Ovol2 Suppresses Cell Cycling and Terminal Differentiation of Keratinocytes by Directly Repressing c-Myc and Notch1*

Ovol2 belongs to the Ovo family of evolutionarily conserved zinc finger transcription factors that act downstream of key developmental signaling pathways including Wg/Wnt and BMP/TGF-β. We previously reported Ovol2 expression in the basal layer of epidermis, where epidermal stem/progenitor cells reside. In this work, we use HaCaT human keratinocytes to investigate the cellular and molecular functions of Ovol2. We show that depletion of Ovol2 leads to transient cell expansion but a loss of cells with long term proliferation potential. Mathematical modeling and experimental findings suggest that both faster cycling and precocious withdrawal from the cell cycle underlie this phenotype. Ovol2 depletion also accelerates extracellular signal-induced terminal differentiation in two- and three-dimensional culture models. By chromatin immunoprecipitation, luciferase reporter, and functional rescue assays, we demonstrate that Ovol2 directly represses two critical downstream targets, c-Myc and Notch1, thereby suppressing keratinocyte transient proliferation and terminal differentiation, respectively. These findings shed light on how an epidermal cell maintains a proliferation-competent and differentiation-resistant state.

The Ovo gene family encodes evolutionarily conserved proteins including members from Caenorhabditis elegans, Drosophila, Zebrafish, chick, and mammals. Ovo proteins contain four DNA-binding C 2 H 2 zinc fingers at the C termini and possess transcriptional regulatory activities (1)(2)(3)(4)(5). Drosophila ovo, the founding member of the family, acts genetically downstream of Wg (fly Wnt homolog) and DER (fly epidermal growth factor receptor homolog) signaling pathways and is required for epidermal denticle formation and oogenesis (6 -8). We and others have identified mammalian Ovol (Ovolike) genes, including Ovol1 (movo1), Ovol2 (movo2), and Ovol3 (movo3) in mice and OVOL1, OVOL2, and OVOL3 in humans (9 -12). Mammalian Ovol/OVOL (referred to as Ovol from here on) genes also appear to reside downstream of key developmental signaling pathways. For instance, Ovol1 is activated by the ␤-catenin-LEF1 complex, downstream effectors of Wnt signaling (13), and is a downstream target of TGF-␤/BMP7-Smad4 signaling (14,15). A functional Ovol1 gene is required for multiple developmental processes, including that of epidermis, hair follicles, kidney, and male germ cell differentiation (2,4,10). In both epidermis and testis, Ovol1 restricts the boundary of late progenitor cells during development by promoting cell cycle exit (2,4). Less is known about the biological function of Ovol2 and Ovol3.
Ovol2 is expressed in myriad embryonic and postnatal tissues (11,16). Consistent with a widespread expression pattern, ablation of the Ovol2 gene results in early embryonic lethality (16), which precludes a comprehensive analysis of its later developmental roles. Interestingly, Ovol2 expression appears to correlate with a proliferative stem/progenitor cell state (16,17). For instance, Ovol2 protein is detected in the basal layer of skin epidermis (17), where proliferative epidermal stem/progenitor cells reside. Moreover, Ovol2 is strongly expressed in inner cell mass as well as its in vitro equivalent, embryonic stem cells (16). The functional significance of Ovol2 expression in proliferating stem/progenitor cells remains to be established.
Mammalian epidermis is an excellent model system to study the molecular circuits that control proliferation and differentiation. Using cultured keratinocytes and mouse models, important regulators of epidermal proliferation and differentiation have been uncovered (18). Among these, c-Myc is the most intriguing because of its multiple and seemingly opposing roles (19). Although c-Myc is expressed in basal cells and clearly important for keratinocyte proliferation, its constitutive overexpression in cultured keratinocytes causes progressively reduced growth, precocious terminal differentiation, and loss of cells that express a high level of ␤1 integrin, a putative epidermal stem cell marker (20). The latter finding has been interpreted to indicate a c-Myc-stimulated premature exit from the stem cell compartment, a notion that is apparently supported by the observation of decreased ␤1 high or label-retaining cells in mice that overexpress c-Myc in the epidermal basal layer (20 -22). Notch signaling, which is initiated by ligand binding to the Notch receptor followed by cleavage and nuclear translocation of the intracellular domain that in turn binds to RBP-J to generate a transactivation complex, is critical for terminal differentiation of keratinocytes (19,(23)(24)(25)(26). Notch receptors (Notch1, 2, and 3) are not normally expressed in proliferating epidermal basal cells, and overexpression of the intracellular domain in these cells promotes a differentiating, spinous cell fate (23). Collectively, these findings suggest that fine-tuning c-Myc expression and suppressing Notch expression/signaling in the basal layer might be important for maintaining a differentiation-refractory progenitor cell state. To date, little is known about molecular mechanisms that directly repress c-Myc and Notch expression in epidermal progenitor cells.
In this work, we use HaCaT keratinocytes to explore the function of Ovol2. We show that Ovol2 regulates independent yet related aspects of keratinocyte proliferation and differentiation, namely suppressing rapid amplification and terminal differentiation but maintaining long term proliferation potential in culture. We present evidence that Ovol2 does so, at least in part, by regulating downstream targets c-Myc and Notch1.
Cell Counts and Clonal Assay-For cell count experiments, HaCaT cells were seeded in triplicate in a 12-well plate and transfected at 20 -30% confluence with 60 nM siRNA and Lipofectamine 2000 (Invitrogen) according to manufacturer's instructions. At 24-h intervals spanning 72 h post-transfection, the cells were trypsinized and counted using a hemacytometer. Clonal assays were performed as previously described (28). For first generation clonal assays, HaCaT cells were seeded at 8 ϫ 10 4 cells/well of a 6-well plate and transfected 24 h later with siRNA. The cells were trypsinized 24 h after transfection (day 2), replated at a density of 500 cells/well in 6-well plates, and retransfected with siRNA on day 8. After 14 days, three of the six wells/sample were fixed in 4% paraformaldehyde for 15 min, washed with 1ϫ PBS, and stained with a solution of 1% Rhodamine B, 1% Nile Blue before washing and air drying. The number and size of colonies were then scored. The remaining three wells were trypsinized, counted, and replated as second generation at a density of 500 cells/well in a 6-well plate.
Immunofluorescence-HaCaT cells were seeded into a 96-well plate and 24 h later transfected with siRNA. Seventytwo hours after transfection, the cells were fixed with 4% paraformaldehyde for 15 min and washed with PBS. Following permeabilization with cold 90% methanol for 5 min at room temperature and PBS washes, the cells were blocked in 10% normal goat serum for 1 h and then incubated with mouse anti-cleaved poly(ADP-ribose) polymerase (Asp 214 ) (1:10; BD Pharmingen) and rat anti-phospho histone H3 (pS28) (1:20; BD Pharmingen) in 1% normal goat serum overnight at 4°C. The cells were washed with PBS, and the nuclei were stained with 4Ј,6-diamidino-2-phenylindole. The plates were run on an In-Cell Plate Reader, and the fluorescence was measured using InCell Developer Software.
FACS Analysis-HaCaT cells were seeded in duplicate at 2.5 ϫ 10 5 cells/10-cm dish and after 24 h were transfected with siRNA. The cells were collected 72 h after transfection. For propidium iodide staining, the cells were washed once with cold PBS, fixed with 70% ethanol for 24 h at Ϫ20°C, followed by two PBS washes. After being stained with a propidium iodide staining solution (50 g/ml propidium iodide, 0.1% sodium citrate, 0.03% Nonidet P-40, 50 g/ml RNase A, PBS) for 30 min at room temperature, the cells were then washed once with PBS. For ␣6 integrin staining, the cells were washed once with 2% fetal bovine serum with PBS and stained on ice for 30 min with a phycoerythrin-conjugated rat anti-CD49f (␣6 integrin) antibody (1:40; BD Pharmingen) and the appropriate isotype control. The cells were washed two times with cold PBS, run on a FACS Calibur flow cytometer (BD), and analyzed with FlowJo software (Tree Star Inc.).
Cyclic Amplification of Selected Targets-Cyclic amplification of selected targets was carried out as previously described (3) using purified recombinant His 6 -Ovol2 protein. Forty Ovol2-selected clones were sequenced to derive an Ovol2 consensus sequence.
Reporter Assays-Assays were performed in 293T and HaCaT cells. The 293T cells were transfected using calcium phosphate as described (29) and HaCaT cells using Lipo-fectamine 2000 (Invitrogen). Typically, transfection experiments were done in 12-well plates with each well transfected with a total of 1.6 g of plasmids including 0.1 or 0.8 g (for 293T or HaCaT cells, respectively) of pGL3-c-Myc (4), pGL3-Hes1 (where luciferase is under the control of the Hes1 promoter; gift from C. C. W. Hughes, University of California, Irvine), or pGL4 Notch1 (where luciferase is under the control of the 2.4-kb Notch1 promoter; gift from G. P. Dotto, Massachusetts General Hospital (30)), 0.08 or 0.3 g (for 293T or HaCaT cells, respectively) of ␤-actin-␤-galactosidase construct (transfection control), and varying amounts of pCB6-Ovol2A, an Ovol2A expression vector (as indicated in the figure legends). pCB6ϩ (empty vector containing the cytomegalovirus promoter) was used as filler DNA. The cells were harvested 24 -36 h after transfection, and luciferase activity was measured in whole cell extracts using the luciferase assay system (Promega). ␤-Galactosidase activity was measured as previously described (31).
Chromatin Immunoprecipitation (ChIP)-HaCaT cells were seeded in 10-cm plates, and each plate was transfected at ϳ90% confluency with 24 g of pCB6-Ovol2A using Lipofectamine 2000. The plates were cross-linked 24 h after transfection with 1% formaldehyde for 10 min at room temperature, and chromatin immunoprecipitates were isolated using the ChIP assay kit (Upstate Biotechnology Inc.) and anti-Ovol2 antibody (16) according to the manufacturer's instructions. PCR was performed using human c-Myc primers (4) and human Notch1 primers containing the following sequences: forward, 5Ј-ACCAGGAGCAGAGGACGTC-3Ј; and reverse, 5Ј-CTTT-CCTGGCACACCTCTTG-3Ј. The following PCR program was used: 94°C for 5 min followed by 31-38 cycles (within the linear range of the primers) of 94°C for 45 s, 60°C for 45 s, and 72°C for 1 min followed by a final extension at 72°C for 7 min.
Real Time PCR-Total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Five g of total RNA was reverse transcribed into cDNA using the Superscript III RNase H reverse transcriptase (Invitrogen) according to the manufacturer's instructions. All of the primers used for quantitative real time PCR were designed to span exon-exon borders to minimize the possibility of nonspecific amplification of genomic DNA. The following primers were used for quantitative real time PCR: glyceraldehyde-3-phosphate dehydrogenase forward, 5Ј-GGCATCCTGGGCTA-CACTGAG-3Ј; glyceraldehyde-3-phosphate dehydrogenase reverse, 5Ј-TGAGGTCCACCACCCTGTTG-3Ј; c-Myc forward, 5Ј-TCTCTCCGTCCTCGGATTCTC-3Ј; c-Myc reverse, 5Ј-GGAGCCTGCCTCTTTTCCAC-3Ј; Notch1 forward, 5Ј-AATGTGGATGCCGCAGTTG-3Ј; and Notch1 reverse, 5Ј-CGGTCCATATGATCCGTGATG-3Ј. Quantitative real time PCRs were performed using iQ SYBR Green Supermix (Bio-Rad) according to the manufacturer's instructions on an iCycler real time quantitative PCR system (Bio-Rad). The thermal profile included 95°C for 3 min, 40 cycles of denaturation at 95°C for 10 s and annealing at 63°C for 1 min with the optics on for fluorescence monitoring. The reactions were run in triplicate and always included a standard curve and no template sample for each primer set as a control for the linear range and nonspecific PCR products, respectively. Glyceraldehyde-3-phosphate dehydrogenase was used to normalize the data. The mean threshold cycle (C t ) for individual reactions was identified using the iCycler IQ sequence analysis software (Bio-Rad).
Three-dimensional Organotypic Culture-Skin equivalents were prepared in 12-well tissue culture inserts (3-m pore Thinsert; Greiner Bio One) as previously described (32). Briefly, gels (400 l/insert) were prepared from collagen type I (BD Biosciences) at a final concentration of 2.5 mg/ml and seeded with 3T3 cells at 2.5 ϫ 10 4 cells/ml. The gels were submerged in Dulbecco's modified Eagle's medium, 10% fetal bovine serum and grown at 37°C in a humidified, 95% air, 5% CO 2 atmosphere. After 48 h, the gels were released from the side of the insert using a pipette tip and allowed to contract for 5 days. The medium was replaced every 2 or 3 days. 2 ϫ 10 5 HaCaT cells (either negative control or Ovol2-depleted)/insert were added and allowed to grow for 2 days before being air-lifted and cultured in medium containing 2.8 mM calcium for 10 days. The medium was changed every other day.
Mathematical Modeling-A cell population model consisting of a system of ordinary differential equations was considered (33, 34) (see supplemental text for equations and details of the model). In the system, after each stem cell (N 0 in total) division, the daughter cells have the probability (p 0 ) to remain a stem cell, or (1 Ϫ p 0 ) to become a progenitor (N 1 ) cell that can undergo a maximum of M divisions before becoming growth arrested (N Mϩ1 ) cells. For a smaller M, the solution of the system could be explicitly and analytically expressed. For a large M, the system was solved using NDSolve function in Mathematica. The parameter fitting was based on FindFit function of Mathematica.

Loss of Ovol2 Leads to a Transient Increase in Growth and a
Loss of Long Term Proliferation-In both mice and humans, a single Ovol2 gene encodes different protein isoforms with presumably opposing transcriptional regulatory activities (5,11). We therefore first determined which isoform of Ovol2 is expressed in epidermal keratinocytes. By Western blot analysis of nuclear extracts from mouse and human keratinocytes and mouse skin, we found that based on the sizes of the proteins seen in each of these samples, Ovol2A repressor is the dominant isoform expressed in epidermis in vivo and cultured keratinocytes in vitro (Fig. 1A). The other isoforms, namely Ovol2B and Ovol2C, were not detected under our experimental conditions.
Next we turned to use siRNAs to knockdown Ovol2 in HaCaT human keratinocytes, an immortalized cell line that displays epidermal progenitor cell activity, i.e. forming a stratified epidermis under organotypic culture conditions (28,35,36). A number of different differentiation-specific markers are expressed in HaCaT cells under organotypic conditions, although stratification is incomplete and imperfect (37). These cells were chosen for our study because in culture they contain a small subset of quiescent stem/progenitor-like cells (28), mimicking epidermal homeostasis in vivo. When compared with cells transfected with a negative control siRNA that does not target any known gene, Ovol2 proteins were effectively depleted with Ovol2-specific siRNAs, and knockdown persisted in culture for at least 6 days for the most effective siRNA ( Fig. 1B and supplemental Fig. S1A). A growth curve analysis revealed a significantly higher number of cells in Ovol2-depleted high density culture, particularly at 72 h after siRNA transfection (1.7-fold Ϯ 0.12) (Fig. 1C). Consistently, the Ovol2 knockdown plates appeared more confluent, and the cells were more tightly packed on the plate than the negative control plates (Fig. 1D). Three different siRNA oligonucleotides (Ovol2 #1 targets exons 3 and 4, Ovol2 #2 targets the 3Ј-untranslated region, and Ovol2 #3 targets exon 2) produced a similar trend, albeit to a lesser extent for the less effective siRNAs #2 and #3 (supplemental Fig. S1, C and D). This finding suggests that the increase in cell number is not an off target effect. Knockdown was also performed in mouse keratinocytes using adenoviral small hairpin RNA against Ovol2, and a similar increase in cell number was observed (data not shown). Subsequent experiments focused on the most effective siRNA (#1) and HaCaT cells.
Because cultured keratinocytes are a heterogeneous population of cells with different proliferation potentials (28,38,39), we performed clonal assays in which cells are plated at a clonal density to determine the proliferation potential of individual cells. Specifically, we transfected HaCaT cells with either control or Ovol2 siRNA at days 1 and 8 to efficiently knockdown Ovol2 during the first 14 days of clonal assay (this we term first generation). We then assessed the proliferation potential of the cells harvested from the first generation by replating a portion of them at a clonal density, but this time with no additional siRNA knockdown (this we term second generation). Consistent with an increased cell number in short term high density cultures described above, we observed a reproducible increase in the number of large (Ͼ 2 mm) colonies in Ovol2-depleted first generation cultures when compared with negative control (Fig. 1, E, top panels, and F, and supplemental Fig. S2, top panel). Trypan blue staining showed that Ovol2 depletion did not affect the viability of the first generation cells (data not shown). Interestingly, after replating an equal number of first generation cells, we now observed a reduction in both colony size and number in the second generation culture derived from cells that had been previously depleted of Ovol2 (Fig. 1, E, bottom panels, and F, and supplemental Fig. S2, bottom panel). This result suggests that depleting Ovol2 leads to a reduction in the number of keratinocytes that possess long term proliferation potential. Together, our results indicate that depletion of Ovol2 results in a transient increase in growth but a decrease in long term proliferation.
Reduced Active Cycling Accompanies the Growth Phenotype of Ovol2-depleted Cells-A priori, an increase in cell number can be caused by a decrease in cell death or an increase in cell proliferation. To distinguish between these possibilities, we stained negative control and Ovol2 knockdown cells cultured under high density conditions for the presence of cleaved poly-(ADP-ribose) polymerase, which symbolizes activated apoptosis (40), and phospho-H3 (pS28), which marks cells in the mitotic phase (specifically the prophase to anaphase transition) of the cell cycle (41). A slight decrease in the number of cleaved poly(ADP-ribose) polymerase positive cells was observed in Ovol2 knockdown cells (0.17% Ϯ 0.03) when compared with negative control cells (0.25% Ϯ 0.07) (Fig. 2, A and B, left columns). This slight change is consistent with, but unlikely to completely account for the increase in cell number. Contradictory to our prediction, there was a significant decrease in the  (Fig. 2, A and B, right  columns). An overall reduction in the number of actively cycling cells was confirmed by FACS analysis of cell cycle profiles, revealing a 34% increase in G 1 /G 0 cells and a 25% decrease in S/G 2 /M cells in the Ovol2-depleted high density culture (Fig.  2, C and D). Again, a slight reduction in apoptotic cells, as monitored by the sub-G 1 population, was observed.
Mathematic Modeling Suggests That Ovol2 Suppresses the Rate of Keratinocyte Cycling but Prolongs the Number of Divisions They Undergo-To explore the cellular basis of these seemingly contradictory observations, we applied a simple cell lineage mathematic model (33,34) to examine keratinocyte stem/progenitor cell evolution. We considered three distinct cell types (42): slow cycling stem cells that can proliferate indefinitely, faster cycling progenitor cells with a set proliferation potential (indicated by M number of cell divisions they are able to undergo), and growth-arrested cells (Fig. 3A). Using this model, we explored various parameters that would allow the recapitulation of our experimental observations. We found that a total of 36 rounds of progenitor cell divisions are necessary to mimic the observed growth of normal HaCaT keratinocytes (data not shown). Under this precondition, we found that decreasing t 1 (cell cycle time for progenitor cells) but not t 0 (cell cycle time for stem cells) was necessary to robustly capture the Ovol2 depletion-induced transient increase in proliferation in high density culture (Fig.   3B, left panel). However, this could not replicate the finding of decreased clonogenicity in second generation Ovol2 knockdown culture (Fig. 3B, right panel). Importantly, using a smaller M in addition to decreasing t 1 recapitulated both the increased growth in high density culture and the decreased clonogenicity in second generation culture of Ovol2-depleted cells (Fig.  3C). In contrast, decreasing p 0 had minimal impact on the growth behavior of Ovol2-depleted cells regardless of whether M was reduced or not ( Fig. 3C and data not  shown). These results suggest that both decreased cell cycle time and compromised proliferation potential of progenitor cells are critical for the observed growth phenotype of Ovol2-deficient culture. Indeed, we observed a remarkable decrease in the expression of ␣6 integrin, which when highly expressed, marks keratinocytes that possess long term proliferation potential and increased clonogenecity (43,44), in Ovol2 knockdown cells (Fig. 3D). This result provides further correlative support that Ovol2 is required for cultured keratinocytes to maintain their proliferation potential.
Ovol2 Suppresses Active Keratinocyte Proliferation by Repressing c-Myc Expression-How does Ovol2 keep keratinocytes proliferation in check? We entertained the possibility that Ovol2 might function by directly repressing a positive regulator of the cell cycle. c-Myc is an excellent candidate because 1) it is repressed by Ovol1, whose zinc finger domain is highly homologous to that of Ovol2 (11,13); and 2) its expression level in keratinocytes has been proposed to govern whether a stem cell should exit into the transit amplification (TA) stage, how well a TA cell proliferates, or whether it should terminally differentiate (20 -22, 45, 46).
We first examined c-Myc expression in control and Ovol2 knockdown cells. Loss of Ovol2 indeed caused an up-regulation of both c-Myc mRNA and c-Myc protein levels (Fig. 4, A  and B). Next we asked whether inhibiting c-Myc activity might rescue the growth phenotype of Ovol2 knockdown cells. A small molecule inhibitor (10058-F4) that interferes with c-Myc binding to its partner in transcriptional activation (47) was added to keratinocytes cultured under clonal conditions with or without Ovol2 knockdown. At an inhibitor concentration where minimal change of colony growth was observed in first generation control cultures, Ovol2 knockdown no longer elicited a significant increase in transient proliferation (Fig. 4, C and D). In contrast, Ovol2 knockdown still resulted in a dramatic reduction of colony formation in second generation cultures regardless of whether the c-Myc inhibitor was present or not (data not shown). Consistent with this finding, addition of the c-Myc inhibitor was unable to rescue the loss of ␣6 high cells caused by Ovol2 depletion (Fig. 4E). Collectively, these results suggest that c-Myc up-regulation accounts at least in part for the Ovol2 depletion-induced transient increase in proliferation but not for the loss of long term proliferation potential.
We next asked whether c-Myc is a direct target of Ovol2. Cyclic amplification of selected targets analysis revealed an Ovol2 consensus DNAbinding sequence that is almost identical to the Ovol1 cognate sequence (3) (Fig. 5A). Because a conserved Ovol1 consensus was previously discovered in a 1.6-kb c-Myc promoter fragment to mediate Ovol1 repression (4), we performed reporter assays to ask whether Ovol2 also represses the c-Myc promoter. Transfection of an Ovol2A-expressing construct repressed the promoter-driven luciferase activity in a dose-dependent manner (Fig. 5B), whereas transfection of a construct expressing Ovol2B or Ovol2C that lacks the functional SNAG repressor domain had no effect (data not shown). When the Ovol consensus binding site was deleted or mutated, repression by Ovol2A was reduced but not abolished (Fig. 5B), suggesting the presence of additional, noncanonical Ovol2A-responsive element(s). Indeed, a minimal c-Myc promoter that contains only 100 bp upstream (proximal region) of the transcriptional start site (4) was repressed by Ovol2A in both 293T and HaCaT cells (Fig. 5C). This region does not contain any Ovol2 consensus but contains recognition sequences for Smad3 and E2F that mediate transcriptional repression (48 -50). Although mutation of the Smad3-binding sequence had no negative effect, mutation of the E2F-binding sequence significantly compromised repression by Ovol2A (Fig. 5D and data not  shown). In ChIP assays, Ovol2 was found to occupy both the distal consensus site and the proximal nonca-nonical site of the endogenous c-Myc promoter, with binding to the noncanonical site being significantly more prominent (Fig. 5E). Taken together, our findings demonstrate that Ovol2A represses c-Myc expression via direct binding to the c-Myc promoter.

Ovol2 Suppresses Keratinocyte Terminal Differentiation by
Repressing Notch1-The growth and differentiation of keratinocytes are tightly linked. To test whether Ovol2 depletion predisposes keratinocytes to terminal differentiation, we treated negative control and Ovol2 knockdown cells with TGF-␤ or calcium to induce growth arrest or terminal differentiation, respectively, and examined the expression of keratin 1 (K1) and loricrin, markers of terminal differentiation. Neither Ovol2 knockdown or TGF-␤ treatment alone was sufficient to induce K1 and loricrin expression (Fig. 6A). However, TGF-␤ was able to induce loricrin expression when Ovol2 was depleted. Moreover, induction of K1 and loricrin expression by calcium treatment was significantly more remarkable in Ovol2-depleted cells. The up-regulation of differentiation markers was also seen for Ovol2 siRNAs #2 and #3 (supplemental Fig. S1E), indi-cating that the premature differentiation is not an off target effect. To mimic conditions in vivo, we next turned to a threedimensional organotypic culture system. Control or Ovol2-depleted HaCaT keratinocytes were grown on a fibroblast-collagen matrix under conditions that allowed the examination of early differentiation and stratification events. At a culturing time point when control samples showed expression of basal marker K14 but no detectable expression of differentiation marker K1 (51), Ovol2-depleted samples stained weakly for K14 but strongly for K1 (Fig. 6B). Collectively, our results suggest that Ovol2 suppresses the differentiation tendency of keratinocytes; in its absence, keratinocytes are more prone to extracellular signal-induced terminal differentiation.
Notch signaling plays a pivotal role in promoting keratinocyte differentiation. We therefore wondered whether this sig-   OCTOBER 16, 2009 • VOLUME 284 • NUMBER 42 naling pathway is affected by the loss of Ovol2. Examination of our microarray data (to be published else where) revealed a 1.9-fold increase in Notch1 transcript levels in Ovol2-depleted keratinocytes, whereas no change was seen for other Notch genes. Real time PCR confirmed Notch1 up-regulation at 24, 48, and 72 h after transfection of Ovol2 siRNA (Fig. 6C and data not shown). Consistent with this finding, we found that Ovol2A was able to repress the Notch1 promoter in a dose-dependent manner in both 293T and HaCaT cells (Fig. 6D). ChIP analysis showed that Ovol2 bound to the endogenous Notch1 promoter (Fig. 6E), suggesting that Notch1 is a direct target of Ovol2. Moreover, the promoter activity of Hes1, a known target of Notch signaling (52), was repressed by Ovol2A in a dose-dependent manner (Fig. 6F). These results demonstrate that Ovol2 modulates Notch1 expression and downstream signaling.

Role of Ovol2 in Keratinocytes
To address the importance of Notch 1 in the role of Ovol2 in regulating differentiation competence, we repeated the calcium induction experiment, this time including DAPT, a ␥-secretase inhibitor that blocks Notch signaling (53,54). The induction in K1 expression by the loss of Ovol2 was completely abolished when DAPT was added (Fig.  6G). Different concentrations of DAPT were used and yielded similar findings (data not shown). In contrast, blocking Notch signaling did not rescue Ovol2 depletion-induced loss of long term colony formation (data not shown). Therefore, we conclude that Ovol2 normally suppresses Notch signaling to prevent premature terminal differentiation of keratinocyte progenitor cells.

DISCUSSION
Our study has elucidated an important function for Ovol2 in human epidermal keratinocyte growth and differentiation. We have shown that Ovol2 regulates three related cellular aspects: proliferation rate, proliferation potential, and differentia- tion tendency (Fig. 7). In doing so, Ovol2 contributes to the overall genetic program that maintains keratinocytes in a proliferation-competent and differentiation-resistant state, that of a stem/progenitor cell.
Balancing Cell Cycling with Long Term Proliferation Potential in a Single Progenitor Cell Population: Role of Ovol2-Loss of Ovol2 leads to an initial transient increase in growth but decreased long term clonogenicity. Similar cellular changes have been reported for keratinocytes deficient for Rac1 or over-expressing c-Myc (20 -22, 55-57).
In a widely accepted model of epidermal homeostasis, it is posited that stem cells that reside in the epidermal basal layers divide infrequently to give rise to TA cells that cycle faster but are destined to terminally differentiate after a few rounds of cell divisions (42). In the framework of this model, the growth phenomenon of c-Myc-and Rac1-altered cells has been explained by a differential effect of these genes on the different cell type populations present in a heterogeneous keratinocyte culture. One might envision that exit of the socalled stem cells from a slow cycling state will increase the percentage of actively cycling TA cells, thereby resulting in a transient burst in growth. However, because the number of cell divisions that a TA cell can undergo is limited (58), this will ultimately cause a depletion of cells with long term proliferation potential. In an alternative model, misregulation of these genes might affect a single population of progenitor cells, making them cycle faster, thereby exhausting the number of cell divisions that they are programmed to undergo. Our mathematic modeling of the behavior of control and Ovol2-depeted keratinocytes presents a scenario that is more consistent with the second model. Reducing the cell cycle time (t 1 ) and proliferation potential (M) of the progenitor cell population indeed can recreate the observed increase in transient growth and decrease in colonogenicity, whereas changing the cell cycle time (t 0 ) and/or exit probability (p 0 ) of a slow cycling stem cell population cannot. From this analysis, we infer that although there clearly is heterogeneity in terms of proliferation rate (slow cycling versus fast cycling) in cultured keratinocytes, such heterogeneity may not be functionally relevant to the growth of these cells. This said, it is important to note that our modeling was done using previously described experimental parameters, namely that ϳ10% of cultured HaCaT keratinocytes are slow cycling stem cells and that these stem cells cycle at a rate that is half of the TA cells (28). These constraints dictate that the rare presence of slow cycling stem cells generates a minimal impact on overall culture growth.
In principle, our findings are consistent with the recently proposed model that a single population of proliferating progenitor cells is sufficient to maintain normal homeostasis of mouse tail epidermis (42,59,60) and suggest that the same might be true for cultured human keratinocytes. More impor-  tantly, our studies uncover an important novel player, transcription factor Ovol2, that regulates the cellular activities of keratinocytes. If a normal progenitor cell has an intrinsically set number of cell cycles that they can undergo (i.e. a set proliferation potential), then when cycling faster (decreasing t 1 ), they will exit the cell cycle faster. This calls for means that keep proliferation rate in check to achieve populational longevity. Moreover, any molecular alterations that minimize this proliferation potential (i.e. decreasing M) will cause premature growth arrest. It is likely that both mechanisms (proliferation rate and potential) account for the observed reduction in the number of total cycling cells as well as long term clonogenicity in Ovol2-depleted culture. As such, our combinatory experimental and modeling approach sets a useful paradigm to study the function of other critical regulators of stem/progenitor cell homeostasis in epidermis as well as in other tissues.
Molecular Mechanism of Ovol2 Function in Controlling Keratinocyte Proliferation-Consistent with the finding of similar phenotypes caused by Ovol2 depletion and c-Myc overexpression, our results demonstrate that c-Myc up-regulation accounts for the transiently increased proliferation in Ovol2 knockdown cells. However, inhibiting c-Myc function in these cells is not sufficient to rescue the loss of long term proliferation potential. These data corroborates our modeling finding that Ovol2 independently regulates proliferation rate and proliferation potential. Although its role in the proliferation rate control is in part mediated by repressing c-Myc expression, Ovol2 must regulate additional molecular events to control proliferation potential.
The finding that Ovol2, like Ovol1, also represses c-Myc is intriguing. Ovol1 is predominantly expressed in suprabasal cells, whereas Ovol2 is predominantly expressed in basal cells. However, our previous analysis of Ovol1/Ovol2 compound mutants has provided evidence for possible functional redundancy and compensation between the two Ovol genes in an in vivo setting (17). Sharing a common molecular target offers a partial explanation for this observation. Importantly, despite regulating a common target, depletion of the two Ovol proteins leads to apparently opposite cell biological outcomes because of their different sites of expression: loss of Ovol1 results in expansion of late epidermal progenitor cells, whereas loss of Ovol2 results in depletion of long term proliferating keratinocytes. This further highlights the importance of intricate c-Myc regulation at multiple stages of keratinocyte progenitor cell evolution: down-regulation of c-Myc is important for both keeping a long-lived progenitor cell population and initiating postmitotic differentiation (19). Further studies are now necessary to examine whether Ovol2 is also required for maintaining epidermal progenitor cells in vivo.
Another interesting finding of our study is that Ovol2 is able to repress transcription via an E2F-binding sequence. We have previously shown that Ovol1 also represses the c-Myc minimal promoter, where this E2F site resides (4). It is possible that DNA binding via an E2F protein underlies both Ovol1 and Ovol2 repression of c-Myc. A likely candidate is E2F4/5, which is known to form a repressor complex with Smad3 and p107 to repress c-Myc expression (48 -50). Additional studies are needed to test this notion.
Ovol2, Notch1, and Keratinocyte Differentiation-Our work highlights a role for Ovol2 in negatively modulating terminal differentiation of keratinocytes even when differentiation-inducing signals are present. However, when such signals are absent, the loss of Ovol2 alone is insufficient to trigger differentiation, suggesting that Ovol2 is not a master switch of differentiation but instead regulates differentiation competence. We have shown that Ovol2 performs this function by repressing Notch1 expression as well as downstream signaling. Previous studies have underscored the importance of down-regulating Notch signaling in basal progenitor cells to prevent premature spinous cell differentiation (23) and shown that epidermal growth factor receptor signaling negatively regulate Notch1 transcription. While epidermal growth factor receptor repression of Notch1 is via an indirect mechanism that involves tumor suppressor p53 (30), we have found Notch1 to be a direct transcriptional target of Ovol2. Together, these findings indicate that multiple mechanisms are in place to repress Notch1 expression and activity in the proliferating epidermal progenitor cells.
The ability to suppress differentiation is conceivably very important for normal tissue progenitor cells to maintain a differentiation-resistant state, but this ability, when hijacked by malignant cells, can lead to differentiation refractory tumor growth. In fact aberrant epidermal growth factor receptor signaling is frequently linked to tumorigenesis. Importantly, Ovol2 not only suppresses terminal differentiation but also poses a limit on the rate of proliferation in keratinocyte progenitor cells. This may be significant for developmental progenitor cells to maintain an undifferentiated state while restricting unwanted growth. Studies of tissue-specific knock-out mouse models are ongoing to examine whether the same holds true in vivo.