EpCAM Induction Functionally Links to the Wnt-Enhanced Cell Proliferation in Human Keratinocytes

Abstract

Accelerating proliferation of primary keratinocytes benefits skin autografts for severely burned patients. Wnt signal, a conserved pathway controlling cell cycle and morphogenesis in embryo, also involves in cell proliferation and tumorigenesis in adult tissues. Here the effects of Wnt signal on the growth of human interfollicular keratinocytes were investigated. We demonstrated that recombinant Wnt3a significantly promoted the growth of primary keratinocytes at a low cell density. A well-characterized GSK-3b inhibitor, BIO, activated the Wnt signals and also enhanced the colony formation of keratinocytes dose dependently. Gene expression profile of the BIO-treated keratinocytes revealed the linkage of BIO with cell mitosis and indicated that epithelial cell adhesion molecule (EpCAM), a Wnt target gene, was significantly upregulated. Compared to the sorted EpCAM^- keratinocytes, the EpCAM⁺ cells showed a higher proliferation rate and efficacy of colony formation. Inhibiting the EpCAM expression by shRNA attenuated the proliferation effect of BIO and the growth advantage of the EpCAM⁺ keratinocytes. These evidences emphasize the positive roles of canonical Wnt and EpCAM on the regulation of cell growth and self-renewal of human keratinocytes.

Introduction

Proliferative keratinocytes mainly reside in the basal layer next to the basement membrane and provide a continuous cell source for skin homeostasis. The primary human keratinocytes derived from non-hair-bearing tissues, such as foreskin, are classified as interfollicular keratinocytes. The epidermal progenitor cells (EpPCs) reside in interfollicular basal layers, hair bulge, or sebaceous glands and act as a permanent reservoir to sustain the homeostasis of keratinocytes and epidermal turnover throughout life (43). The proliferative potential and the stemness of human interfollicular EpPCs are mainly reflected by in vitro clonal growth assays (22). For instance, when isolated keratinocytes are subcloned at a low cell density, human EpPCs derived from neonatal skin give rise to large circular colonies, termed “holoclones” (4,26). In contrast, small irregularly shaped clones, termed “paraclones,” are generated by putative transit amplifying (TA) keratinocytes (4,26). The TA cells are the undifferentiated keratinocytes that detach from the basement membrane in epidermis and only have a limited capacity of self-renewal and proliferation (31).

To meet the clinical requirement of skin autografts for burn patients, efficient scaling up of primary human keratinocytes and isolating the EpPCs have been persistently pursued. Previous studies have identified that interfollicular EpPCs express high levels of α6 (37) and b1 integrins (12). The sorted b1^high keratinocytes give rise to holoclones, whereas the β1^low cells are prone to produce paraclones (11). In addition, EpPCs can be enriched from primary keratinocytes by sorting a subpopulation with a high expression of α6 integrin and low expression of cluster of differentiation 71 (CD71) (18). Moreover, both melanoma chondroitin sulfate proteoglycan (MCSP) (6,15) and the epidermal growth factor receptor antagonist leucine-rich repeats and immunoglobulin-like domains 1 (Lrig1) (10) have been indicated as markers of quiescent interfollicular EpPCs.

Although the above markers have helped advance our understanding and the isolating of EpPCs from the epidermis, a definite molecular signature of human EpPCs and their interfollicular niche is still largely unclear. In mice, genetic approaches and cell-tracing experiments have elucidated the niche of hair bulge and their resided stem cells. Among the molecular regulators in the early hair induction, wingless-type mouse mammary tumor virus (MMTV) integration site family (Wnt)/β-catenin signaling is substantial in the specification and the activation of hair morphogenesis (1,40,44). During the adult hair cycle, activation of Wnt signaling drives resting stem cells in the bulge to enter a proliferative stage and initiates the growth of a new hair. Notably, stabilizing β-catenin in interfollicular keratinocytes leads to de novo hair formation independent of the stem cells in the hair bulge of mice (3), highlighting the substantial role of Wnt in both the niche of hair bulge and the proliferation of interfollicular EpPCs. Moreover, the importance of Wnt signal in the growth of keratinocytes is further supported by several pathological findings in humans. For example, elevated Wnt-related molecules have been detected in the hyperproliferated keratinocytes in psoriasis patients (7,8,30). The activation of Wnt/β-catenin signal was also shown to be essential for maintaining the stemness of cancer stem cells in squamous cell carcinoma (21,35).

Nevertheless, inconsistent findings have also been reported regarding the role of Wnt on the proliferation of primary human keratinocytes. A recent study indicated that applied recombinant human Wnt3a or Wnt5a did not affect the growth or the migration of the keratinocytes (7). Here to clarify the role of the Wnt signal in human keratinocytes, a recombinant purified Wnt3a and its small molecule agonist, 6-bromoindirubin-3×-oxime (BIO), were applied on foreskin-derived primary keratinocytes. Wnt-steered growth promotion and enhanced colony formation were observed at a low cell density, but not at a high cell density of culture. A Wnt downstream molecule, epithelial cell adhesion molecule (EpCAM), was dramatically upregulated in the BIO-treated keratinocytes. Interestingly, sorted EpCAM⁺ cells showed enhanced growth rate and clonal expansion, suggesting that EpCAM may serve as a new surface marker for human EpPCs.

Materials and Methods

Cell Culture and Reagents

The method of epidermis separation from foreskin and subsequent keratinocyte isolation is well documented (18). The use of human foreskin for isolating keratinocytes was reviewed and approved by Institutional Review Board (IRB) of Veteran General Hospital in Kaohsiung City, Taiwan (VGHKS99-CT3-05). Fifty adult skin tissues were collected with informed consent following ethical and institutional guidelines. The isolated keratinocytes were cultured in keratinocyte serum-free (KSF) medium (Invitrogen, Carlsbad, CA, USA), supplemented with a kit containing 2 mM l-glutamate, 25 μg/ml bovine pituitary extract (Invitrogen), and 0.2 ng/ml human recombinant epidermal growth factor (hrEGF) (Invitrogen). The keratinocytes were passaged with 0.25% trypsinethylenediaminetetraacetic acid (EDTA) (Invitrogen) when the cells reached 75% confluence on culture plates (Corning, Corning, NY, USA). The cells before third passages were used in this study. The media was supplemented with 50 or 100 ng/ml human recombinant Wnt3a (R&D Systems, Minneapolis, MN, USA), 1 nM bone morphogenetic protein 4 (BMP4, R&D Systems), 0.1, 0.2, or 0.5 μM BIO (Merck-Calbiochem, Whitehouse Station, NJ, USA), or 0.5 μM 1-methylBIO (MeBIO; Merck-Calbiochem) every 2 days as required; the vehicle for BIO and MeBIO was dimethyl sulfoxide (DMSO) and PBS for others. Human female embryonic kidney 293T (HEK293T; Food Industry Research and Development Institute, Hsinchu, Taiwan) cells and female breast cancer MCF7 cells (Food Industry Research and Development Institute) were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen), 100 U/ml penicillin, and 100 μg/ml streptomycin. Cells were maintained at 37°C in 5% CO₂. All chemicals were obtained from Sigma-Aldrich (St. Louis, MO, USA) unless noted elsewhere. Cell enumeration of suspended keratinocytes was estimated from triplicate independent samples by using a hemocytometer (Paul Marienfed, Lauda-Koenigshofen, Baden-Württemberg, Germany).

Immunocytostaining

Cells were fixed in 4% cold paraformaldehyde and permeabilized with 0.3% Triton X-100 in phosphate-buffered saline (PBS). After treatment with blocking solution (2% horse serum, 0.2% Triton X-100 in PBS) for 30 min, immunocytochemistry was performed using the primary antibodies, including involucrin (1: 1,000; Sigma-Aldrich), Ki67 (1:100; Dako, Glostrup, Denmark), or EpCAM (11032:1032200; clone HEA-125, Miltenyi Biotec, Bergisch Gladbach, Germany). Cells were washed twice with 0.1% Tween-20 in PBS to remove unconjugated primary antibody. Appropriate fluorescence-tagged secondary antibodies (1:3,000; all from Jackson ImmunoResearch, West Grove, PA, USA) were used for visualization. DAPI (4×,6-diamidino-2-phenylindole) was used for nuclear counterstaining. Images of immunostaining were captured using an upright microscope (Nikon ECLIPSE 80I; Tokyo, Japan) or confocal microscope (LSM510 Meta, Zeiss, Jena, Germany).

Luciferase Assays

Top-flash plasmid (Millipore, Billerica, MA, USA), a Wnt reporter plasmid containing repeated transcription factor 3 (TCF3)-binding sites and a downstream luciferase gene, was applied in this study. The plasmid was transiently transfected into the keratinocytes by Lipofectamin-2000 (Invitrogen). At 16 h posttransfection, the keratinocytes were further treated with recombinant proteins or BIO derivatives for 24 h as previously described. Total cell lysates were collected and subjected to the dual luciferase assay system (Promega, Madison, WI, USA). Protein concentration was determined by Bradford method (Bio-Rad protein assay) (Bio-Rad, Hercules, CA, USA). Luciferase activity in transfected cells was measured by Beckman-Paradigm machine (Beckman-Coulter, Pasadena, CA, USA) and analyzed by its installed software (Multimode detection software; Beckman-Coulter).

cDNA Microarray

Total RNA was extracted using TRIzol C&T (Protech, Taipei, Taiwan) from the keratinocytes treated with DMSO-treated (mock) and 0.1 μM BIO for 6 days. The quality of RNA was carefully evaluated by electrophoresis and Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). Twenty to 50 μg of mRNA from each sample was used in each array (Human Gene 2.0 ST Array, Affymetrix, Santa Clara, CA, USA) by the Affymetrix Gene Expression Service Lab at Academic Sinica (Taipei, Taiwan; http://ipmb.sinica.edu.tw/affy). The microarray images were scanned and digitized using an Affymetrix scanner (GeneChip scanner 3000). The data were analyzed with Microarray Suite version 5.0 (MAS 5.0) using Affymetrix default analysis settings and global scaling as normalization method. The trimmed mean target intensity of each array was arbitrarily set to 500. The array data were deposited to Gene Expression Omnibus (GEO) repository (http://www.ncbi.nlm.nih.gov/geo/; Serial No. GSE40233).

BrdU Incorporation Assay

Cell proliferation was assessed via a 5-bromo-2×-deoxyuridine (BrdU) incorporation assay. BrdU (100 μg/ml) was incubated with the cells for 1 h or 16 h at 37°C. After washing with PBS three times, cells were fixed in 4% formaldehyde for 5 min. The cells were stained with anti-BrdU antibody (1:1,000; clone MoBU, Invitrogen) 4°C overnight. The BrdU signal was enhanced by secondary fluorescein isothiocyanate (FITC) conjugated anti-mouse immunoglobin antibody (1:3,000; Invitrogen). Cell nuclei were stained with DAPI for 5 min.

Cell Cycle Analysis

For cell cycle analysis, cells were fixed overnight in ice-cold 70% ethanol at 4°C. After washout of the ethanol, cell pellets were suspended in 5 ml PBS. The keratinocytes were treated with 20 μg/ml of RNase A (Roche Applied Science, Indianapolis, IN, USA) for 30 min at 25°C to remove intracellular RNAs. Nuclear DNA was stained with 1 μg/ml propidium iodide (Fluka-Sigma-Aldrich). At least 10,000 to 20,000 events were counted during data collection using Influx cell sorter (Becton-Dickinson, Franklin Lakes, NJ, USA), and the data were analyzed by FlowJo software (Treestar, Ashland, OR, USA).

Colony Formation

Human keratinocytes were seeded at a density of 10³ cells/cm² on either six-well plates (Corning) or 10-cm culture dishes (Corning). These cells were cultured for 8–10 days in KSF medium at 37°C. The number of colonies was manually recorded after fixing with 4% paraformaldehyde and staining with 0.1% crystal violet in 10% formalin. In addition, to estimate the total surviving cells, the crystal violet that resided in cells was extracted with acetic acid and quantified with a spectrophotometer using a 595-nm filter. Additionally, the area of survived cells on culture plates was also calculated using Image-Pro Plus software (Media Cybernetics, Rockville, MD, USA).

Flow Cytometry and Cell Sorting

To mark EpCAM⁺ keratinocytes, 5 μg/ml anti-human EpCAM (Miltenyi Biotec) was added and incubated with the isolated keratinocytes for 40 min at 37°C in the dark. After twice washing with PBS, the EpCAM signal was enhanced by adding FITC-conjugated donkey anti-mouse IgG antibody for 40 min at 37°C (Invitrogen). The individual EpCAM⁺ and EpCAM^- cells were sorted using the Influx cell sorter (Becton Dickinson).

Recombinant Lentivirus Production and Infection

Four short hairpin RNA (shRNA) clones for human EpCAM were obtained from the RNAiCore at Academia Sinica, including TRCN0000073733 (clone 1), TRCN0000073734 (clone 2), TRCN0000073736 (clone 3), and TRCN0000073737 (clone 4). Recombinant lentiviruses were generated by transfection in HEK293T cells with pLKO.1-shRNA, psPAX2, and PMD2G triple plasmids (RNAiCore) by Lipofectamine 2000 (Invitrogen). The cell supernatant was collected and concentrated by ultracentrifugation at 35,000 × g in a swinging bucket rotor for 60 min at 4°C. The viral pellet was resuspended in DMEM (Invitrogen). Final viral titer was determined by the cell viability assay as suggested by the RNAiCore protocol using puromycin selection. Infection of recombinant lentivirus in the keratinocytes was aided with 8 μg/ml polybrene (Sigma-Aldrich) at indicated multiplicity of infection (MOI). Noninfected viruses were removed by changes of the culture media at 16 h postinfection.

Western Blot Analysis

Infected cells with individual lentivirus-shRNA were lysed with radioimmunoprecipitation assay (RIPA) buffer [50 mM Tris pH 7.5, 150 mM NaCl, 10 mM EDTA, 1% Tergitol-type NP-40, 0.1% sodium dodecyl sulfate (SDS)] plus a cocktail of proteinase inhibitors (Sigma-Aldrich). Denatured proteins were separated by 10% SDS-polyacrylamide gel electrophoresis (PAGE; Sigma-Aldrich) and then transferred to polyvinylidene difluoride (PVDF) membranes (GE-Amersham, Pittsburgh, PA, USA). Samples were detected with antibodies to EpCAM (1:200; clone HEA-125, Miltenyi Biotec) and glyceraldehyde 3-phosphate dehydrogenase (1:1,000; GAPDH; Novus Biologicals, Litteton, CO, USA). The EpCAM immunoreactive bands were revealed using secondary horseradish peroxidase-conjugated antibodies (1:3,000; Jackson ImmunoResearch) and enhanced chemiluminescence (ECL) reagents (GE-Amersham, Pittsburgh, PA, USA). The intensity of the chemiluminescence was analyzed by Quantity One software (version 4.6.1, Bio-Rad).

Statistical Analysis

One-way analysis of variance (ANOVA) was performed to determine significant differences from at least two independent experiments with the Tukey post hoc test. Statistical significance was established at p<0.05.

Results

Recombinant Wnt3a Promoted the Growth of Keratinocytes

The primary human keratinocytes derived from adult human foreskins were cultured at a low cell density, 10³ cells/cm², on 10-cm culture dishes. Compared to the mock treatment, 50 ng/ml and 100 ng/ml Wnt3a treatment significantly increased the cell numbers by 1.5- and 2.3-fold on culture day 8, respectively (p<0.05) (Fig. 1A). Wnt3a also reduced the expression of involucrin, a marker of mature keratinocyte (Fig. 1B, C), indicating that the canonical Wnt signal sustains the self-renewal capacity of human keratinocytes. In contrast, treating with 1 nM recombinant BMP4 for 8 days increased the involucrin expression of the culture cells (p<0.01) (Fig. 1D, E).

Figure 1.

Recombinant Wnt3a promoted the cell growth of human primary keratinocytes. The primary keratinocytes were treated with human wingless-type mouse mammary tumor virus (MMTV) integration site family member 3a (Wnt3a) protein in 10-cm culture dishes at a cell density of 10³ cells/cm². The cell numbers of the keratinocytes were estimated from independent triplicate experiments by using a hemocytometer (A). After adding phosphate-buffered saline (PBS) (mock, B), 1 nM bone morphogenetic protein 4 (BMP4) (C), or 100 ng/ml Wnt3a (D) for 8 days, the matured keratinocytes were detected by the staining of involucrin and the ratio of involucrin⁺ cells to total cells was calculated from three independent experiments (E). The cell growth of primary keratinocytes was also estimated at a higher cell density (10⁴ cells/cm²) for 8 days in the presence of Wnt3a or BMP4 (F). Scale bar: 50 μm. The statistics were analyzed by one-way ANOVA. *p < 0.05.

Interestingly, this growth enhancement effect of Wnt3a on keratinocytes was not observed when the culture density was elevated 10-fold (Fig. 1F). Adding 50 or 100 ng/ml Wnt3a did not significantly increase the cell number on culture day 8. Treatment with 1 nM BMP4 significantly reduced the cell growth of primary keratinocytes (Fig. 1F). These results indicated that the Wnt-mediated cell growth in keratinocytes is cell density dependent.

BIO Enhanced the Clonal Formation and Expansion of Keratinocytes

The activation of canonical Wnt signal in keratinocytes was recapitulated by applying the small molecule BIO, an indirubin derivative showing glycogen synthase kinase (GSK)-3β phosphorylation inhibition (23). Revealed by the TCF3-driven luciferase reporter, nuclear Wnt signals in 0.1 and 0.2 μM BIO-treated keratinocytes were enhanced 19.1- and 26.0-fold, respectively (Fig. 2A). The inactive BIO analog compound, MeBIO, only moderately activated the TCF3 promoter activity and showed much less potency than BIO (Fig. 2A). Notably, this Wnt activation effect was not observed in the presence of 0.5 μM BIO. We found that BIO concentrations higher than 0.5 μM resulted in considerable cell apoptosis and intracellular caspase-3 activation in primary keratinocytes (data not shown).

Figure 2.

BIO compound enhanced clonal formation of human primary keratinocytes. The activated Wnt signal was determined by the transcription factor (TCF) promoter of top-flash plasmid in the 6-bromoindirubin-3×-oxime (BIO)- or 1-methyl-BIO (MeBIO)-treated keratinocytes, comparing to that of mock-treated control (DMSO) (A). The growth kinetics of BIO- or MeBIO-treated cells were plotted according to the estimated cell numbers within 8 days (B). The cells with 0.1 μM or 0.2 μM BIO treatment were trypsinized and counted on day 8 (C). The colonies of mock-, BIO-, and MeBIO-treated keratinocytes in six-wells/plates were visualized by the staining of crystal violet (D). The counting of colony number of the treated keratinocytes per well was performed on day 10 culture (E). The represented cell morphologies of colonies were shown for the primary keratinocytes treated with mock (DMSO) (F), 0.5 μM MeBIO (G), 0.1 μM BIO (H), and 0.2 μM BIO (I), respectively. The statistics were analyzed by one-way ANOVA. *p < 0.05.

Similar to the growth enhancement by recombinant Wnt3a, 0.1 μM BIO promoted the proliferation of the keratinocytes at a density of 10³ cells/cm², whereas this growth-enhancing effect was not observed in MeBIO-treated cells (Fig. 2B). We further demonstrated that the pro-cell growth effect of BIO was dose dependent (Fig. 2C). These evidences strongly indicated the specificity of BIO-steered Wnt activation in the regulation of keratinocyte growth.

In addition to the total cell numbers, the efficacy of clonal expansion of BIO-treated keratinocytes was also evaluated. We discovered that 0.1 μM BIO dramatically promoted the colony formation of keratinocytes (5.6-fold). However, 0.5 μM MeBIO only moderately enhanced the colony numbers (2.1-fold) (Fig. 2D, E). The mock and MeBIO-treated colonies exhibited irregular paraclone shapes, and the cells in these colonies were more dispersed (Fig. 2F, G). In contrast, BIO-treated colonies were circular and compact, which are classical holoclone characteristics (Fig. 2H, I).

Wnt Signal Accelerated the Cell Cycle of Keratinocytes

Comparison of gene expression profiles between mock and BIO-treated primary keratinocytes revealed that gene families involving proteolysis–ubiquitination, transcription activation, and cell mitosis were significantly upregulated (Table 1 and Fig. 3). Regarding cell cycle control, we observed that several genes related to growth-promoting factors, such as epithelial mitogen (epgn), feline encephalitis virus-related tyrosine kinase (fer), rat sarcoma viral oncogene homolog (ras), v-yes-1 Yamaguchi sarcoma viral oncogene homolog 1 (yes1), and v-ets avian erythroblastosis virus E26 oncogene homolog 1 (ets1), were highly induced (Table 1). In addition, some genes of transcriptional activators, such as replication factor C (activator 1) 1, 145 kDa (rfc1), eukaryotic translation initiation factor 2A, 65 kDa (eif2a), TAF2 RNA polymerase II, thymine–adenine–thymine adenine (TATA) box-binding protein (TBP)-associated factor, 150 kDa (taf2), and activating transcription factor 2 (atf2), were highly expressed in BIO-treated keratinocytes (Table 1). Notably, several adhesion molecules were overexpressed in the BIO-treated cells, such as activated leukocyte cell adhesion molecule (ALCAM) and EpCAM (Table 1). α6 Integrin, a well-characterized EpPC surface molecule (14), was also enhanced by the BIO treatment. These surface proteins enhanced by Wnt activation could be candidate markers for isolating putative EpPCs by flow cytometry.

Figure 3.

The activated signal pathways in BIO-treated keratinocytes. Comparative data of whole genome cDNA microarray were mapped according to the fitness to the annotated processes in the GeneGo data bank (server: portal.genogo.com). The bars with different lengths correspond to the significance of upregulated signal pathways in BIO-treated keratinocytes.

Table 1.

BIO-Steered Upregulated Genes in Human Keratinocytes

Fold	Gene	mRNA Accession	Description
7.92	EPGN	NM_001013442	Epithelial mitogen homolog (mouse)
7.06	FER	NM_005246	fer (fps/fes related) tyrosine kinase
5.59	RASA1	NM_002890	RAS p21 protein activator (GTPase activating protein) 1
5.34	YES1	NM_005433	v-yes-1 Yamaguchi sarcoma viral oncogene homolog 1
5.34	SLK	NM_014720	STE20-like kinase (yeast)
5.27	BRCC3	NM_024332	BRCA1/BRCA2-containing complex, subunit 3
4.95	ERBB2IP	NM_018695	erbb2 interacting protein, transcript variant 2
4.89	RFC1	NM_002913	Replication factor C (activator 1) 1, 145kDa
4.87	EIF2A	NM_032025	Eukaryotic translation initiation factor 2A, 65kDa
4.63	MAPK6	NM_002748	Mitogen-activated protein kinase 6
4.19	TAF2	NM_003184	TAF2 RNA polymerase II
4.07	ETS1	NM_001143820	v-ets erythroblastosis virus E26 oncogene homolog 1
3.90	SOS1	NM_005633	Son of sevenless homolog 1 (Drosophila)
3.72	ATF2	NM_001880	Activating transcription factor 2
5.10	ALCAM	NM_001627	Activated leukocyte cell adhesion molecule
4.23	EPCAM	NM_002354	Epithelial cell adhesion molecule
4.15	ITGA2	NM_002203	Integrin, α 2
5.04	ITGB6	NM_000888	Integrin, β 6
3.11	ITGAV	NM_002210	Integrin, α V
2.64	ITGA6	NM_000210	Integrin, α 6
3.68	FZD6	NM_003506	Frizzled homolog 6 (Drosophila)
3.34	ARMC1	NM_018120	Armadillo repeat containing 1
2.36	ADAM17	NM_003183	ADAM metallopeptidase domain 17

BIO, 6-bromoindirubin-3×-oxime.

EpCAM Represented a New Surface Marker for EpPCs

The elevated expression of EpCAM, an identified downstream target of Wnt (42), was especially examined (Table 1). The overexpressed EpCAM transcripts were further validated by the immunocytostaining on the cultured keratinocytes after the treatment of DMSO (mock), MeBIO, or BIO (Fig. 4A, B, C). In addition, flow cytometry analysis quantitatively revealed that the EpCAM positivity was 20.1 ± 4.2% in mock-treated keratinocytes, while 62.5 ± 6.8% in BIO-treated keratinocytes (Fig. 4D, E, F).

Figure 4.

EpCAM expression was induced by the BIO treatment. The surface epithelial cell adhesion molecule (EpCAM) expression was detected in living keratinocytes after treatment of mock (DMSO, A), 0.5 μM MeBIO (B), and 0.1 μM BIO (C) for 6 days. Quantitative examination of the EpCAM expression was conducted using flow cytometry (D, E, F). FITC, fluorescein isothiocyanate; SSC, side scatter.

To examine the proliferation potency of EpCAM-expressing keratinocytes, flow cytometric sorting was applied to collect the individual EpCAM⁺ and EpCAM^- cells at a purity higher than 90% (Fig. 5A, B). The 5 × 10⁴ sorted cells were grown on 10-cm culture dishes (10³ cells/cm²). After 10 days of culture, expanded EpCAM⁺ cells reached about 1 × 10⁶ cells, ca. twofold increase in cell population of EpCAM^- cells (Fig. 5C). This proliferation advantage was further supported by the higher incorporation of BrdU in the nuclei of EpCAM⁺ cells after BrdU labeling for 1 h or 16 h (Fig. 5D, E). Moreover, cell cycle analysis also consistently revealed that the ratio of G₂/M phase for the EpCAM⁺ population was higher than that of the EpCAM^- cells (Fig. 5F). Interestingly, we found that the ratios of Ki67-expressing cells in both subpopulation remained largely unchanged (Fig. 5G). These results indicated that generally both EpCAM⁺ and EpCAM^- cells were proliferative, and the high BrdU⁺ ratio of EpCAM⁺ cells (Fig. 5D, E, lanes 1 and 2) was mainly due to the rapid cell mitosis in EpCAM⁺ cells, instead of the prodifferentiation property of EpCAM^- cells. Importantly, the BIO-steered EpCAM⁺ population showed high proliferation potency as the EpCAM⁺ cells from mock treatment (Fig. 5C, D, E, lane 3 vs. lane 1), indicating that BIO-mediated growth enhancement (Fig. 2) may be contributed to by the induced hyperproliferative EpCAM⁺ cells.

Figure 5.

EpCAM represented a new marker of EpPCs. According to the expression intensity of EpCAM, subpopulation of primary keratinocytes was sorted by flow cytometry (A, B). Sorted cells were cultured, and the individual cell numbers were calculated on culture day 10 (C). The mitotic cells, labeled with 5-bromo-2×-deoxyuridine (BrdU) for 1 h (D) or 16 h (E), were manually calculated under fluorescent microscope. Total cells were revealed by the number of 4×,6-diamidino-2-phenylindole positive (DAPI⁺) nuclei. Cell cycle distribution in the sorted cells was revealed by the nuclear intensity of propidium iodide and analyzed by FlowJo software (F). The Ki67⁺ cells indicated the proliferative cells (G). All the data represented three independent experiments, and their statistics were analyzed by one-way ANOVA. *p < 0.05.

EpCAM Was Essential for the Maintenance of Keratinocyte Stemness

To further elucidate the role of EpCAM in regulating the growth of keratinocytes, lentiviruses carrying EpCAM shRNA were applied to inhibit the EpCAM expression. We evaluated the shRNA inhibition efficiency in MCF7 cells (Fig. 6A, lane 1) because relative low expression of EpCAM was detected in primary keratinocytes (Fig. 6A, lane 2). Among the four shRNA-carrying lentiviral clones, clones 2, 3, and 4 reduced ca. 75%, 60%, and 85% EpCAM expression in MCF7 cells at MOI 5, respectively, but clone 1 (shRNA1) failed to downregulate the EpCAM expression (Fig. 6A, B). This nonspecific shRNA1 lentivirus (LV-shRNA1) was used as the control virus for the following experiments. We found that after LV-shRNA4 infection at MOI 5, the clonal expanding capacity of keratinocytes was significantly reduced under both mock and BIO treatment conditions, compared to the LV-shRNA1-infected cells (Fig. 6C, D, E). These results strongly suggested that the BIO-mediated growth enhancement was mediated, at least in part, by the EpCAM expression.

Figure 6.

EpCAM was essential for the stemness of keratinocytes. The endogenous EpCAM expression in breast cancer MCF7 cells (A, lane 1) and human primary keratinocytes (A, lane 2) was examined by Western blot. The EpCAM inhibition efficacy of four shRNA lentiviruses was tested in the infected MCF7 cells at MOI 5 (A, lanes 3–6). Quantitative analysis of EpCAM expression was referred from the relative optical intensity in (A) after normalization with constitutive glyceraldehyde 3-phosphate dehydrogenase (GAPDH) expression (B). Distribution area of cultured keratinocytes on culture plates was visualized by crystal violet staining in the presence of DMSO (mock) or 0.1 mM BIO (C). The area of cells (D), amount of incorporated crystal violet (E), and the total cell numbers on culture day 10 (F) were estimated and compared among the lentiviral infected cells with different shRNA clones. All the data represented two independent experiments, and their statistics were analyzed by one-way ANOVA. *p < 0.05.

In addition, after the treatment of 0.1 mM BIO, both EpCAM⁺ and EpCAM^- keratinocytes were sorted and consequently infected with the LV-shRNAs at MOI 5. On culture day 10, the cell numbers of all LV-shRNA-infected EpCAM^- keratinocytes were not significantly varied (Fig. 6F). However, for the EpCAM⁺ cells, the shRNA2, shRNA3, and shRNA4 lentiviral infections caused significant attenuation of cell growth, compared to the LV-shRNA1-infected cells (Fig. 6F). This study emphasized that EpCAM not only acts as a new marker of EpPCs but also functionally links to the proliferation of human EpPCs.

Discussion

Wnt pathway represents a conserved signal for controlling cell division of stem cells, and Wnt upregulation in adult somatic cells often associates with tumorigenesis (5). Previous reports showed that overexpressing a stabilized β-catenin or adding Wnt16b in human keratinocytes significantly prolonged the cell growth and clonogenicity (38,45). Recent dermatological studies also suggest that highly expressed Wnt proteins are associated with basal cell carcinoma (17,25) and hyperproliferative psoriatic keratinocytes (7,30,32). In this study, we demonstrate that exogenous Wnt3a efficiently promotes the cell growth and cell division of keratinocytes, emphasizing the importance of Wnt pathway in the regulation of human keratinocyte proliferation.

Both canonical Wnt/β-catenin and noncanonical Wnt/ planar cell polarity (PCP) pathways are involved in the growth promotion in normal and pathological epidermis (7,17,32,39). It has been demonstrated that putative EpPCs express a higher level of non-cadherin-associated β-catenin and TCF/lymphoid enhancer-binding factor 1 (LEF; TCF1α) transcription activity (45). A recent study further showed that ablation of β-catenin signaling disrupted the stemness and tumorigenesis of epidermal cancer stem cells (21). Particular Wnt ligands and their receptors, Frizzled (fzd) proteins, are specifically expressed in distinct layers in normal epidermis (32). For instance, noncanonical Wnt5a and fzd6 expression are restricted in the basal cell layer of epidermis, implicating a role in keratinocyte proliferation and self-renewal (32). In addition, in the psoriatic plaques, both upregulated nuclear β-catenin and Wnt5a-related signals have been detected (7,8,30). Interestingly, global expression profile of psoriatic skin showed that Wnt5a upregulation was accompanied with reduced axin2 expression and nuclear β-catenin, suggesting a transition from canonical Wnt toward noncanonical Wnt signaling during the progress of psoriasis (7).

Although adding Wnt3a/BIO and providing a stabilized β-catenin (45) both activate canonical Wnt signal and stimulate proliferation of human keratinocytes, notable differences between these two studies were found. In contrast to the Wnt ligand and GSK-3β inhibition by BIO, activating Wnt signal through a recombinant retrovirus carrying N-terminally truncated β-catenin has little effect on the cell morphology, involucrin expression, α6 integrin expression, colony size, and overall colony numbers (45). The cell cycle kinetic analysis and BrdU incorporation were also not affected by the introduced β-catenin mutant. We postulated that in spite of the high efficacy on activating the TCF/LEF promoter, the truncated β-catenin mutant may possess less affinity to other transcription activators and result in less competency to activate some genes containing Wnt-related response elements.

It is interesting to find that providing the Wnt ligand did not increase the growth of keratinocytes at a high cell density culture. Accumulated evidence demonstrated that the cell density of a keratinocyte culture is a critical factor for the cell growth and differentiation (29,33). Continuous culture at confluence densities steers terminal differentiation and commitment of human keratinocytes (29,33). In addition, lipid metabolism and cellular responses to certain chemicals are altered at different growth densities (28,41). A study of glycosaminoglycan (GAG) biosynthesis in the culture of human epidermal keratinocytes demonstrated that newly synthesized sulfated GAGs, including heparan sulfate and chondroitin sulfate, strikingly declined from preconfluent to confluent growth condition (27). Heparan sulfate constitutes heparin sulfate proteoglycans (HSPGs), which serve as a reservoir and functional modulator for extracellular Wnt molecules (16). Although the exact details of the mechanism involved are still obscure for the regulation of Wnt stimuli and the downstream signals in human keratinocytes, both the developmental status of cultured cells and the availability of HSPGs on the cell membrane may be critical factors for the nonresponse to Wnt3a/BIO at a high cell density culture.

The kinetic cell growth and clonal expansion assays both supported the contribution of activated Wnt signal in the cell proliferation of human keratinocytes. In addition, recombinant Wnt3a simultaneously reduced the expression of involucrin, suggesting that the Wnt pathway may also prevent the mitotic keratinocytes toward cell differentiation. Notably, only the primary keratinocytes before their third in vitro passage and short-term growth effects within 2 weeks were evaluated for the Wnt stimuli here. We did not clarify the susceptibility of Wnt for the subpopulation of the cells, such as EpPCs, TA cells, and committed keratinocytes. The postulated dedifferentiation effect of Wnt signals requires further experiments by performing quantitative estimation of involucrin⁺ cells after long-term in vitro culture and testing the Wnt effect on individual subpopulation after third passages (45). Whether the Wnt activation can revert TA cell fate to EpPCs or dedifferentiate committed keratinocytes to become mitotic TA cells or EpPCs will be further explored.

We observed that in Wnt3a/BIO-treated keratinocytes, several cell adhesion molecules, such as ALCAM, EpCAM, and integrins, were upregulated (Table 1). These adhesion molecules may help the initial cell attachment, cell division, and clonal expansion. EpCAM upregulation is frequently associated with the occurrence of human adenocarcinoma and basal cell carcinoma in human skin (2), and EpCAM is also characterized as one of the signatures of cancer stem cells (9,24) and human pluripotent embryonic stem cells (19). Importantly, previous studies in human skin have demonstrated that EpCAM expression is restricted in some cells of the basal layer of the epidermis but is not detected in the differentiated keratinocytes in spinous, granular, or cornified layers (13,34), suggesting a potential role in the growth regulation of EpPCs.

In contrast to b1 integrin, which labels both suprabasal keratinocytes and EpPCs in vivo, EpCAM showed a similar pattern as MCSP, which marks only proliferative basal keratinocytes (11,12,15). However, inhibition of the MCSP expression did not attenuate the clonogenicity and proliferation of the EpPCs (15). In contrast, sorted EpCAM⁺ keratinocytes showed higher BrdU incorporation than EpCAM^- cells, and, in particular, the downregulation of EpCAM significantly reduced cell number and colony formation. This result is consistent with a recent finding in hepatic cancer cells showing that EpCAM has dual roles as a cell adhesion molecule and a proliferation-promoting factor (20,24). Our results indicated that EpCAM not only represents a new marker for EpPCs but also functionally links to the stemness and growth potency of EpPCs.

In primary human keratinocytes, we provided the first evidence demonstrating that the cell growth enhancement by Wnt/β-catenin signal was mediated through the EpCAM expression. The linkage between the EpCAM expression and Wnt signal was strengthened by a recent finding showing that two TCF4-binding loci were identified in the EpCAM upstream gene sequences (20). The EpCAM-mediated cell growth depends on both extracellular and intracellular cleavages by tumor necrosis factor-α-converting enzyme/a disintegrin and metalloprotease domain-containing protein 17 (TACE/ADAM17) and presenilin-2 to produce EpCAM extracellular (EpEX) and intracellular domain (EpICD), respectively. Extracellular EpEX is a soluble agonist for EpCAM signal by promoting the intracellular cleavage of EpCAM. The dissociated EpICD forms a nuclear complex with β-catenin, Lef-1, and four and a half LIM domains 2 (FHL2) to drive the c-myc and cyclins gene expression (20) (summarized in Fig. 7). In BIO-treated keratinocytes, we observed that the genes of epcam and ADAM17 were both elevated (Table 1), suggesting that abundant EpCAM was synthesized, and sufficient EpEX ligands were proteolytically produced. In addition, the enhanced cell proliferation was usually associated with the dissociation of cadherin adhesion, which leads to accumulation of intracellular β-catenin and nuclear EpICD/β-catenin/Tcf-1 complexes (36). These evidences illustrated that in human keratinocytes, Wnt/BIO stimuli may control the activation of EpCAM signal at multiple levels, including RNA transcription, extracellular ligand production, and cytosolic accumulation of the nuclear effectors.

Figure 7.

Intracellular signal networks of Wnt/β-catenin and EpCAM pathways for EPC maintenance. (A) Exogenous wingless-type mouse mammary tumor virus (MMTV) integration site family (Wnt) stimuli increase the EpCAM expression and the efficacy of clonal keratinocyte expansion. (B) The interaction of Wnt and EpCAM pathways in the EpPCs. LRP, low-density lipoprotein receptor-related protein; Dsh, disheveled; APC, adenomatous polyposis coli; GSK, glycogen kinase synthase; EpEX, EpCAM extracellular domain; EpICD, EpCAM intracellular domain; PS2, presenilin 2; TACE, tumor necrosis factor-α-converting enzyme; FHL2, four and a half LIM domains 2; Tcf/lef, transcription factor/lymphoid enhancer-binding factor 1.

Conclusion

Wnt signal is well characterized in the hair follicle but is still lacking strong support to show its importance in interfollicular keratinocytes. We have illustrated the physiological importance of Wnt and EpCAM in human keratinocytes and the necessary role of EpCAM for the Wnt-steered cell proliferation and clonogenicity. The discovery of EpCAM also sheds new light on the exploration of interfollicular niche for the EpPCs. Further investigation of Wnt/EpCAM in EpPCs will broaden our understanding in the field of skin homeostasis, wound healing, and skin cancer initiation.