Cyclin E drives human keratinocyte growth into differentiation

ORIGINAL ARTICLE

Cyclin E drives human keratinocyte growth into differentiationA Freije1,2, L Ceballos1,2, M Coisy3, L Barnes3, M Rosa1,2, E De Diego4, JM Blanchard3 and A Gandarillas1,2,3,5

Human epidermis is continuously exposed to environmental mutagenic hazard and is the most frequent target of humancancer. How the epidermis coordinates proliferation with differentiation to maintain homeostasis, even in hyperproliferativeconditions, is unclear. For instance, overactivation of the proto-oncogene MYC in keratinocytes stimulates differentiation. Herewe explore the cell cycle regulation as proliferating human keratinocytes commit to terminal differentiation upon loss ofanchorage or overactivation of MYC. The S-phase of the cell cycle is deregulated as mitotic regulators are inhibited in the onsetof differentiation. Experimental inhibition of mitotic kinase cdk1 or kinases of the mitosis spindle checkpoint Aurora B or Polo-like Kinase, triggered keratinocyte terminal differentiation. Furthermore, hyperactivation of the cell cycle by overexpressing theDNA replication regulator Cyclin E induced mitosis failure and differentiation. Inhibition of Cyclin E by shRNAs attenuated theinduction of differentiation by MYC. In addition, we present evidence that Cyclin E induces DNA damage and the p53 pathway.The results provide novel clues for the mechanisms committing proliferative keratinocytes to differentiate, with implications fortissue homeostasis maintenance, HPV amplification and tumorigenesis.

Oncogene advance online publication, 20 February 2012; doi:10.1038/onc.2012.22

Keywords: skin; stem cells; DNA damage; MYCER; cell size; endoreplication

INTRODUCTIONEpidermal homeostasis involves continuous cell renewal, expan-sion and enlargement.1,2 In the basal layer of the epidermis,mainly quiescent, reside the epidermal stem cells. Some daughtersof the stem cells activate their cell cycle, undergo a short clonalexpansion phase of rapid proliferation and migrate into thesuprabasal layers. As they leave the basal layer, they ceaseproliferation, initiate terminal differentiation and increase theircellular size.3,4 The mechanisms that coordinate epidermalkeratinocyte cell growth and differentiation are unclear. Activelyproliferating cells have a limited potential, as they are committedto differentiate after four or five rounds of cell divisions byunknown mechanisms.1 Epidermal proliferation and differentia-tion are controlled by cell adhesion to the basement membranevia integrins. Terminal differentiation can be rapidly induced bydepriving keratinocytes of cell anchorage in suspended cultures.5

Proto-oncogene MYC, a potent inducer of cell growth and cellcycle,6 stimulates epidermal stem cell differentiation.7,8 Activationof ectopic MYC in basal cells favours differentiation versusproliferation, what causes a rise of differentiation markers after4 -- 5 days. We previously proposed that this is the time required bykeratinocyte to undergo the clonal expansion phase of rapidproliferation that precedes terminal differentiation.7 MYC drivesdifferentiation in part by inhibiting cell adhesion.7,8 However, thecell cycle mechanisms by which MYC elicits this function have notbeen elucidated.

We have previously shown that postmitotic keratinocytescontinue DNA synthesis and become polyploid during differentia-tion in vitro9 and in vivo,10,11 and that the process of endoreplica-tion is stimulated by MYC. As for MYC, other molecules involved incell cycle progression caused epidermal hyperplasia, but were not

able to disrupt differentiation (for example, E2F, Cyclin D1, MDM2,cdk2, cdk4; references in Zanet et al.11).

Cyclin-dependent kinases cdk2 and cdk1 are key regulators ofS-phase and mitosis, respectively.12 In order to gain insight into themechanisms linking cell growth with differentiation, we havecompared the dynamics of Cyclin/cdk complexes during thecommitment to differentiation in normal keratinocytes or afterMYC overactivation. We have inhibited cell cycle kinases, includingkinases of the mitosis spindle checkpoint, and explored the effectson differentiation. The results reveal that the cell cycle isderegulated in the onset of differentiation, involving strongaccumulation of the S-phase regulator Cyclin E. Overexpressionof Cyclin E in human basal keratinocytes induced a mitosis block,re-replication and differentiation. This response appears to be partof a novel keratinocyte anti-proliferative mitosis-differentiationcheckpoint. Interestingly, MYC might promote keratinocyte differ-entiation by inducing Cyclin E. In turn, accumulation of Cyclin Ewould trigger the mitosis checkpoints by causing DNA damageand inducing the p53/p21 pathway. This mechanism might explainhow actively proliferating keratinocytes commit to differentiationand how the epidermis is protected from cell growth alterations.

RESULTSCdk complexes of S-phase and mitosis during suspension-induceddifferentiationIntegrins mediate cell adhesion to the basal membrane of theepidermis and sustain keratinocyte proliferation.13 Inactivation ofintegrin function in suspension rapidly triggers terminal differ-entiation. Within 5 h keratinocytes irreversibly lose their capacity

Received 2 June 2011; revised 6 January 2012; accepted 8 January 2012

1Cell Cycle, Stem Cell Fate and Cancer Laboratory. Institute for Training and Research of the Fundacion Marques de Valdecilla (IFIMAV-FMDV), Santander, Spain; 2MolecularBiology Department of Universidad de Cantabria (UC), Santander, Spain; 3Institut de Genetique Moleculaire de Montpellier (CNRS/UMPII), Montpellier, France; 4Servicio de CirugıaPediatrica, Hospital Universitario Marques de Valdecilla (HUMV), Santander, Spain and 5Laboratoire de Dermatologie Moleculaire, UPRES EA 3754, Institut Universitaire deRecherche Clinique, Inserm ADR Languedoc-Roussillon, Montpellier, France. Correspondence: Dr A Gandarillas, Cell Cycle, Stem Cell Fate and Cancer Laboratory, Institute forTraining and Research of the Fundacion Marques de Valdecilla (IFIMAV-FMDV), Av. Herrera Oria s/n., Santander 39011, Spain. E-mail: [email protected]

Oncogene (2012) 1 -- 13& 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12

www.nature.com/onc

http://dx.doi.org/10.1038/onc.2012.22

mailto:[email protected]

http://www.nature.com/

to proliferate and commit to differentiate.5 By 24 h all cells aredifferentiated. We placed primary keratinocytes in suspension toanalyse changes in the cell cycle during the transition todifferentiation. Epidermal differentiation is associated with mor-phological changes and with the induction of the cornifiedprecursor involucrin that can be assessed by flow cytometry(Figure 1a and Supplementary Figure 1a).9,14 Cellular size andcomplexity (light scattering), and the expression of involucrinincreased markedly after 12 h in suspension.

Bromodeoxyuridine (5-bromo-20-deoxyuridine, BrdU) incorpora-tion experiments showed that DNA synthesis stays active duringdifferentiation (Figure 1a and Supplementary Figure 1a). Labelling-adherent cycling cells by a BrdU pulse before suspension showedthat differentiating cells accumulated in G2/M by 12 h andsubsequently, bypassed mitosis and became polyploid (Supple-mentary Figure 1a). Therefore, DNA synthesis stayed active duringdifferentiation beyond the time of commitment (6 h). Concomi-tantly, we analysed the dynamics of cdk2 and cdk1 complexes.Cyclin E/cdk2 complexes control DNA replication, Cyclin A binds to

cdk2 or to cdk1 during the S/G2 transition, and Cyclin B/cdk1controls mitosis.12 The Cyclin A/cdk2 complex has been previouslyshown to depend on cell anchorage,15 also in keratinocytes.16,17

Expression of Cyclins A and B was high for the first 6 h insuspension but it was suppressed by 12 h (Figures 1b and 2a andSupplementary Figure 1b). Strikingly, Cyclin E followed anopposite pattern, and markedly accumulated from 6 to 24 h(Figures 1b and 2a and Supplementary Figure 1b). Similar resultswere observed by confocal microscopy analyses of stratifiedcultures (Figure 1c and Supplementary Figure 2). Interestingly, theswitch from mitotic Cyclins to high Cyclin E was associated withthe increase in cell size and involucrin expression, characteristicsof epidermal differentiation (Figures 1a-c and 2a and Supplemen-tary Figures 1-3).3,4

Consistent with the changes in Cyclin expression, the expres-sion of mitotic kinase cdk1 was downregulated in suspended cells(Figure 1b). Although the expression of Cyclin E and cdk2 proteinswas sustained, all the kinase activities were inhibited after 6 h insuspension (Figure 1d). The inhibition of Cyclin A-associated

Figure 1. Dynamics of cdk complexes of S-phase and mitosis, during the keratinocyte commitment to suspension-induced differentiation.(a) Quantitation of the flow-cytometry analyses of primary human keratinocytes placed in suspension for the periods of time indicated. Tophistogram: percentage of involucrin-expressing cells or cells with a differentiated morphology (High Scatter). Bottom histogram: percentage ofcells undergoing DNA synthesis (BrdU-incorporation) in G2/M (propidium iodide, PI) or polyploid; cells were incubated with BrdU for 3h beforeharvest, or for the last 12h in the case of 24h. Data are the average of duplicate samples and representative of four independent experiments withcells from three different individuals. Error bars are s.e.m. The plots and populations are shown in Supplementary Figure 1. (b) Expression of theregulators indicated in suspended keratinocytes as determined by western blotting; CA, Cyclin A; CB, Cyclin B; CE, Cyclin E; p-Rb, phosphorylatedRb. b-actin (b-act) was used a loading control. *Cyclin B (CB) was first immunoprecipitated owing to low expression. (c) Normal humankeratinocytes were double stained for Cyclin E (CE, in green) and involucrin (Invol, in red) or Cyclin A (CA; in red) and analysed by confocalmicroscopy. The images show a differentiating layer. Note that Cyclin E accumulates (arrows) and that stratified cells express mitotic Cyclin A(arrowhead; see Supplementary Figures 2a--c). Scale bar, 50mm. (d) Representative analyses of kinase activity on histone H1 associated with theregulators indicated; histogram bars shows the quantitation of the radioactivity of duplicates, small bars being s.e.m. (e) Immunoprecipitation (IP)for the regulators indicated on the top and WB for the regulators indicated on the left. PST: motive common to cdk1 and cdk2; i, a: inactive oractive forms of the regulators indicated according to mobility shift due to phosphorylation. Results with cells from two different individuals.

Cyclin E and keratinocyte differentiationA Freije et al

2

Oncogene (2012), 1 -- 13 & 2012 Macmillan Publishers Limited

activity preceded the inhibition of Cyclin B-associated activity,which was still high by 6 h. Both became undetectable by 12 h.However, a remaining low Cyclin E-associated activity wasdetectable by 12 h. Cyclin B and Cyclin E activities increasedslightly and tumour suppressor Rb and histone H3 werehyperphosphorylated as soon as cells were placed in suspensionby trypsin treatment (0 h: Figures 1b and 2a). Inactivation of Rb byphosphorylation marks cell cycle entry and total Rb is down-regulated during keratinocyte differentiation.16,18 Consistent withprevious reports,16,17 cdk inhibitors p21CIP1 (hereafter, p21) andp27KIP1 (hereafter, p27) were upregulated in suspension(Figure 1b). However, whereas p27 reached maximum levels lateduring differentiation, p21 was transiently induced by the time ofcommitment (6 h).

We next analysed the dynamics of Cyclin/cdk complexes byimmunoprecipitation (IP) and western blotting (WB; Figure 1e andSupplementary Figure 4a). Cyclin A bound to both cdk2 and cdk1(Figure 1e). By 12 h, the detection of both kinases in Cyclin A IPswas much reduced and by 24 h, only the inactive form of cdk1was barely detectable (Figure 1e). Cyclin B/cdk1 peaked by 6 h.Increasing levels of cdk2 were detected in Cyclin E IPs up to 24 h(Figure 1e; PSTAIR is a homologous motive of the cdks; Cyclin Ebinds to cdk2). It is worth noting that the binding of the cdkinhibitor p21 to Cyclin B complexes peaked by 6 h whereas the

binding of p21 to Cyclin E complexes peaked by 12 h (Figure 1e).Reverse experiments by pulling down p21 showed consistentresults (Figure 1e).

p21 was barely detectable in Cyclin E IPs after 24 h, suggestingthat most of the Cyclin E protein was free of the inhibitor(Figure 1e). Conversely, IP for p21 revealed a neat band of Cyclin Eby 24 h, suggesting that most of the p21 protein was bound toCyclin E/cdk2 (Figure 1e). In summary, the transient increase ofp21 appears to suffice to inhibit the mitotic Cyclin/cdk complexesby 6 h, but not the high levels of Cyclin E that are subsequentlyproduced. Consistently, immunodepletion of p21 in cell lysatesremoved most of the Cyclin A/cdk complex by 6 h, but leftsignificant levels of Cyclin E/cdk2 complex by 12 h (SupplementaryFigure 4a). Interestingly, downregulation of Cyclin A in suspendedmouse keratinocytes requires p21.17

Cdk complexes of S-phase and mitosis during MYC-induceddifferentiationConstitutive activation of proto-oncogene MYC drives epidermalstem cells into the clonal amplification phase (referred to as transitamplifying cells or TAC) that precedes terminal differentiation.7,8

Once terminal differentiation initiates, MYC is downregulated(Figure 2a).19,20 Keratinocytes expressing MYCER constitute an

Figure 2. Dynamics of cdk complexes of S-phase and mitosis during MYC-promoted differentiation. (a) WB for the regulators indicated on theleft of parallel samples from keratinocytes in suspension or after activation of MYCER by OHT, for the periods of time indicated. Invol bands aredue to differentiation crosslinking. Regulators abbreviated as in Figure 1; broken lines: phase of rapid cell amplification. (b) Representativeanalyses of kinase activity on histone H1 associated with the regulators indicated, after activation of MYCER as in a; bar histogram shows thequantitation of duplicate samples; small bars are s.e.m. (c) Expression of the cell cycle inhibitors indicated after activation of MYCER as in a; (d)Or in freshly isolated keratinocyte populations from human skin enriched for stem cells (SC), rapidly amplifying cells (TAC), or terminallydifferentiating cells (TDC); MIX: unsorted. Results with cells from two different individuals. Bars in c and d indicate quantitation (% of total) ofthe different mobility forms detected for Rb: hypophosphorylated (white; active, a), phosphorylated (grey), hyperphosphorylated (black;inactive, i). C, negative antibody control on adherent cells; E, exponentially growing HDF; M, human dermal fibroblasts (HDF) blocked inmitosis by Nocodazole for 24 h.


3

Oncogene (2012), 1 -- 13& 2012 Macmillan Publishers Limited

inducible differentiation system in the presence of cell adhesionthat is more gradual than the suspension method. An increase inthe proportion of terminally differentiating cells is observed over 5days after the activation of MYCER, the time estimated for stemcells to amplify before differentiation (see involucrin inFigure 2a).7,9 We have examined the dynamics of cdk complexesduring MYC-promoted differentiation. As the induction ofdifferentiation by MYC is more gradual than in the suspensionsystem, the changes were not as marked but were consistent. Atransient increase of the expression of Cyclins A and B wasobserved over 2 -- 3 days after MYCER activation (Figure 2a andSupplementary Figure 4c). In contrast, MYC continuously inducedthe expression of Cyclin E (Figure 2a). This is consistent withprevious reports showing that MYC induces Cyclin E expres-sion.21,22 The kinase activities associated with Cyclin A, Cyclin Band cdk2 increased over the first 2 days and dropped by 7 daysbelow the initial levels (Figure 2b). Cyclin E-associated activity alsoincreased and decreased to levels similar to the start of theexperiment (Figure 2b). All regulators thus reflected an earlyactivation of the cell cycle by MYC. Consistently, the hyperpho-sphorylated (inactive) form of Rb became predominant by 6 -- 12 hand phospho-Rb Ser780 and phospho-cdk2 Y15 increased afterMYCER activation (bars, Figure 2c and Supplementary Figure 4b).The inhibition of the kinase activities after 3 days might be due tothe striking transient induction of total p21 levels (Figure 2c). p21bound to both Cyclin A and Cyclin E complexes in the presence ofactivated MYC (Supplementary Figure 4c). In contrast, p27 wasundetectable in the MYCER keratinocytes (Figure 2c).

Altogether, the results suggest that the transient induction ofp21 might be a keratinocyte response to cell cycle deregulation.Expression of p21 has been previously described in theepidermis.23 To further study this issue, we isolated keratinocytesfrom the normal human skin and selected stem cells, transitamplifying cells and terminally differentiating cells on the basis oftheir adhesion capacity via integrins.14 Strong expression of p21was detected in the slowly adhering population, previouslydefined as transit amplifying cells (TAC; Figure 2d). The presenceof p21 was again accompanied by a higher phosphorylation of Rb(Figure 2d). These results were therefore consistent in suspension-induced differentiation, MYC-induced differentiation and sponta-neously occurring differentiation.

Mitosis block and differentiationThe results reported above suggest that keratinocyte differentia-tion involves mitotic inhibition and cell cycle deregulation. Tofunctionally explore this issue, we treated normal primarykeratinocytes with cdk inhibitors in order to block either DNAreplication cdk2 or mitosis cdk1. The specificity of these inhibitorson cdk2 or cdk1 vary depending on the concentration and the celltype.24 However, AG555 was identified as a cdk2 inhibitor.25,26

Roscovitine and NG97 were selected by searching for inhibitors ofCyclin B/cdk1 and have been shown to block the mitosismachinery.27 -- 29 As it is not technically possible to test the activityof the inhibitors in vivo, we tested their capacity to inhibit S-phaseor mitosis by analysing their effect on DNA replication and cellcycle.30,31 Treating primary keratinocytes with AG555 for 48 h atthe concentration used in Figure 3a significantly inhibitedprogression through S-phase, causing an accumulation of cellsin G1/S and a reduced G2/M phase. In contrast, althoughRoscovitine and NG97 have the capacity to inhibit both cdk1and cdk2, they provoked a marked accumulation of keratinocytesin G2/M and polyploidy and yet allowed DNA replication(Figure 3a and Supplementary Figures 5 and 6). AG555 slightlyinhibited cell size and terminal differentiation as measured bylight scattering, involucrin expression and cellular morphology(Figure 3a and Supplementary Figures 5 and 6). In contrast, bothRoscovitine and NG97 strongly induced basal cells to undergo

terminal differentiation by 48 h, as it was evident by lightscattering, involucrin expression and cell morphology (Figure 3aand Supplementary Figures 5 and 6).

To further determine whether inhibition of cdk1 triggeredterminal differentiation, we expressed specific short hairpin RNAs(shRNAs) in primary keratinocytes by lentiviral infection. Twodifferent shRNAs were able to inhibit the expression ofendogenous cdk1 as analysed by real-time reverse transcrip-tion -- polymerase chain reaction (RT -- PCR; Figure 3b). The cellcultures showed areas of rounded cells typical of mitosis arrest(Supplementary Figure 7). Inhibition of cdk1 provoked an increasein the proportion of cells expressing the terminal differentiationkeratins K1 and K10 as measured either by RT -- PCR or bypercentage of positive cells (Figure 3b).

Genotoxic agents might cause keratinocyte differentiation bytriggering a mitosis checkpoint.11 Mitosis kinases Aurora B orPolo-like kinase 1 (Plk1), which are involved in chromosomesegregation, participate in the mitotic spindle checkpoint32,33 andsome of their inhibitors are being tested in cancer clinicaltrials.34 To explore whether the keratinocyte mitotic checkpointsare involved in the signal to differentiation, we inhibitedthese molecules by specific chemical inhibitors. A 48 h treatmentwith Aurora B kinase inhibitor, ZM44743935 or Plk1 inhibitor,BI253636 strongly induced an accumulation of keratinocytes in G2/M and polyploidy and a remarkable increase of cell size anddifferentiation (Figure 4a and Supplementary Figures 5 and 6).

To confirm the specificity of the effects, we inhibited Aurora B inkeratinocytes by transfection of specific small interference RNA(siRNA). Transient inhibition of Aurora B produced 24 h aftertransfection a significant increase in the proportion of cellsexpressing the differentiation marker involucrin (Figure 4b).

The induction of terminal differentiation by the mitoticinhibitors was irreversible as cells did not recover colony growthafter being released from the 48 h treatment with Roscovitine orBI2536 (Figure 4c and Supplementary Figure 6). In contrast, thecolonies grew bigger after the G1/S block by the AG555 inhibitor(Figure 4c and Supplementary Figure 6).

Overexpression of Cyclin E-GFPWe have shown that epidermal differentiation is associated with amarked accumulation of Cyclin E in vitro (Figures 1a and 2a andSupplementary Figures 5 and 6) and in vivo.11 To test whether adirect hyperactivation of the cell cycle is able to drive differentia-tion, we constructed a retroviral green fluorescent protein (GFP)form of Cyclin E under a constitutive retroviral promoter for itsexpression in basal primary keratinocytes (Figures 5a and b andSupplementary Figure 8a). In vitro assays showed that Cyclin E-GFPwas functional in activating cdk2 (Supplementary Figure 8b andnot shown). The exogenous Cyclin E-GFP localised to the nucleusand was sensitive to posttranscriptional regulation, as it wasabsent in metaphasic cells (Supplementary Figure 9a, INSET), asdescribed for the wild-type protein.12 Keratinocytes overexpres-sing Cyclin E-GFP (KCEGFP) were significantly larger and presentedlarge or various nuclei (Figure 5b and Supplementary Figure 9a).KCEGFP cultures widely expressed the terminal differentiationmarker involucrin (Figure 5b).

Flow-cytometry analyses of unselected cultures after infectionwith Cyclin E-GFP (GFPþ /�; Figure 5c) or stably selected bypuromycin (KGFP/KCEGFP; Supplementary Figure 9b) showed thatCyclin E caused an accumulation of cells in the G2/M phase of thecell cycle and induced a significant proportion of polyploid cells. Inaddition, overexpression of Cyclin E induced a significant increase inthe rate of large cells (High Scatter) and involucrin expression(Figure 5c). Overexpression of Cyclin E also induced an increase inthe rate of DNA synthesis, consistent with cell cycle activation(Figure 5d and Supplementary Figure 9c), and a decrease in Cyclin A,consistent with terminal differentiation (Figure 5d; see Figure 1a).


4


Overexpression of Cyclin E caused a reduction in the clonogenicamplification capacity (Figure 6a). This reduction was not due tosenescence (Supplementary Figure 10) or apoptosis (no sub-G1 inthe DNA content profiles; Figure 5c and Supplementary Figure 9b).Basal KCEGFP were able to attach and to proliferate but formed ahigher proportion of abortive-differentiated colonies (Figure 6a).The increased terminal differentiation rate was further confirmedby a higher expression of keratins K1 and 10 (Figure 6b and

Supplementary Figure 9d) and by assays on the capacity to formcornified envelops, the final product of epidermal differentiation(Figure 6c).37 Therefore, all the parameters analysed show that theactivation of the cell cycle by Cyclin E in basal keratinocytes resultsin terminal differentiation.

MYC has been shown to induce Cyclin E.21,22 Activation ofconditional MYCER significantly induced Cyclin E in keratinocytes(Figures 2a and 6d). We questioned whether Cyclin E might

Figure 3. Changes in cell cycle and differentiation after treating keratinocytes with cdk inhibitors. (a) Representative flow-cytometry analysesof DNA content, DNA synthesis (BrdU incorporation), cell size and morphology (light scattering) and involucrin expression after treatingkeratinocytes for 48 h with the cdk inhibitors 50 mM AG555 (AG), 30mM Roscovitine (ROSCO; Supplementary Figure 5), or 1 mM NG97 (NG). Barhistograms: quantitation of BrdU-positive cells, polyploid cells (44N) or differentiating cells (High Scatter). Numbers are from the regionsshown in the histograms above, means of duplicates and representative of experiments with cells from two different individuals. The circle inthe morphology histograms indicates the proliferative basal population. The involucrin positive region (p) was determined by a negativeisotype antibody control (CD8). Bars are s.e.m. Ctr: control cells. (b) Inhibition of cdk1 by lentiviral infection with shRNA 2. Top histograms:expression of cdk1 or differentiation markers K1 and K10 as measured by RT--PCR in control cells (Ctr, white) or in the presence of shRNA tocdk1 (grey). Bottom histogram: percentage of keratin K1-expressing cells. Microphotographs: expression of K1 and nuclear staining by DAPI ofthe same field, as indicated.


5


Figure 4. Changes in cell cycle and differentiation after inhibiting keratinocyte mitosis kinases of the spindle checkpoint. (a) Representativeflow-cytometry analyses of DNA content, DNA synthesis, morphology and involucrin expression as in Figure 3a, after treating keratinocytes for48 h with inhibitors of Aurora B kinase (AURB), ZM447439 (ZM; 2 mM) or Polo-like kinase 1, BI2536 (BI; 100 nM ; Supplementary Figure 5). Barhistograms and errors as in Figure 3a. Representative of experiments with cells from two different individuals. (b) Effects 24 h after transfectionwith siRNA to downregulate Aurora B Kinase (siAURB). Bar histograms: expression of AURB as measured by RT--PCR, or percentage ofinvolucrin positive cells by immunofluorescence and cell counting. Ctr: non-targeting control siRNA. Bars are s.e.m. Microphotographs showrepresentative images of immunofluorescent staining for involucrin (Invol; red) and DAPI (nuclear DNA, blue). Scale bar: 50 mm.(c) representative samples of clonogenicity capacity of keratinocytes released after a 48 h treatment with the kinase inhibitors indicated. Dataare from experiments of duplicate samples with cells from two different individuals. Ctr, control; ROSCO, Roscovitine.

Figure 5. Overexpression of Cyclin E-GFP in basal primary keratinocytes. (a) Detection of the Cyclin E-GFP fusion protein (arrowhead) by WBwith antibodies to Cyclin E (CE) or to GFP, as indicated on the left in keratinocytes infected with the GFP-only vector (KGFP) or in keratinocytesinfected with the Cyclin E-GFP construct (KCEGFP). (b) Upper panels: microphotographs of phase contrast of the cells indicated. Note thatCEGFP keratinocytes are significantly larger and they have large or various nuclei. Bottom panels: Immunofluorescence for CE (in green) andinvolucrin (Invol, in red); DAPI in blue. Bars: 100 mm. Ctr: uninfected keratinocytes. (c) Representative flow-cytometry analyses for lightscattering (SSC-A) versus GFP (upper-left dot plot, GFP positive cells in green), involucrin versus GFP (upper-right dot plot, GFP positive cells ingreen), DNA content (propidium iodide), morphology (Light Scatter) or involucrin as indicated, in GFP expressing (GFPþ ) or non-expressing(GFP-) keratinocytes. Bar histograms show the quantitations of the percentage of cells in G2/M, polyploidy, with differentiated morphologicalfeatures (large and complex cells; High Scatter) or positive for involucrin (Invol), as indicated. Regions and controls as in Figure 3a.(d) Expression of Cyclin E and Cyclin A as measured by RT--PCR (CE, CA) and percentage of positive cells for DNA replication, as measured byBrdU-incorporation as in Figure 3a; KGFP (white); KCEGFP (grey). Numbers are means of duplicate samples, representative of experiments withcells from two different individuals. Small bars are s.e.m.


6



7


contribute to the differentiation effect produced by MYC. To thisend, we transduced MYC in keratinocytes in combination withshRNAs to Cyclin E. As shown in Figures 6e and f, inhibition ofCyclin E by shRNAs attenuated the increase of keratinocytedifferentiation markers induced by MYC.

Cyclin E and DNA damageWe have shown that triggering the spindle mitosis checkpointcauses in keratinocytes mitotic failure and differentiation (Figure 4and Supplementary Figures 5 and 6). This effect is also provokedby genotoxic agents such as bleomycin or a topoisomerase IIinhibitor.11 We investigated whether overexpression of Cyclin Ecaused DNA damage by measuring the phosphorylation ofhistone H2ax (P-gH2ax), an early marker of DNA repair.38 Assummarised in Figures 7a and b, KCEGFP strikingly accumulatedDNA damage as compared with control keratinocytes over-expressing GFP only. Interestingly, ectopic Cyclin E induced theexpression of the p53/p21 pathway (Figure 7c). The p53 pathwayis induced in response to DNA damage and triggers mitoticcheckpoints.39

We next tested whether an accumulation of DNA damagemight trigger the mitosis block and terminal differentiation effectsobserved in KCEGFP cells. To this aim, we treated keratinocytes

with doxorubicin (DOXO), a known DNA-damaging agent.40 DOXOtreatment strongly induced DNA damage by 24 h (P-gH2ax;Figure 8a), the p53/p21 pathway (Figure 8b) and a mitosis arrest(Figure 8c). As a result, keratinocytes were rapidly induced toundergo terminal differentiation (Figure 8c), resembling theeffects observed after a direct mitosis block (Figures 3a and 4a).Interestingly, inhibition of the mitotic checkpoint componentAurora B kinase by ZM447439 did not induce p53 as significantly(ZM; Figure 8b), consistent with the model that p53 acts upstreamof the mitosis block (Figure 8d).

DISCUSSIONThe cell cycle deregulates prior to differentiationIt is unclear how keratinocytes coordinate cell multiplication withpostmitotic differentiation. Integrins maintain both cell adhesionto the basement membrane and keratinocyte proliferation.1 Wehere have shown that primary keratinocytes continue DNAsynthesis after loss of anchorage, as they commit to terminaldifferentiation.9 While differentiation progressed, keratinocytesaccumulated in G2/M and became polyploid. Consistently, wehave found that keratinocytes continue DNA replication in thedifferentiating layers of the human epidermis.11

Figure 6. Effects of ectopic Cyclin E on keratinocyte growth and differentiation. (a) Clonogenic capacity of control KGFP or KCEGFP. 2000 cellsof either total (upper dish) or basal (lower dishes) populations were plated per well (see Materials and methods). Bar histograms represent thepercentage of actively growing colonies (grey) versus the percentage of abortive, differentiated colonies (white). Dotted line indicates theproportion of total colonies produced per thousand cells plated � 10. Representative of triplicate samples from experiments with cells fromtwo different individuals. (b) mRNA expression of differentiation markers K1, K10 or involucrin as measured by RT--PCR in KGFP (white bars) orKCEGFP (grey bars) cells cultured in low calcium medium. (c) Quantitation of the production rate of cornified envelopes by control KGFP(white bars) or KCEGFP (grey bars). (d) Immunofluorescence for Cyclin E after activation of MYCER by OHT for 4 days. (e) Detection of MYCGFP,Cyclin E (CE) or involucrin by RT --PCR in primary keratinocytes, uninfected or infected with MYCGFP and CE-shRNA1 or CE-shRNA2 asindicated. (f ) Microphotographs: staining of the differentiation marker K1 in cells overexpressing MYCGFP in the absence or presence of CE-shRNA1; scale bars, 50 mm. bar histogram: quantitations of the percentage of K1-expressing cells are data from the two independentexperiments.


8


Although the activities of cell cycle kinases cdk2 (S phase) andcdk1 (mitosis) were both inhibited in suspended keratinocytes,a remaining low-intensity Cyclin E-associated activity persistedduring differentiation. The decrease of the kinase activities insuspended keratinocytes was likely elicited by cell cycle inhibitorssuch as p21 or p27. Induction of p21 during keratinocytedifferentiation has been reported previously.16,17,23,41 In ourexperiments, p21 was transiently induced in suspended keratino-cytes and bound Cyclin A/cdk1 and Cyclin B/cdk1 at the time ofthe commitment to differentiation (6 h), whereas the binding tocdk2 peaked later on. Interestingly, p21 is required for thedownregulation of Cyclin A in mouse keratinocytes.17 Althoughp21 inhibits cdk2 complexes, it is able to bind and inactivate themitosis complex Cyclin B/cdk1 and yet allow DNA replica-tion.39,42,43 The induction of p21 in suspended keratinocytesmight not be sufficient to completely inhibit the large amounts ofCyclin E/cdk2 produced, but it might be sufficient to inhibit thedecreased amounts of Cyclin A and Cyclin B. Consistently, thelevels of cdk2 required to drive DNA replication appear to belower than the levels of cdk1 required to trigger mitosis.44,45 Thelow activity of Cyclin E/cdk2 might thus sustain keratinocyte DNAre-replication once differentiation has initiated. Within the samelines, cdk inhibitors Roscovitine and NG97 blocked mitosis and yetallowed DNA replication.

Overactivation of MYC in keratinocytes induced cell cyclechanges consistent with the suspension system. The changes weremore marked in suspension-induced differentiation, as it is a rapidprocess that includes complete lack of cell adhesion. In contrast,ectopic activation of MYC is thought to drive keratinocytes intothe clonal amplification phase of rapid proliferation that precedes

terminal differentiation.7,8 Activation of MYCER in keratinocytesfor 2 days caused hyperphosphorylation of Rb and a transientinduction of the Cyclins, all changes that mark cell cycle entry.Subsequently, ectopic MYC induced p21 and downregulation ofmitotic Cyclins A and B. These changes are intriguing, as MYC hasbeen reported to inhibit p21 expression and to induce Cyclin B.46 -- 48

However, our results are consistent with the model that MYCfunction depends on the cell type and the cell context.10,49,50

Interestingly, induction of p21 has also been found in MYCER-induced senescence in the absence of cdk2.51 The upregulation ofp21 in keratinocytes was concomitant with the cell cycleactivation by MYC and it was also found in the rapid proliferativepopulation isolated from human basal epidermis.

Activation of MYC in keratinocytes induced expression ofCyclin E, consistent with previous reports in other cell types.21,22

The regulation of Cyclins during suspension-induced keratinocytedifferentiation is consistent with what we have observedin suprabasal epidermis11 and with the modified cycle that occursin endoreplication.52 -- 54 Some of these changes have alsobeen observed in keratinocytes harbouring HPV molecules.55

Postmitotic DNA replication allows a large increase in cellsize in endoreplicating systems, including plant epidermis (forexample, Zanet et al.,10 Schellmann and Hulskamp54 andreferences therein). The switch from proliferative Cyclins to highCyclin E observed in our experiments correlated with theinduction of the differentiation marker involucrin and mightexplain the cell size increase of differentiating keratinocytes.3,4

This in turn reconciles the function of MYC in promotingepidermal differentiation with its well-established function indriving cellular growth.

Figure 7. Effects of ectopic Cyclin E on DNA damage and the expression of the p53 pathway. (a) Expression of the DNA repair marker P-gH2axas assessed by immunofluorescence in stably selected KGFP or KCEGFP cells; bars: 100 mm. (b) Percentage of positive cells for P-gH2ax scoredin: i--- the GFP-negative or -positive populations within the same cultures 48 or 72 h after infection with CEGFP; or ii--- in cells stably expressingGFP or CEGFP (ST); Ctr: uninfected cells. (c) WB for expression of the regulators indicated on the left in uninfected, KGFP or KCEGFP cells. UI,uninfected keratinocytes. b-actin (b-act) was used as the loading control.


9


All the cell cycle regulators analysed increased early after loss ofanchorage or MYC activation as keratinocytes committed todifferentiation. This included Cyclin/cdk activities and phosphor-ylation of Rb. In addition, we and others have shown that p53antagonist MDM2 is upregulated during keratinocyte differentia-tion (Dazard et al.41 and references therein). Altogether, the resultsshow that there is a loss of inhibition in the S-phase of the cellcycle as keratinocytes commit to differentiation and block mitosis(Figure 8d). This response to lack of anchorage might account forthe loss of proliferative potential as keratinocytes lose adherenceto the basal membrane via integrins.5

A mitosis-differentiation checkpointThe inhibition of mitosis after loss of cell anchorage or MYCactivation in keratinocytes might be a response to counterbalance

cell cycle hyperactivation. In our experiments, a block of mitosis byinhibiting cdk1 or components of the spindle checkpoint Aurora Bkinase or Polo-like kinase, or by causing genotoxic insult bydoxorubicin (herein), Bleomycin or topoisomerase II inhibitor ICRF-193,11 triggered a rapid increase of the keratinocyte cell size anddifferentiation. This striking rapid response was not observedwhen keratinocytes were treated with inhibitors of G1/S, AG555(herein) or hydroxyurea.9 In contrast, the clonogenic potential ofkeratinocytes augmented once they were released from a G1arrest, induced by the cdk inhibitor AG555. This inhibitor slightlyreduced cell size and involucrin expression. Along the same lines,overexpression of p21 in keratinocytes slowed down the cell cycleand inhibited differentiation.16,23 Consistently, inhibition of CyclinE by shRNAs, in our experiments, attenuated the differentiationeffect caused by MYC. The results altogether suggest thatacceleration of the cell cycle supports and promotes keratinocyte

Figure 8. Effects of DNA damage on primary keratinocytes. (a) Immunofluorescence for the DNA repair marker P-gH2ax in control (DMSOtreated) keratinocytes, or keratinocytes treated with doxorubicin (DOXO) for 24 h; scale bar, 50 mm. (b) WB for expression of the regulatorsindicated on the right side in keratinocytes treated with DMSO, DOXO, or the Aurora B kinase ZM447439 (ZM) for 48 h. GAPDH (GDH) was usedas the loading control. (c) Representative flow-cytometry analyses of DNA content, involucrin expression and morphology after treatingkeratinocytes with DOXO for 48 h; top bar histograms: quantitations of the percentage of involucrin-positive cells, morphologicallydifferentiated cells (High Scatter) and G2/M cells; regions and controls as in Figure 3a. Numbers are means of duplicates, representative ofexperiments with cells from two different individuals. Bars are s.e.m.; botton bar histograms: expression of mRNA of differentiation markersfilaggrin and involucrin as determined by RT--PCR after a 24 h treatment with DMSO (Ctr, white bars); or with DOXO (grey bars). (d) Model forthe sequence of events following a hyperactivation of the keratinocyte cell cycle: cell cycle deregulation leads to accumulation of DNAdamage and induction of the p53/pathway; the mitosis checkpoints block mitosis and trigger terminal differentiation.


10


differentiation (Figure 8d). The model is consistent with the notionthat epidermal stem cells are mainly quiescent, and that theiractivation already initiates the differentiation programme. Themodel is in turn consistent with the fact that MYC promoteskeratinocyte differentiation.

An uninhibited cell cycle might be a part of the rapidproliferation phase that precedes keratinocyte terminal differen-tiation. As a response, a keratinocyte mitosis checkpoint mighttrigger differentiation. This checkpoint might contribute tomaintain homeostasis and be part of a tissue self-defencemechanism in the case of genotoxic stress or oncogenicactivations, similar to the role that apoptosis or senescence hasin other systems.56, 57 These processes have been found to sharesome mechanisms.58 However, MYC has pleiotropic effects in cellmetabolism and growth.6 To challenge the model and to testwhether a direct activation of the cell cycle would trigger adifferentiation response, we overexpressed Cyclin E in basalkeratinocytes.

Cyclin E makes the linkCyclin E accumulates during keratinocyte differentiation bothin vitro (herein) and in human epidermis.11 Overexpression ofCyclin E-GFP induced terminal differentiation of proliferativekeratinocytes although the exogenous fusion protein wasdegraded at mitosis as the natural protein.12 Despite the increasein DNA replication index upon ectopic Cyclin E, the clonogeniccapacity diminished and cells accumulated in G2/M andpolyploidy owing to mitosis failure. The loss of proliferativecapacity caused by ectopic Cyclin E overexpression was notassociated with senescence or apoptosis (see also Gandarillas58)and it was more marked than the inhibition caused by over-activation of MYC.7 Interestingly, sustained Cyclin E expressionresulted in accumulation of DNA damage and induction of p53/p21, a pathway known to halt the cell cycle in G2/M in response toDNA damage (Figure 8d).39 Interestingly, p53 also increased afteractivation of MYCER in keratinocytes and in the rapid-proliferativefraction of basal human epidermis.41

Cyclin E is involved in the activation of DNA replication originsand it must be downregulated for cells to undergo cell cyclearrest.12 Premature Cyclin E expression accelerates cell cycle entry.In addition, oncogenes might promote genomic instability bycausing DNA damage.57 Recent reports have also concludedthat acceleration of the cell cycle by Cyclin E can causeDNA damage.59,60 Interestingly, in our experiments, causing DNAdamage in keratinocytes with genotoxic agents such as doxor-ubicin induced p53 and terminal differentiation. As discussedabove, we have shown that a direct block of mitosis rapidly resultsin terminal differentiation. By causing DNA damage, accumulationof Cyclin E might trigger the keratinocyte mitosis checkpoints andin turn, terminal differentiation (Figure 8d). This would create alink between rapid cell amplification and the commitment toterminal differentiation, thus constituting an automatic compen-satory mechanism to maintain homeostasis. This model wouldcontribute to explaining how the epidermis maintains tissuestructure in benign hyperproliferative situations. Additionalalterations breaking the mitosis block response might be requiredto promote carcinogenesis.

The differentiation effect induced by deregulated Cyclin E doesnot only have implications for the mechanisms protecting theepidermis against hyperproliferative alterations. For long, a cell-autonomous clock-like mechanism has been proposed as one ofthe possible models determining stem cell fate. The levels ofCyclin E drive cell fate in Drosophila oogenesis.53 A crossroadbetween mitosis, differentiation and DNA endoreplication isimportant in a variety of cell systems, including the plantepidermis or human megakaryocytes.52 -- 54 In myeloid leukaemiacells, Cyclin E drives endoreplication61 and Myc and p21 cooperate

to induce megakaryocityc differentation and polyploidy.50 Onco-gene-induced accumulation of DNA damage by replication stressmight trigger a similar response leading to apoptosis orsenescence. The mechanisms we report here might thereforehave a more global role in homeostasis beyond the epidermis.

It is tempting to speculate that within the epidermis, athreshold of S-phase regulators, such as Cyclin E during the cellcycle hyperactivation that precedes terminal differentiation, mightlimit the number of keratinocyte divisions and thus controlepidermal cell fate as a clock.

MATERIALS AND METHODSCell culture, plasmids and antibodiesPrimary keratinocytes were isolated from neonatal human foreskin andcultured in the presence of a mouse fibroblast feeder layer (inactivated bya 2 h treatment with 4 mg/ml mitomycin C), 10% serum and 1.2 mM Ca2þ ,as described.62 Low passages (1 -- 5) of keratinocytes from four differentindividuals were utilised. Terminal differentiation was induced bysuspending disaggregated keratinocytes in culture medium supplementedwith methyl cellulose (Sigma-Aldrich, Inc., St Louis, MO, USA) asdescribed.5,41 For some studies keratinocytes were transferred to low-calcium (o0.1 mM) Defined Keratinocyte-Serum-Free-Medium (Invitrogen,Carlsbad, CA, USA), following the manufacturer’s instructions: (i) siRNAtransfections (small interfering RNA); (ii) shRNA (small hairpin RNA)lentiviral infections; and (iii) quantitation of the expression of keratins K1and K10 by immunofluorescence or RT -- PCR.

Activation of conditional MYCER63 in primary keratinocytes wasachieved by addition of 100 nM 4-hydroxytamoxifen (OHT; variant Z,Research Biochemicals International, Natick, MA, USA) to the culturemedium for the lengths of time indicated.

Stem cells (SC), ‘transit amplifying cells’ (TAC) and terminally differentia-tion cells (TDC), were freshly isolated from human foreskin and sorted onbasis of their adhesion capacity on collagen IV, as described.14,41

Primary keratinocytes were treated for 24 or 48 h with cdk inhibitors ormitosis kinase inhibitors: 50mM of AG555 (ICN Biochemicals, Aurora, OH,USA), 30 mM of Roscovitine (Sigma-Aldrich, Inc.) and 1 mM of NG97 (kind giftof J Meijer) in DMSO; 2 mM of the Aurora B kinase inhibitor ZM447439(Tocris Bioscience, Bristol, UK) or 100 nM of the Polo-like kinase inhibitorBI2536 (Axon MedChem BV, Netherlands) in DMSO. For DNA damageanalyses cells were treated for 24 or 48 h with 0.5mM of DOXO (Sigma-Aldrich, Inc.) in DMSO. Parallel control cultures were always subjected tothe DMSO vehicle. Optimal concentrations were tested for each inhibitor.

For clonogenicity assays, 5000 or 10 000 of total or basal keratinocyteswere plated per T6 well triplicates. Basal cells were counted on basis oftheir size (o18 mm). After 12 -- 15 days, the cultures were coloured withRhodanile blue as described.10

Keratinocytes were subjected to retroviral infection with pBabeconstructs as described.7 pBabe-GFP-CyclinE was constructed to expressCyclin E-GFP fusion protein in basal primary keratinocytes from pBabe-li-GFP and pBabe-CyclinE (kind gifts from Pierre Roux and Bruno Amati) asdescribed in Supplementary Figure 8. The studies were carried out eitherby analysing unselected pools 2 -- 5 days after infections or by stablyselecting GFP or CEGFP expressing keratinocytes by 1 mg/ml puromycin.

The production of cornified envelopes was measured as described inSun and Green.37

The following antibodies were used: anti-Cyclin A2 (CY-A1, Sigma-Aldrich, Inc.; immunofluorescence, IF and western blotting, WB), anti-CyclinA2 antibody (H432, Santa Cruz Biotechnology, Santa Cruz, CA, USA; IP),anti-Cyclin B1 (GNS1, Santa Cruz Biotechnology; IF, IP and WB), anti-CyclinE1 (C19, Santa Cruz Biotechnology; IF, IP and WB), anti-Cyclin E1 (HE12,Santa Cruz Biotechnology; IF and WB), anti-BrdU (BD Biosciences, San Jose,CA, USA; flow cytometry, FC and IF), anti-cdk1 (p34cdc2, Sigma-Aldrich,Inc.; WB), anti-cdk2 (M2, Santa Cruz Biotechnology; WB), anti-phospho-cdk2(Y15, Abcam, Cambridge, UK; WB), anti-GFP (Invitrogen; FC, IF and WB),anti-ER (Santa Cruz Biotechnology; WB), anti-phospho-Histone H2AX(Ser139, JBW301, Millipore, Bedford, MA, USA; IF), anti-phospho-H3 (Ser-10, Santa Cruz Biotechnology; WB), anti-involucrin (SY5, Sigma-Aldrich, Inc.;


11


FC, IF and WB), anti-K1 (Covance, Vienna, VA, USA; IF), anti-MYC (Santa CruzBiotechnology; WB), anti-p21CIP (C19, Santa Cruz Biotechnology; Depletionassay and WB), anti-p21CIP (CP74, Sigma-Aldrich, Inc.; WB), anti-p27(F8, Santa Cruz Biotechnology; WB), anti-p53 (FL-393, Santa CruzBiotechnology; WB), anti-PSTAIR (pstair, Sigma-Aldrich, Inc.; WB), anti-Rb(C15, Santa Cruz Biotechnology; WB), anti-phospho-Rb (ser780, CellSignalling Technology, Beverly, MA, USA; WB), anti-b-actin (I-19, SantaCruz Biotechnology WB), anti-GAPDH (FL-335, Santa Cruz Biotechnology;WB) and anti-CD8 ( Sigma-Aldrich, Inc; FC).

Flow cytometryAfter removal of the mouse feeder layer, 106 trypsinized keratinocyteswere fixed, stained for BrdU and DNA (propidium iodide), for GFP or forinvolucrin, and analysed by flow cytometry as described in SupplementaryData. All antibody stainings were controlled by the use of isotype negativeimmunoglobulins (mouse anti-cd8 or rabbit serum). In order to gate outcell aggregates, the area of the fluorescent pulse of propidium iodide (DNAcontent; PE-A) was plotted versus the width of the fluorescent pulse (PE-W;see Supplementary Figure 11).

ImmunofluorescenceKeratinocytes were grown on glass coverslips and fixed and stained asdescribed in Supplementary Data.

Whole-cell extracts and WBCells were washed twice with phosphate-buffered saline and incubated for30 min on ice in lysis buffer. Cell lysates were processed and subjected toWB for the molecules indicated as described in Supplementary Data.

IP and in vitro kinase assaysCell lysates (200mg) were incubated with specific antibodies to IP cell cycleregulators as described in Supplementary Data. The IPs were subsequentlysubjected to WB to determine interacting proteins or the associated kinaseactivity as described in Supplementary Data.

siRNA transfection assays and shRNA infectionsTransfections with siRNA were performed on primary keratinocytes asdescribed previously64 using Stealth siRNA duplex for Aurora B Kinase(Invitrogen) and for a non-targeting random control siRNA. Finally, 24 hafter transfection the expression of the gene was measured by RT -- PCR asdescribed below.

To inhibit endogenous regulators and to overexpress MYC, the followinglentiviral plasmids were acquired: control shRNA plKO1-TRC1 fromAddgene (plasmid 10879, Cambridge, MA, USA), pLKO.1 cdk1 shRNA andCyclin E shRNA from Sigma-Aldrich, Inc. (Mission shRNA) and MycGFP fromGeneCopoiea (EX-Z2845-Lv103; Rockville, MD, USA). Lentiviral productionwas performed by transient transfection of producer 293T cells usingjetPEIt as indicated by the manufacturer (Polyplus-Transfection, Illkirch,France). After 48 h, the viral supernatant was added to primarykeratinocytes cultured in DSFM serum-free medium (completed with8mg/ml Polybrene; Sigma-Aldrich, Inc.). The medium was changed 4 h afterinfection and the experiments were carried out 5 days after infection.

Reverse transcription -- polymerase chain reactionTotal RNA was isolated using the Trizol Reagent (Invitrogen) according tothe manufacturer’s instructions. In all, 2 mg of total RNA was reverse-transcribed with the iScriptt cDNA synthesis kit (Bio-Rad, Hercules, CA,USA) in a 20-ml reaction for 30 min at 42 1C. The cDNAs (2ml) wereamplified by real-time PCR (Bio-Rad iQt SYBR Green supermix) usingb-actin or S14 mRNA levels to normalise the data (Supplementary Table 1).

Cellular b-galactosidase activitySenescence was assessed by analyses of endogenous b-galactosidaseactivity and visualised by brightfield microscopy.65

CONFLICT OF INTERESTThe authors declare no conflict of interest.

ACKNOWLEDGEMENTSWe are especially grateful to Jean-Claude Rossi and Javier Leon for their strongsupport. We thank Bruno Amati, Pierre Roux and Jonathon Pines for providing DNAconstructs, J. Leon for reagents, Anne Blangy, Vjeko Dulic, Veronique Baldin andThierry Lorca for technical advice, Daniel Fisher for helpful suggestions, Kathy Brooksand Marıa Aramburu for technical support at flow cytometry, J. Leon and PhilippePasero for critical reading of the manuscript, and Anna Solinıs and Diana Solinıs forreviewing the English writing. Human skin biopsies were provided with consent by StJean and Hopital La Peronye (Montpellier, France) and by the Chirurgical PaediatricService of Hospital Marques de Valdecilla (EdD; HMDV, Santander, Spain). MC was arecipient of a predoctoral fellowship from ARC (France). AG belongs to the INSERM(France) and the FMDV-IFIMAV (Spain) and was financially supported by the INSERMand the ISCIII (Spain). AF and LC were employed by FMDV-IFIMAV and supported bythe ISCIII, MR was recipient of a postdoctoral contract by the FMDV-IFIMAV. This workwas funded by ARC (AG; France), La Ligue contre le cancer (JMB; France), and Institutode Salud Carlos III (AG; ISCIII-Programa Regiones Emergentes and Fondo deInvestigaciones Sanitarias PI080890; Spain).

REFERENCES1 Watt FM, Lo Celso C, Silva-Vargas V. Epidermal stem cells: an update. Curr Opin

Genet Dev 2006; 16: 518 -- 524.2 Fuchs E, Horsley V. More than one way to skin. Genes Dev 2008; 22: 976 -- 985.3 Banks-Schlegel S, Green H. Involucrin synthesis and tissue assembly by

keratinocytes in natural and cultured human epithelia. J Cell Biol 1981; 90:732 -- 737.

4 Watt FM, Green H. Involucrin synthesis is correlated with cell size in humanepidermal cultures. J Cell Biol 1981; 90: 738 -- 742.

5 Adams JC, Watt FM. Fibronectin inhibits the terminal differentiation of humankeratinocytes. Nature 1989; 340: 307 -- 309.

6 Eilers M, Eisenman RN. Myc’s broad reach. Genes Dev 2008; 22: 2755 -- 2766.7 Gandarillas A, Watt FM. c-Myc promotes differentiation of human epidermal stem

cells. Genes Dev 1997; 11: 2869 -- 2882.8 Watt FM, Frye M, Benitah SA. MYC in mammalian epidermis: how can an

oncogene stimulate differentiation? Nat Rev Cancer 2008; 8: 234 -- 242.9 Gandarillas A, Davies D, Blanchard JM. Normal and c-Myc-promoted human

keratinocyte differentiation both occur via a novel cell cycle involving cellulargrowth and endoreplication. Oncogene 2000; 19: 3278 -- 3289.

10 Zanet J, Pibre S, Jacquet C, Ramirez A, de Alboran IM, Gandarillas A. EndogenousMyc controls mammalian epidermal cell size, hyperproliferation, endoreplicationand stem cell amplification. J Cell Sci 2005; 118(Part 8): 1693 -- 1704.

11 Zanet J, Freije A, Ruiz M, Coulon V, Sanz JR, Chiesa J et al. A mitosis block linksactive cell cycle with human epidermal differentiation and results in endoreplica-tion. PLoS One 2010; 5: e15701.

12 Murray AW. Recycling the cell cycle: cyclins revisited. Cell 2004; 116: 221 -- 234.13 Watt FM. Role of integrins in regulating epidermal adhesion, growth and

differentiation. EMBO J 2002; 21: 3919 -- 3926.14 Jones PH, Watt FM. Separation of human epidermal stem cells from transit

amplifying cells on the basis of differences in integrin function and expression.Cell 1993; 73: 713 -- 724.

15 Guadagno TM, Ohtsubo M, Roberts JM, Assoian RK. A link between cyclin Aexpression and adhesion-dependent cell cycle progression. Science 1993; 262:1572 -- 1575.

16 Harvat BL, Wang A, Seth P, Jetten AM. Up-regulation of p27Kip1, p21WAF1/Cip1and p16Ink4a is associated with, but not sufficient for, induction of squamousdifferentiation. J Cell Sci 1998; 111 (Part 9): 1185 -- 1196.

17 Hauser P, Ma L, Agrawal D, Haura E, Cress WD, Pledger WJ. Efficientdown-regulation of cyclin A-associated activity and expression insuspended primary keratinocytes requires p21(Cip1). Mol Cancer Res 2004; 2:96 -- 104.

18 Paramio JM, Lain S, Segrelles C, Lane EB, Jorcano JL. Differential expression andfunctionally co-operative roles for the retinoblastoma family of proteins inepidermal differentiation. Oncogene 1998; 17: 949 -- 957.

19 Gandarillas A, Watt FM. Changes in expression of members of the fos and junfamilies and myc network during terminal differentiation of human keratinocytes.Oncogene 1995; 11: 1403 -- 1407.

20 Hurlin PJ, Foley KP, Ayer DE, Eisenman RN, Hanahan D, Arbeit JM. Regulation ofMyc and Mad during epidermal differentiation and HPV-associated tumorigen-esis. Oncogene 1995; 11: 2487 -- 2501.


12


21 Jansen-Durr P, Meichle A, Steiner P, Pagano M, Finke K, Botz J et al. Differentialmodulation of cyclin gene expression by MYC. Proc Natl Acad Sci USA 1993; 90:3685 -- 3689.

22 Hanson KD, Shichiri M, Follansbee MR, Sedivy JM. Effects of c-myc expression oncell cycle progression. Mol Cell Biol 1994; 14: 5748 -- 5755.

23 Dotto GP. p21(WAF1/Cip1): more than a break to the cell cycle? Biochim BiophysActa 2000; 1471: M43 -- M56.

24 Knockaert M, Greengard P, Meijer L. Pharmacological inhibitors of cyclin-dependent kinases. Trends Pharmacol Sci 2002; 23: 417 -- 425.

25 Kleinberger-Doron N, Shelah N, Capone R, Gazit A, Levitzki A. Inhibition of Cdk2activation by selected tyrphostins causes cell cycle arrest at late G1 and S phase.Exp Cell Res 1998; 241: 340 -- 351.

26 Ben-Bassat H, Rosenbaum-Mitrani S, Hartzstark Z, Levitzki R, Chaouat M, Shlomai Zet al. Tyrphostins that suppress the growth of human papilloma virus 16-immortalized human keratinocytes. J Pharmacol Exp Ther 1999; 290: 1442 -- 1457.

27 Planchais S, Glab N, Trehin C, Perennes C, Bureau JM, Meijer L et al. Roscovitine, anovel cyclin-dependent kinase inhibitor, characterizes restriction point and G2/Mtransition in tobacco BY-2 cell suspension. Plant J 1997; 12: 191 -- 202.

28 Rosania GR, Merlie Jr J, Gray N, Chang YT, Schultz PG, Heald R. A cyclin-dependentkinase inhibitor inducing cancer cell differentiation: biochemical identificationusing Xenopus egg extracts. Proc Natl Acad Sci USA 1999; 96: 4797 -- 4802.

29 Whittaker SR, Te Poele RH, Chan F, Linardopoulos S, Walton MI, Garrett MD et al.The cyclin-dependent kinase inhibitor seliciclib (R-roscovitine; CYC202) decreasesthe expression of mitotic control genes and prevents entry into mitosis. Cell Cycle2007; 6: 3114 -- 3131.

30 Meijer L. Chemical inhibitors of cyclin-dependent kinases. Trends Cell Biol 1996; 6:393 -- 397.

31 Planchais S, Glab N, Inze D, Bergounioux C. Chemical inhibitors: a tool for plantcell cycle studies. FEBS Lett 2000; 476: 78 -- 83.

32 Barr FA, Sillje HH, Nigg EA. Polo-like kinases and the orchestration of cell division.Nat Rev Mol Cell Biol 2004; 5: 429 -- 440.

33 Taylor S, Peters JM. Polo and Aurora kinases: lessons derived from chemicalbiology. Curr Opin Cell Biol 2008; 20: 77 -- 84.

34 Perez de Castro I, de Carcer G, Montoya G, Malumbres M. Emerging cancertherapeutic opportunities by inhibiting mitotic kinases. Curr Opin Pharmacol 2008;8: 375 -- 383.

35 Gadea BB, Ruderman JV. Aurora kinase inhibitor ZM447439 blocks chromosome-induced spindle assembly, the completion of chromosome condensation, and theestablishment of the spindle integrity checkpoint in Xenopus egg extracts. MolBiol Cell 2005; 16: 1305 -- 1318.

36 Lenart P, Petronczki M, Steegmaier M, Di Fiore B, Lipp JJ, Hoffmann M et al. Thesmall-molecule inhibitor BI 2536 reveals novel insights into mitotic roles of polo-like kinase 1. Curr Biol 2007; 17: 304 -- 315.

37 Sun TT, Green H. Differentiation of the epidermal keratinocyte in cell culture:formation of the cornified envelope. Cell 1976; 9 (4 Part 1): 511 -- 521.

38 Fernandez-Capetillo O, Lee A, Nussenzweig M, Nussenzweig A. H2AX: the histoneguardian of the genome. DNA Repair (Amst) 2004; 3: 959 -- 967.

39 Taylor WR, Stark GR. Regulation of the G2/M transition by p53. Oncogene 2001; 20:1803 -- 1815.

40 Swift LP, Rephaeli A, Nudelman A, Phillips DR, Cutts SM. Doxorubicin-DNA adductsinduce a non-topoisomerase II-mediated form of cell death. Cancer Res 2006; 66:4863 -- 4871.

41 Dazard JE, Piette J, Basset-Seguin N, Blanchard JM, Gandarillas A. Switch from p53to MDM2 as differentiating human keratinocytes lose their proliferative potentialand increase in cellular size. Oncogene 2000; 19: 3693 -- 3705.

42 Medema RH, Klompmaker R, Smits VA, Rijksen G. p21waf1 can block cells at twopoints in the cell cycle, but does not interfere with processive DNA-replication orstress-activated kinases. Oncogene 1998; 16: 431 -- 441.

43 Ullah Z, Lee CY, Depamphilis ML. Cip/Kip cyclin-dependent protein kinaseinhibitors and the road to polyploidy. Cell Division 2009; 4: 10.

44 Krasinska L, Cot E, Fisher D. Selective chemical inhibition as a tool to study Cdk1and Cdk2 functions in the cell cycle. Cell Cycle 2008; 7: 1702 -- 1708.

45 Coudreuse D, Nurse P. Driving the cell cycle with a minimal CDK control network.Nature 2010; 468: 1074 -- 1079.

46 Yin XY, Grove L, Datta NS, Katula K, Long MW, Prochownik EV. Inverse regulationof cyclin B1 by c-Myc and p53 and induction of tetraploidy by cyclin B1overexpression. Cancer Res 2001; 61: 6487 -- 6493.

47 Herold S, Wanzel M, Beuger V, Frohme C, Beul D, Hillukkala T et al. Negativeregulation of the mammalian UV response by Myc through association withMiz-1. Mol Cell 2002; 10: 509 -- 521.

48 Seoane J, Le HV, Massague J. Myc suppression of the p21(Cip1) Cdk inhibitorinfluences the outcome of the p53 response to DNA damage. Nature 2002; 419:729 -- 734.

49 Beer S, Zetterberg A, Ihrie RA, McTaggart RA, Yang Q, Bradon N et al.Developmental context determines latency of MYC-induced tumorigenesis. PLoSBiol 2004; 2: e332.

50 Munoz-Alonso MJ, Ceballos L, Bretones G, Frade P, Leon J, Gandarillas A.MYC accelerates p21(CIP) -induced megakaryocytic differentiation involvingearly mitosis arrest in leukemia cells. J Cell Physiol 2012; 227: 2069 -- 2078.

51 Campaner S, Doni M, Hydbring P, Verrecchia A, Bianchi L, Sardella D et al. Cdk2suppresses cellular senescence induced by the c-myc oncogene. Nat Cell Biol2010; 12: 54 -- 59; sup pp 1 -- 14.

52 Edgar BA, Orr-Weaver TL. Endoreplication cell cycles: more for less. Cell 2001; 105:297 -- 306.

53 Lilly MA, Duronio RJ. New insights into cell cycle control from the Drosophilaendocycle. Oncogene 2005; 24: 2765 -- 2775.

54 Schellmann S, Hulskamp M. Epidermal differentiation: trichomes in Arabidopsis asa model system. Int J Dev Biol 2005; 49: 579 -- 584.

55 Noya F, Chien WM, Broker TR, Chow LT. p21cip1 Degradation in differentiatedkeratinocytes is abrogated by costabilization with cyclin E induced by humanpapillomavirus E7. J Virol 2001; 75: 6121 -- 6134.

56 Evan G, Littlewood T. A matter of life and cell death. Science 1998; 281:1317 -- 1322.

57 Halazonetis TD, Gorgoulis VG, Bartek J. An oncogene-induced DNA damagemodel for cancer development. Science 2008; 319: 1352 -- 1355.

58 Gandarillas A. Epidermal differentiation, apoptosis, and senescence: commonpathways. Exp Gerontol 2000; 35: 53 -- 62.

59 Iglesias-Ara A, Zenarruzabeitia O, Fernandez-Rueda J, Sanchez-Tillo E, Field SJ,Celada A et al. Accelerated DNA replication in E2F1- and E2F2-deficientmacrophages leads to induction of the DNA damage response and p21(CIP1)-dependent senescence. Oncogene 2010; 29: 5579 -- 5590.

60 Bester AC, Roniger M, Oren YS, Im MM, Sarni D, Chaoat M et al. Nucleotidedeficiency promotes genomic instability in early stages of cancer development.Cell 2011; 145: 435 -- 446.

61 Garcia P, Frampton J, Ballester A, Cales C. Ectopic expression of cyclin E allowsnon-endomitotic megakaryoblastic K562 cells to establish re-replication cycles.Oncogene 2000; 19: 1820 -- 1833.

62 Rheinwald JG. Methods for clonal growth and serial cultivation of normal humanepidermal keratinocytes and mesothelial cells. In: Baserga R (ed.). Cell Growth andDivision. IRL Press: Oxford, 1989, pp 81 -- 94.

63 Littlewood TD, Hancock DC, Danielian PS, Parker MG, Evan GI. A modifiedoestrogen receptor ligand-binding domain as an improved switch forthe regulation of heterologous proteins. Nucleic Acids Res 1995; 23:1686 -- 1690.

64 Simpson CL, Kojima S, Getsios S. RNA interference in keratinocytesand an organotypic model of human epidermis. Methods Mol Biol 2010; 585:127 -- 146.

65 Itahana K, Campisi J, Dimri GP. Methods to detect biomarkers of cellularsenescence: the senescence-associated beta-galactosidase assay. Methods MolBiol 2007; 371: 21 -- 31.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)


13


http://www.nature.com/onc

Documents

Cyclin E drives human keratinocyte growth into differentiation