl alcohol and stained for 1 h with 0.3% Oil Red O (Sigma). After three consecutive washes in deionized water, the stained cells were photographed using bright field microscopy.

RESULTS

Transduction of ΔNp63α Induces EMT Phenotype in NHEK

We constructed retroviral vectors expressing ΔNp63α (LXSN-ΔNp63α) or TAp63α (LXSN-TAp63α). Rapidly proliferating NHEK were infected with these viral vectors along with the empty vector (LXSN). To confirm expression of the transgenes, we performed Western blotting using pan-p63 and ΔNp63-specific antibodies (Fig. 1A). SCC9 was included as a control because it lacks endogenous p63 expression. NHEK infected with empty vector exhibited two bands approximately at 63 and 75 kDa and lacked expression of endogenous TAp63. Transduction of ΔNp63α led to a specific band near 63 kDa in both SCC9 and NHEK, and TAp63α transduction led to a single band at ∼80 kDa. ΔNp63α overexpression in the transduced cells was confirmed by RT-qPCR (supplemental Fig. S1A). Interestingly, ectopic ΔNp63α expression in NHEK reduced the level of the ΔNp63 band at ∼75 kDa. NHEK infected with LXSN or LXSN-TAp63α (denoted NHEK/LXSN and NHEK/TAp63α, respectively) replicated for 32 PDs and senesced (Fig. 1B).

FIGURE 1.

Transduction of ΔNp63α induces EMT in NHEK. A, SCC9 and NHEK were infected with retroviral vectors (LXSN, LXSN-ΔNp63α, or LXSN-TAp63α), and Western blotting was performed with WCEs using two different primary antibodies (4A4 and N16). Pan-p63/4A4 can detect both TA and ΔN isoforms of p63, whereas ΔNp63/N16 is specific for ΔN isoforms. n.s., nonspecific band. GAPDH is used as a loading control. Size markers (kDa) indicated are based on the N16 blot. B, rapidly proliferating NHEK were infected with LXSN, LXSN-ΔNp63α, or LXSN-TAp63α and selected with G418 (200 μg/ml). The G418-resistant cells were maintained in serial subcultures until senescence, and their replication kinetics were plotted against time in culture. Noted on the graph are the point at which the cells were infected with the viral vectors (*) and the point at which the EMT cells were first identified only in ΔNp63α-transduced cells (**). C, photomicrographs were taken from the cultures infected with LXSN, LXSN-ΔNp63α, or LXSN-TAp63α at PDs 25 and 32. Original magnification was ×100.

Replicating cultures of NHEK expressed ΔNp63 isoforms, which were drastically lost in cells approaching subculture-induced senescence (supplemental Fig. S1B). NHEK infected with LXSN-ΔNp63α (NHEK/ΔNp63α) followed the replication profiles of NHEK/LXSN and NHEK/TAp63α and did not show any phenotypic alteration, including cell morphology or replication kinetics. When the culture entered the senescing phase past PD 30, NHEK/ΔNp63α exhibited morphological changes resembling mesenchymal cells (Fig. 1C). These cells were designated NHEK/ΔNp63α-M to reflect their mesenchymal phenotype.

Occurrence of EMT by ΔNp63α transduction in NHEK was determined by assessing the expression of respective molecular markers. Western blotting was performed for FN, E-Cad, and keratin 14 (K14) in NHEK/LXSN or NHEK/ΔNp63α using WCE (Fig. 2A). FN expression level increased in NHEK/LXSN with increase in PD numbers. ΔNp63α transduction led to stronger induction of FN, higher than that of NHOF. E-Cad and K14 levels were unaltered in NHEK/LXSN during subcultures. At PD 26, NHEK/ΔNp63α showed similar levels of E-Cad and K14 as the controls but completely lost expression of these proteins at higher PDs. Loss of E-Cad and K14 expression coincided with the point at which NHEK/ΔNp63α acquired the mesenchymal morphology at PD 32 (Figs. 1B and 2A). Likewise, expression of Snail, which is linked with EMT (32), was observed in NHEK/ΔNp63α-M at PD 39 but not at PD 26 nor in NHEK/LXSN (Fig. 2B). In addition, NHEK/ΔNp63α-M cells expressed a negligible level of involucrin, which is a marker of keratinocyte differentiation (23), and their response to increased Ca²⁺ level was drastically reduced (Fig. 2C). These data indicate that ectopic ΔNp63α expression in NHEK triggers EMT.

FIGURE 2.

Transduction of ΔNp63α alters the epithelial or mesenchymal markers. Rapidly proliferating NHEK were infected with LXSN or LXSN-ΔNp63α, and the infected cells were serially subcultured. Western blotting was performed with the cells at varying PD levels for FN, E-Cad, and K14 (shown in A) and for Twist and Snail (B). Parental NHEK and NHOF were included for comparison. C, NHEK/LXSN (PD 25) and NHEK/ΔNp63α (PDs 25 and 38) were exposed to 1.5 mm Ca²⁺ for 3 days, and Western blotting was performed for involucrin to assess Ca²⁺-induced keratinocyte differentiation. For all blotting experiments, β-actin was used as a loading control.

Enhanced cell migration is a hallmark of EMT phenotype (33). Therefore, in the next experiment, we performed an in vitro scratch assay with NHEK/LXSN, NHEK/TAp63α, NHEK/ΔNp63α, and NHEK/ΔNp63α-M to compare cell migration capacity. After the cultures reached confluence, a scratch wound was made to induce cell migration, and the cultures were photographed at 0, 12, and 24 h after wounding (Fig. 3A). During the 24-h period, NHEK/LXSN and NHEK/TAp63α showed limited migration, and the wound closure was not complete. NHEK/ΔNp63α showed a similar degree of limited migration, but NHEK/ΔNp63α-M acquired enhanced migration, which led to complete closure of wound after 24 h. When we assessed cell migration in transwell inserts, NHEK/ΔNp63α-M demonstrated notably increased migration compared with the other cell types (Fig. 3B). These data provide functional evidence of EMT in NHEK/ΔNp63α-M.

FIGURE 3.

EMT phenotype induced by ΔNp63α enhances cell migration of NHEK. A, rapidly proliferating NHEK were infected with LXSN, LXSN-TAp63α, or LXSN-ΔNp63α, and the infected cells were cultured to confluence. ΔNp63α-M denotes NHEK transduced with ΔNp63α exhibiting the mesenchymal phenotype. A scratch wound was made in the middle of the culture to initiate cell migration, and wound closure was determined at 12 and 24 h after wounding. Representative areas are shown in the photomicrographs. Original magnification was ×100. Relative wound closure (%) with respect to the initial wound (time 0) was quantitated by measuring the wounded area not occupied by the migrated cells. Error bars, S.D. B, NHEK/LXSN, NHEK/ΔNp63α, and NHEK/ΔNp63α-M were tested for cell migration in transwells. NHEK/ΔNp63α-M, cells transduced with ΔNp63α and exhibiting the mesenchymal phenotype. After 24 h postseeding, the cells that have migrated across the transwells were stained as shown in the photomicrographs. The migrated cells were counted and graphed. Original magnification was ×100. Error bars, S.D.

Next, we tested the possibility that NHEK/ΔNp63α-M cells represent contaminating sebocytes, which comprise sebaceous glands of skin. We assayed for the expression of sebocyte markers (e.g. keratin 7 (K7) and peroxisome proliferator-activated receptor γ (PPARγ)) (34, 35). As shown in supplemental Fig. S2A, K7 and PPARγ were uniquely expressed in primary human sebocytes but not in NHEK, even after the occurrence of EMT by ΔNp63α transduction. Morphologically, sebocytes differ from NHEK exhibiting the EMT phenotype (supplemental Fig. S2B). Thus, sebocyte contamination in NHEK culture is unlikely.

TGF-β Signaling Is Required for ΔNp63α-mediated EMT in NHEK

Our data show that EMT phenotype in NHEK/ΔNp63α did not appear until the cells entered the senescing phase past PD 30 (Fig. 1C). This indicates that ΔNp63α requires an additional endogenous factor that only appears in senescing cells. Because TGF-β accumulates in subcultured NHEK to mediate senescence-associated differentiation (36), we hypothesized that ΔNp63α cooperates with TGF-β to induce EMT. To test this possibility, we compared the kinetics at which EMT occurred in NHEK/ΔNp63α with and without exogenous TGF-β. ΔNp63α transduction alone led to EMT phenotype after 25 days postinfection, and the number of EMT cells had since increased exponentially (Fig. 4A). NHEK/ΔNp63α-M appeared after 6 days of TGF-β treatment (5 ng/ml), notably faster than the control culture, although the exponential kinetics remained the same. Exogenous TGF-β treatment caused growth inhibition and terminal differentiation in control NHEK (Fig. 4B and supplemental Fig. S3). TGF-β also led to growth suppression in NHEK/ΔNp63α before emergence of cells demonstrating the EMT phenotype. Thus, ΔNp63α did not evade TGF-β-induced growth arrest per se but allowed EMT to occur among those that exhibit growth arrest in ΔNp63α-transduced culture, whereas the control cells remained growth-arrested. Western blotting showed that TGF-β treatment led to loss of E-Cad and increased FN and Snail levels only in NHEK/ΔNp63α and not in NHEK/LXSN or NHEK/TAp63α (Fig. 4C). When NHEK/ΔNp63α-M were exposed to TGF-β receptor I inhibitor (TβRI-i), the cells regained the epithelial morphology, and E-Cad and K14 expression levels were restored (Fig. 5, A and B).

FIGURE 4.

Exogenous TGF-β treatment accelerates ΔNp63α-induced EMT in NHEK. A, NHEK were infected with LXSN-ΔNp63α and were cultured in the presence of TGF-β1 (5 ng/ml). Kinetics of EMT was determined by counting the cells with the transformed morphology every 3 days. B, NHEK/LXSN and NHEK/ΔNp63α cells were cultured with or without TGF-β1 (5 ng/ml) for 10 days and stained for E-Cad (green). Same cultures were counterstained with Hoechst 33342 (blue). Original magnification was ×100. C, NHEK/LXSN, NHEK/ΔNp63α, and NHEK/TAp63α were cultured with or without TGF-β1 (5 ng/ml) for 10 days. Using WCE, Western blotting was performed for E-Cad, FN, Snail, and p16^INK4A. β-actin was used as a loading control.

FIGURE 5.

TGF-β signaling is required for ΔNp63α-mediated EMT in NHEK. A, NHEK/ΔNp63α at PD 27 (pre-EMT) and PD 34 (post-EMT) were exposed to TβRI-i at 1 μm for 14 days. Photomicrographs show reversion of NHEK/ΔNp63α-M to cobblestone-shaped epithelial morphology after the drug treatment. Original magnification was ×100. B, Western blotting was performed for p63, E-Cad, and K14 using WCE of NHEK/ΔNp63α-M with or without TβRI-i treatment for 14 days. β-Actin served as the loading control. C, NHEK/ΔNp63α (pre-EMT) were infected with Babe-B0 or Babe-Bmi-1 and serially subcultured to induce EMT. Western blotting was performed for E-Cad and Snail at PDs 38 and 46 for NHEK/ΔNp63α/B0 and NHEK/ΔNp63α/Bmi-1, respectively. NHEK/LXSN and NHEK/ΔNp63α without exogenous Bmi-1 transduction were included as controls. D, NHEK were infected with Babe-B0 or Babe-Bmi-1 as the primary (1°) infection and then superinfected with LXSN-ΔNp63α. Western blotting was performed for E-Cad to assess the occurrence of EMT at PDs 35 and 38 for NHEK/B0/ΔNp63α and at PDs 39 and 46 for NHEK/Bmi-1/ΔNp63α. β-Actin was included as a loading control for all blotting experiments.

We recently showed that Bmi-1 inhibits the TGF-β signaling in normal human oral keratinocytes (NHOK) (36). To determine whether Bmi-1 can prevent ΔNp63α-mediated EMT in NHEK, we infected NHEK with LXSN-ΔNp63α and then subsequently with retroviral vector expressing Bmi-1 or the empty vector (B0). Conversely, in another experiment, we transduced Bmi-1 first and then ΔNp63α in NHEK. In both cases, ΔNp63α transduction led to loss of E-Cad and Snail induction, but ectopic expression of Bmi-1 prevented EMT in these cells, whereas the control vector (B0) had no effect (Fig. 5, C and D). Therefore, intact TGF-β signaling is required for ΔNp63α-mediated EMT in NHEK and for the maintenance of the EMT phenotype.

We measured the TGF-β1 level secreted from cultured NHEK and found that ΔNp63α transduction increased the level of TGF-β1 secretion compared with the control cells at the same PD levels (Fig. 6). TGF-β1 level was further increased in cells at higher PDs in control cells and those transduced with ΔNp63α exhibiting the EMT phenotype. Also, ΔNp63α transduction led to enhanced phosphorylation of Smad2/3 proteins and TGF-β1-responsive promoter activities in cells exposed to exogenous TGF-β1. Thus, ΔNp63α transduction allowed for continued production of TGF-β and may sensitize the cells to the TGF-β signaling pathway, thereby resulting in EMT in late passage cultures.

FIGURE 6.

ΔNp63α enhances TGF-β signaling in NHEK. A, secreted TGF-β was measured in nutrient medium conditioned with NHEK/LXSN or NHEK/ΔNp63α at varying PDs as indicated by ELISA. B, Western blotting was performed with NHEK/LXSN or NHEK/ΔNp63α exposed to TGF-β1 in the presence or absence of TβRI-i for phosphorylated and total Smad2/3. β-Actin was used as loading control. C, activities of TGF-β-responsive promoter reporter construct (p3TP-Luc) were measured in NHEK/LXSN, NHEK/ΔNp63α, and NHEK/TAp63α with or without TGF-β1 treatment (5 ng/ml). Firefly luciferase activity was normalized against Renilla luciferase activity driven by the SV40 promoter. Bars represent mean ± S.D. (error bars) of triplicates.

ΔNp63α-mediated EMT Confers Stem Cell Phenotype and Multipotency in NHEK

Next, we explored the possibility that NHEK/ΔNp63α-M cells acquired stem cell properties. Immunophenotyping was performed for CD24 and CD44 surface markers by flow cytometry. MSCs like dental pulp stem cells (DPSC) and stem cells of apical papillae (SCAP) were marked with the CD24^low/CD44^high phenotype in comparison with NHEK/LXSN, which exhibited 52 and 3% positive staining for CD24 and CD44, respectively (Fig. 7A). NHEK expressing exogenous Snail (NHEK/Snail) and NHEK/ΔNp63α-M demonstrated CD24^low/CD44^high in relation to NHEK/LXSN. Also, when we determined SP content by a Hoechst efflux assay (37), ΔNp63α-mediated EMT led to an increase of SP level to 4.3%, whereas the control cells and NHEK/ΔNp63α cells exhibited 2.7 and 1.6% SP level, respectively (Fig. 7B). We found increased expression of the genes associated with stemness, such as Lin28 and Nanog in NHEK/ΔNp63α-M, compared with NHEK/LXSN or NHEK/ΔNp63α (Fig. 7C). Western blotting confirmed up-regulation of Nanog in NHEK upon ΔNp63α transduction (supplemental Fig. S4A).

FIGURE 7.

EMT induced by ΔNp63α confers stem cell-like properties in NHEK. A, flow cytometry was performed with MSCs (DPSC and SCAP), NHEK/LXSN, NHEK/Snail, and NHEK/ΔNp63α-M for surface expression of CD24 and CD44. Numerical values in each group represent CD24^low (upper panels, horizontal axis) and CD44^high (lower panels, vertical axis). B, a Hoechst dye exclusion assay was performed in DPSC, SCAP, NHEK/LXSN, NHEK/ΔNp63α (pre-EMT), and NHEK/ΔNp63α-M in the presence or absence of reserpine (40 nm). SP cells are noted in green, and their abundance is shown as numerical values for each group (%). C, RT-qPCR was performed for Oct4, Nanog, and Lin28 in NHEK/LXSN, NHEK/ΔNp63α at PD 26 (pre-EMT), NHEK/ΔNp63α-M at PD 41, NHOF, SCAP, and DPSC. D, relative telomere length was determined by qPCR analysis of 30 ng of DNA of HOK-16B, NHEK (PD16), NHEK/ΔNp63α-M, NHOF, and DPSC. Average telomere versus single copy gene (T/S) ratio was used to compare relative telomere length. E, telomerase activities in cells were determined with increasing amounts of cell extracts (0.125–1 μg) using a SYBR Green Q-TRAP assay. Error bars, S.D. from triplicate samples.

We previously showed differences in telomere length dynamics between NHOK and NHOF in that keratinocytes exhibit shorter (∼6 kbp) telomeres than fibroblasts (∼8–10 kbp) and that keratinocytes replicate with limited telomere shortening (24, 38). In the current study, we compared the relative telomere lengths in NHEK before and after EMT by the qPCR method. HOK-16B, an immortalized counterpart of NHOK (39), demonstrated lower telomere length compared with NHEK, consistent with our prior report (40) (Fig. 7D). Rapidly proliferating NHEK exhibited significantly shorter telomere length than NHOF, and NHEK/ΔNp63α-M demonstrated extended telomere length from parental NHEK. This extension of telomere length did not occur from activation of telomerase enzyme, which was strongly detected in HOK-16B and weakly in NHEK, but only a negligible amount was found in NHEK/ΔNp63α-M, NHOF, and DPSC (Fig. 7E). These data indicate that EMT conversion through ΔNp63α in NHEK led to increased telomeric length, independent of telomerase activation.

MSCs are characterized by their multipotent differentiation capacities (31, 41). We tested if NHEK/ΔNp63α-M gained multipotency like MSCs. ALP activity was measured in confluent culture after 7 days of exposure to calcifying conditions to induce osteo-/odontogenic differentiation (Fig. 8, A and B). ALP activity was strongly induced in NHEK/ΔNp63α-M and the MSCs, whereas it was absent in parental NHEK, NHEK/Snail, and NHOF. Mineralization capacity was compared by Alizarin Red staining after 28 days of exposure to the calcifying conditions (Fig. 8C). NHEK/ΔNp63α-M demonstrated robust mineralization, whereas it was not evident in NHEK/Snail. Assessment of the genes related to osteo-/odontogenic differentiation revealed strong induction of the gene expression in NHEK/ΔNp63α-M compared with NHEK or NHEK/Snail (Fig. 8D). Western blotting confirmed expression of DSP, an odontogenic differentiation marker (42), in NHEK/ΔNp63α-M and DPSC, and absence of the protein expression in NHEK and NHEK/Snail (Fig. 8E). Adipogenic differentiation potential was assessed in NHEK/ΔNp63α-M by inducing the cells in adipogenic differentiating medium. After 21 days of induction, the cultures were stained for cytoplasmic lipid droplets with Oil Red O stain (supplemental Fig. S4B). Robust accumulation of lipids was noted in NHEK/ΔNp63α-M cells but not in NHOF, which served as a negative control. These data suggest that EMT triggered by ΔNp63α in NHEK confers stem cell properties, similar to MSCs.

FIGURE 8.

ΔNp63α induces osteo-/odontogenic differentiation of NHEK through EMT. A, ALP activity was detected by in situ staining in NHEK (parent), NHEK/Snail, NHEK/ΔNp63α-M, NHOF, SCAP, and DPSC. B, photomicrographs of ALP staining are shown for each tested group. C, NHEK/Snail and NHEK/ΔNp63α-M were cultured in the basal or calcifying medium to induce mineralization and stained with Alizarin red. Original magnification was ×100. D, RT-qPCR was performed for the osteo-/odontogenic differentiation markers (BSP, ColIα, DMP-1, DSPP, ALP, OC, and ON) in DPSC, NHEK/Snail, and NHEK/ΔNp63α-M. E, Western blotting was performed for DSP in NHEK (parent), NHEK/Snail, NHEK/ΔNp63α-M, and DPSC cultured for 7 days in the basal (−) or calcifying (+) medium. The blot yielded multiple DSP signals at the ∼230, 95, and 38 kDa marks. GAPDH was used as a loading control.

DISCUSSION

The current study reports a novel function of ΔNp63α in transforming NHEK to gain a stem cell-like phenotype through EMT. This finding was obtained by ectopic expression of ΔNp63α in primary human keratinocytes, whereas parallel experiments using TAp63α showed no such effect. Occurrence of EMT in ΔNp63α-transduced cells was assessed by the appearance of mesenchymal shape and enhanced migratory capacity and altered expression of epithelial or EMT markers. These cells also exhibited stemness and multipotent capacity as demonstrated by CD24^low/CD44^high, Hoechst efflux assay, stem cell markers, and osteo-/odontogenic and adipogenic differentiation. Interestingly, EMT conversion in NHEK by ΔNp63α led to telomere length extension in the absence of telomerase activation. This observation may be viewed as an appearance of the EMT phenotype because mesenchymal cells generally have longer telomeres than do epithelial cells (43). Because telomere length appeared to be elongated in NHEK/ΔNp63α-M cells compared with that of parental NHEK in the absence of telomerase activation, EMT may have occurred selectively in cells that possessed longer telomeres, such as undifferentiated keratinocyte stem cells.

Our data indicate that NHEK/ΔNp63α-M cells are capable of differentiation into adipogenic cells, and we ruled out the possibility of contamination of sebocytes (supplemental Fig. S2). Likewise, it is possible that primary NHEK cultures are contaminated with fibroblasts. This possibility is ruled out in the current study because NHEK/ΔNp63α-M cells demonstrate different phenotype and gene/protein expression profile compared with fibroblasts. Multipotency (osteo-/odontogenic and adipogenic differentiation) was shown in NHEK/ΔNp63α-M cells but not in NHOF (Fig. 8 and supplemental Fig. S4). NHEK/ΔNp63α-M cells express higher levels of Nanog/Lin28 compared with NHOF (Fig. 7C). We also demonstrate that EMT cells transduced with ΔNp63α continue to express the α-isoform of ΔNp63, whereas NHOF completely lack p63 expression. Furthermore, we only observe EMT in ΔNp63α-transduced NHEK and not in the control (LXSN), whereas both cultures were derived from the same primary culture.

p63 regulates proliferation and differentiation of mature keratinocytes in normal tissue homeostasis as well as re-epithelialization in cutaneous wound healing (44, 45). Wound re-epithelialization is mediated by Slug, which is required for EMT (46). EMT is involved in skin health and disease, including malignant cancers (47). A recent study showed that ΔNp63α may cooperate with oncogenic Ras to promote tumor-initiating stemlike proliferation in skin keratinocytes (48). A high level of p63 is associated with the stem cell compartment of corneal epithelia (49). Also, p63 expression is elevated in a subset of head/neck squamous cell carcinomas (SCC) that show aggressive phenotype and poor prognosis (50). Similarly, Massion et al. (51) reported frequent amplification and overexpression of p63 in lung SCC compared with normal counterpart, suggesting its involvement in lung tumorigenesis. The current study may shed light on the mechanism by which p63 may regulate epithelial regeneration and contribute to carcinogenesis upon its aberrant expression through induction of EMT. As such, enhanced p63 expression is observed in physiological and diseased conditions, although detailed study is necessary to prescribe the mechanistic role of p63 overexpression in such biological processes.

ΔNp63α-mediated EMT appeared only in senescing cultures even if the gene was transduced in a very young culture. This finding suggested possible involvement of TGF-β in ΔNp63α-mediated EMT because endogenous TGF-β production is highly increased in senescing culture of NHOK (36). The addition of exogenous TGF-β accelerated the occurrence of EMT in the ΔNp63α-transduced cells, whereas senescence was induced in control NHEK. Hence, ΔNp63α altered cellular response to TGF-β. Several studies reported that TGF-β alone is capable of triggering EMT in human epithelial cells, such as ovarian surface epithelium and HMLE (16, 52). In the current study, NHEK underwent premature senescence and terminal differentiation when exposed to TGF-β (supplemental Fig. S3). Non-proliferative and differentiation states were maintained in cells even after 14 days of TGF-β treatment. Likewise, TGF-β alone failed to induce EMT in HaCaT cells, spontaneously immortalized human keratinocytes, and led to proliferation arrest (15). Therefore, ΔNp63α transduction appears to be required for EMT in NHEK exposed to exogenous TGF-β, whereas inhibition of TGF-β signaling prevented or reversed EMT triggered in NHEK/ΔNp63α.

Earlier studies showed that p63 is required for epithelial morphogenesis and limb development (4, 5). Both prior studies demonstrate that mice lacking p63 die prenatally due to lack of epithelial structure development, such as skin, prostate, and breast. This finding raised the possibility that p63 may be involved in epithelial cell fate determination. Our data confirmed that ΔNp63 expression was epithelial cell-specific because its expression was completely absent in NHOF. ΔNp63 protein expression in NHEK was mainly detected near 63 and 75 kDa, and the ΔNp63α isoform corresponded to the band located at 63 kDa (Fig. 1A). Upon transduction of cells with ΔNp63α, endogenous ΔNp63 protein signal at 75 kDa was notably diminished. Also, when EMT was induced in NHEK/ΔNp63α cells by TGF-β exposure, the ΔNp63 signal at 75 kDa completely vanished, and the protein expression was restored when the cells regained their epithelial phenotype after exposure to TβRI-i (Fig. 5B). It is possible that the ΔNp63 protein migrating at 75 kDa plays a crucial role in maintaining the epithelial cell identity and may be targeted by exogenous ΔNp63α to induce EMT. Recently, Lindsay et al. (17) demonstrated an effect of ΔNp63γ on EMT in immortal breast epithelial cell line, MCF10A. In this prior study, overexpression of ΔNp63γ led to EMT only when both ΔNp63α and ΔNp63β were depleted, and no phenotypic change was induced by ΔNp63γ transduction alone. In the current study, overexpression of ΔNp63α led to EMT in NHEK in a TGF-β-dependent manner. The difference between the two studies may stem from cell type specificity (i.e. primary keratinocytes versus immortal mammary epithelial cell line). These findings and our current data indicate that there are clear mechanistic differences by which p63 regulates epithelial cell fate in breast epithelial cells and skin keratinocytes, and ΔNp63 isoforms do possess distinct roles in regulating EMT. Further research is necessary to delineate the detailed mechanisms underlying this phenomenon.

In addition, Davis et al. (15) reported EMT in human immortal and malignant keratinocytes by TGF-β1. In this prior study, TGF-β1-mediated EMT required Ras activation, and TGF-β1 caused G₁ cell cycle arrest in cells with or without active Ras. Also, continuous exposure of transformed cells to TGF-β was required to maintain the EMT phenotype; otherwise, the cells spontaneously reverted to the epithelial phenotype. On the contrary, NHEK/ΔNp63α-M cells replicated in the presence of exogenous TGF-β and did not require continuous exposure to TGF-β to maintain their EMT phenotype. Both active Ras and MAPK pathway were found to be important for EMT in cells exposed to TGF-β (15), but ΔNp63α did not activate either pathway in NHEK (data not shown). EMT was induced in these cells simply by serial subculture in the absence of exogenous TGF-β, through accumulation of endogenous TGF-β. Reversion to the epithelial phenotype was not spontaneous in NHEK/ΔNp63α-M cells but occurred after exposure to TβRI-i. We recently showed that Bmi-1 inhibits the intracellular TGF-β signaling (36). Accordingly, ΔNp63α transduction did not cause EMT in cells expressing exogenous Bmi-1, regardless of the order of gene transduction between Bmi-1 and ΔNp63α. This is in contrast to a recent study showing that Bmi-1 alone is capable of inducing EMT in nasopharyngeal epithelial cells (53). Long term culture of NHOK was maintained after Bmi-1 transduction, but no sign of EMT was ever noted, even during cellular immortalization with Bmi-1 and human papillomavirus E6 (40). These data indicate that cells do respond differently to Bmi-1 transduction based on tissue origin.

The current study is the first demonstration that NHEK gained the stem cell-like phenotype through ΔNp63α-mediated EMT. Recent studies have reported the acquisition of stem cell properties in HMLE, which is a SV40-immortalized human mammary epithelial cell line, through EMT (16, 22). In these prior reports, EMT was induced by overexpression of Twist, Snail, or TGF-β1 in HMLE using retroviral vectors, and the transformed cells exhibited characteristics similar to MSCs. NHEK/ΔNp63α-M cells also demonstrated a phenotype resembling MSCs, such as changes in the CD24/CD44 surface marker profile, expression of osteo-/odontogenic markers, ALP activity, and mineralization. These cells expressed a markedly enhanced level of Nanog and Lin28, both of which are involved in nuclear reprogramming and acquisition of pluripotency (54). We checked if other factors could induce EMT and stem cell properties in NHEK. Twist was constitutively expressed in parental NHEK at a high level, and its expression was not altered in cells after the occurrence of EMT (Fig. 2B). Exogenous Snail expression induced EMT in NHEK, as judged by reduced E-Cad expression and morphological changes. However, the Snail-transduced cells failed to express ALP activity or mineralize in vitro, suggesting that EMT induced by Snail was not sufficient for the stem cell-like phenotype. In fact, EMT appears to be only partially induced by Snail transduction in NHEK. In this regard, ΔNp63α-mediated EMT appears to be the unique mechanism to confer stem cell-like properties in NHEK. Our current study raises a possibility that NHEK/ΔNp63α-M cells can be used for regenerative therapies due to their stem cell-like properties. These cells and other primary human epithelial cells exhibiting EMT may be described by a new term, “induced MSCs” (iMSCs), to reflect their phenotypic resemblance to native MSCs. Because NHEK can be readily obtained from human skin, iMSCs derived from NHEK may be an alternate and ample source for stem cell-based therapies. Further research will elucidate the regenerative potential of iMSCs and their therapeutic benefit.