Normal and c-Myc-promoted human keratinocyte differentiation both occur via a novel cell cycle involving cellular growth and endoreplication

Normal and c-Myc-promoted human keratinocyte di�erentiation bothoccur via a novel cell cycle involving cellular growth and endoreplication

Alberto Gandarillas*,1, Derek Davies2 and Jean-Marie Blanchard1

1Institut de GeÂneÂtique MoleÂculaire, Centre National de la Recherche Scienti®que (CNRS), F34293 Montpellier Cedex 5, France;2FACS Laboratory, Imperial Cancer Research Fund, London, WC2A 3PX, UK

The relationship between cell cycle and di�erentiation inhuman keratinocytes is poorly understood. It is believedthat keratinocytes suppress DNA replication and cellcycle arrest in G0 before they initiate terminaldi�erentiation. However, a temporal separation betweenboth events has not been established. Moreover, c-Mycpromotes keratinocyte di�erentiation without causingcell cycle arrest. To address these paradoxes we haveanalysed cell cycle control during normal and c-Myc-promoted di�erentiation. Continuous activation of c-Mycor initiation of terminal di�erentiation results in a blockof G2/M, cellular growth, endoreplication and poly-ploidy. Keratinocytes abandon G1, continue replicatingDNA as they di�erentiate terminally and becomepolyploid. In fact, simply blocking mitosis with nocoda-zole resulted in increased cell size, terminal di�erentia-tion and endoreplication. This indicates that terminaldi�erentiation associates with defective cell cycle pro-gression and provides a novel insight into c-Myc biology.Oncogene (2000) 19, 3278 ± 3289.

Keywords: epidermis; mitosis; cell size; ¯ow cytometry;cell fate; endoreduplication

Introduction

Human epidermis is a strati®ed epithelium in whichproliferation and terminal di�erentiation are tightlycompartmentalised. Within the basal layer continuousproliferation throughout adult life accounts for the lossof cells that shed as squames from the surface of skin(Watt, 1989; Fuchs and Byrne, 1994). A constantbalance between both processes is crucial to maintainthe normal structure of epidermis and its alterationresults in skin diseases. Proliferation is sustained byinteraction with the basement membrane via integrins(Watt and Jones, 1993) and keratinocytes undergoterminal di�erentiation when this interaction is inter-rupted, unlike endothelial or simple epithelial cells thatundergo apoptosis in suspension (Adams and Watt,1989; Frisch and Francis, 1994; Gandarillas et al.,1999; reviewed in Gandarillas, 2000). As keratinocytescease proliferation and migrate into the suprabasallayers they switch from a cytoskeleton formed bykeratins K5 and K14 to another one formed bykeratins K1 and K10. Concomitantly, they enlargeand initiate the expression of precursors and othercomponents of the corni®ed envelope, such as

involucrin, loricrin and ®laggrin (Watt and Green,1981; Fuchs and Byrne, 1994).

Within the basal layer, epidermal stem cells aremainly quiescent but have an unlimited potential toself-renewal (Lavker and Sun, 1983; Hall and Watt,1989). After division their progeny stays as stem cellsor enter an intermediate state, the transit amplifyingcell, which proliferates continuously but undergoesterminal di�erentiation after a small number of celldivisions. Although little is known about the control ofthe keratinocyte cell cycle, basal keratinocytes arebelieved to withdraw from the cell cycle, switch o�DNA synthesis and arrest in G0 before they initiateterminal di�erentiation (Fuchs, 1993; Gandarillas andWatt, 1995; Hurlin et al., 1995; Hauser et al., 1997;Harvat et al., 1998). However, some ®ndings do notsatisfy this model. Dover and Watt (1987) found notemporal separation between proliferation and terminaldi�erentiation in culture. Similarly, a subpopulation ofepidermal keratinocytes has been found to expressterminal di�erentiation markers and yet undergoS phase (Regnier et al., 1986; Bata-Csorgo et al.,1993; Dover and Watt, 1987). Suprabasal mitotic®gures or thymidine incorporation have also beenreported (Pinkus and Hunter, 1966; Penneys et al.,1970).

Analyses of cell cycle regulators in keratinocyteshave also provided confusing results. Cell cycleinhibitors P16, P21, P27 have been proposed to havea role in keratinocyte di�erentiation (Missero et al.,1996; Hauser et al., 1997). However, althoughconstitutive expression of these molecules in keratino-cytes causes cell cycle arrest, terminal di�erentiation isnot stimulated (Harvat et al., 1998; Di Cunto et al.,1998). Similarly, inhibition of cell cycle by c-mycantisense oligonucleotides or TGFb does not have apositive e�ect on keratinocyte terminal di�erentiation(Reiss and Sartorelli, 1987; Pietenpol et al., 1990;Hashiro et al., 1991).

A further paradoxical line of evidence has arisenfrom our recent study of the role of c-Myc inepidermal di�erentiation (Gandarillas and Watt,1997). c-Myc promotes proliferation and apoptosis ina variety of cell types and is frequently ampli®ed inhuman malignancies (Henriksson and LuÈ scher, 1996).However, the precise mechanisms by which c-Mycperforms its functions are unclear. c-Myc promotesentry of cells in S phase, possibly through inducingpositive regulators of cell cycle (Amati et al., 1998).Recently, it has been suggested that c-Myc may inducethe expression of molecules involved in proteinsynthesis and cellular growth (see Grandori andEisenman, 1997; Dang, 1999). Endogenous c-Myc ishardly detectable in epidermis, and it is down-regulated

Oncogene (2000) 19, 3278 ± 3289ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00

www.nature.com/onc

*Correspondence: A GandarillasReceived 10 January 2000; revised 4 April 2000; accepted 12 April2000

at late stages of in vitro-induced keratinocyte di�er-entiation (see Gandarillas and Watt, 1997). Weconstitutively expressed wild type or conditional c-Myc in all layers of primary keratinocytes by targetingstem cells by retroviral infection (Gandarillas andWatt, 1997). In both cases, and unexpectedly, activa-tion of c-Myc for more than 5 days did not stimulateproliferation or apoptosis but di�erentiation. c-Myc®rst drove keratinocytes to leave the stem cellcompartment and become transit amplifying cells,which subsequently initiated terminal di�erentiationafter four to ®ve rounds of cell division. This increaseof terminal di�erentiation, however, did not associatewith a DNA replication arrest.

Aiming to solve these paradoxes and to understandthis novel function of c-Myc, we have investigated therelationship between cell cycle and di�erentiation inhuman primary, normal or c-Myc expressing keratino-cytes. The application of ¯ow-cytometry techniques tokeratinocytes (Nakatani et al., 1992; Bata-Csorgo etal., 1993; Jones and Watt, 1993; Gandarillas and Watt,1997; Harvat et al., 1998) has provided a powerful toolto study these processes in detail. The resultsconsistently show that the initiation of keratinocyteterminal di�erentiation precedes cell cycle arrest,resulting in cycles of DNA replication withoutcompletion of mitosis. This phenomenon has beenreferred to as endoreplication or endoreduplication inother biological systems (Gra®, 1998; Traas et al.,1998; Hawkins et al., 1996) including some humantissues (Kirk and Clingan, 1980; Ho�man, 1989; Jacket al., 1990; Solari et al., 1996). Continuous activationof c-Myc stimulated keratinocyte endoreplication andproduced an increase of cell size, as a result of ablockade in mitosis. The results reveal a novelrelationship between keratinocyte cell cycle anddi�erentiation and provide new insights into thebiological consequences of c-Myc activation.

Results

Differentiating normal and Myc keratinocytes are foundin any phase of the cell cycle

Both the ratio of DNA synthesis and the proportion ofcells in S phase were only slightly decreased during c-Myc-promoted terminal di�erentiation (Gandarillasand Watt, 1997). We have explored whether constitu-tive activity of c-Myc stimulated the cell cycle ofterminally di�erentiating cells. For this purpose wehave made use of two conditional mycER fusionproteins and constitutively expressed them in keratino-cytes by retroviral infection: c-mycER (containing wild-type c-Myc) or 106ER (containing deletion mutant

Dmyc106-143 that lacks the transactivation domain ofc-Myc; Littlewood et al., 1995; Gandarillas and Watt,1997). Primary keratinocytes were cultured in thepresence of a feeder layer of mouse ®broblasts and1.5 mM calcium. In such conditions keratinocytesproliferate and stratify, mimicking epidermis in vivo(Rheinwald, 1989; Watt, 1989). c-mycER was activatedin primary keratinocytes upon 4-OH-tamoxifen (OHT)for 9 days and cells double stained for the terminaldi�erentiation marker involucrin and DNA content.The cell cycle distribution of involucrin positive ornegative cells was then quantitated by ¯ow-cytometry(Table 1). Terminally di�erentiating keratinocytesexpressing activated c-mycER were found in everyphase of the cell cycle, and a signi®cant proportion ofinvolucrin positive cells were undergoing S phase. Thenumbers however, were similar in di�erentiating andnon di�erentiating cells expressing the empty vectorpuro-resistance (KpBabe) and in cells expressing thenegative myc mutant (K106; Table 1). Therefore, thepresence of di�erentiating keratinocytes in S phase wasnot speci®c of c-Myc activation. There was, however, aslight accumulation of c-Myc di�erentiating cells inG2/M.

Keratinocytes can be induced to terminally di�er-entiate when they are placed in suspension in theabsence of cell adhesion. When normal, controlkeratinocytes expressing the empty vector were inducedto terminally di�erentiate in suspension for 24 h, theywere found to contain a proportion of cells in S phase(9%, Table 1) and to accumulate in G1 (62.8%) or G2(26.0%). Interestingly, during 24 h in suspension aproportion of pBabe keratinocytes abandoned G1 andcompleted S phase to accumulate in G2 (total cells inG1 decreased from 75.1% to 62.8%; total cells inS phase from 11.3% to 9%; total cells in G2+Mincreased from 12.2% to 26.0%).

Normal and Myc differentiating keratinocytes continueDNA synthesis and become polyploid

Continuous activation of c-mycER in keratinocytesdrives stem cells to become transit amplifying cells andleave the proliferative compartment after 5 days(Gandarillas and Watt, 1997), when they initiateterminal di�erentiation. At this stage bi- or tri-nucleatecells were frequently observed (Figure 1a). Confocalanalyses of c-Myc staining showed stratifying, di�er-entiating keratinocytes to often contain more than onenucleus or a large single nucleus (Figure 1b). Multi-nucleate di�erentiating keratinocytes were also foundin normal cultures (Figure 1a) but not as frequently.

Multinucleate cells must be a consequence of adefective mitosis that results in polyploidy. Toinvestigate this possibility, we analysed normal,

Table 1 Cell cycle distribution of terminally di�erentiating or non-di�erentiating keratinocytes as quantitated by ¯ow-cytometry

KpBabe K106ER KmycER KsuspInvol7 Invol+19.3 (0.4) Invol7 Invol+17.1 (0.6) Invol7 Invol+50.5 (1.5) Total+55.4 (1.2)

G1 75.4 (0.1) 74.4 (1.3) 72.7 (0.4) 70.8 (1.3) 77.6 (1.8) 71.2 (1.8) 62.8 (1.3)S 11.8 (0.1) 10.8 (0.5) 16.1 (0.5) 16.0 (1.0) 12.4 (0.8) 11.6 (0.4) 9.0 (0.4)G2+M 11.9 (0.3) 13.7 (0.6) 10.1 (0.0) 10.8 (0.3) 9.2 (1.1) 16.3 (1.4) 26.0 (1.2)

Numbers next to Invol+ are the percentage of involucrin positive cells within the whole population. Numbers in table are percentages of cells inthe di�erent phases of the cell cycle with respect to the involucrin positive or negative subpopulations. Data are means of triplicate samples;s.e.m. are shown in parentheses. Ksusp: Cells in suspension for 24 h. All other cells were OHT-treated for 9 days

Oncogene

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3279

stratifying cultures of primary keratinocytes by ¯ow-cytometry, after staining DNA with propidium iodideand without gating out events beyond G2/M (Figure2). Cell aggregates were excluded on basis of theWidth/Area (Figure 2a; see e.g., Ormerod, 1990). Wehave found that normal, stratifying cultures of primaryhuman keratinocytes contain around 8% of polyploidcells, with a slight variation depending on thekeratinocyte strain (Figure 2a). A peak of cells with8 N DNA content and a small proportion of cells with12 N DNA content were also observed. Events beyond4 N were sorted and visualized with a conventional¯uorescence microscope to con®rm that they weresingle polyploid cells (see below). Moreover, cellsuspensions that had been stained for DNA wereplaced on slides and analysed with a Laser ScanningCytometer, which detected a signi®cant proportion ofsingle, polyploid nuclei (see below).

As keratinocytes terminally di�erentiate and stratifythey progressively increase their cell size (Banks-Schlegel and Green, 1981) and it is possible to sortthem on basis of their Side Scatter pro®le (Jones andWatt, 1993). Polyploid keratinocytes were restricted tothe suprabasal population (Figure 2b). Parallel sampleswere double stained for DNA and the terminaldi�erentiation marker involucrin. The DNA contentof involucrin negative or positive cells was determined

(Figure 2b, third panel). Polyploid cells were restrictedto the involucrin positive population. At a given time,they constituted more than a fourth of terminallydi�erentiating cells (26.3% on basis of light scatter;29.3(+1.6%) on basis of involucrin staining; Figure2a,b). Therefore polyploidy was not a randomconsequence of the conditions in culture, but it wastightly associated with terminal di�erentiation. Thissuggests that di�erentiating keratinocytes undergoendoreplication (cycles of DNA replication in theabsence of mitosis; see Introduction). Interestingly,keratinocyte size correlated with DNA content, asobserved in endoreplicating tissues (Kirk and Clingan,1980; Traas et al., 1998). Keratinocytes in the di�erentphases of the cell cycle could be identi®ed on basis oftheir light scatter properties (three colours in Figure2a). Finally, the majority of very large cells, that are ata late stage of terminal di�erentiation (Banks-Schlegeland Green, 1981; Watt and Green, 1981) werepolyploid (red dots in Figure 2a,b), and only a smallproportion were in G1 (green dots in Figure 2a,b).

To con®rm that di�erentiating keratinocytes are ableto synthesise DNA, stratifying cultures were incubatedwith BrdU, for 2.5 h, and double stained for DNAcontent and BrdU (Figure 2c). BrdU incorporation inbasal or terminally di�erentiating cells was thendetermined by means of physical parameters as inFigure 2a,b. Although slightly more DNA-synthesisingcells were found within the basal, non-di�erentiatingcompartment (23.5%; Figure 2c) a similar proportionof diploid and polyploid terminally di�erentiating cellswere positive for DNA synthesis (21.3%; Figure 2c).23.1% of DNA synthesising cells were suprabasal (onbasis of light scatter; Figure 2c), compared to 22.2%suprabasal cells in the whole population (Figure 2a).These observations suggest that the capacity ofdi�erentiating cells to replicate DNA is not signi®-cantly lower than that of basal cells. It is also worthnoting that BrdU or thymidine incorporation techni-ques are misleading to estimate keratinocyte prolifera-tion, unless only the basal population is considered.

Keratinocytes thus become polyploid and occasion-ally multinucleate as they undergo terminal di�erentia-tion. Confocal analyses of double staining for DNAand a terminal di�erentiation marker that is expressedat the plasma membrane (CD44-V3; Hudson et al.,1996), showed multiple simultaneous anaphases insingle, strati®ed keratinocytes (Figure 3a). Simulta-neous multiple anaphases occur in some multinucleatecells undergoing endomitosis (see e.g. Kirk and Cling-an, 1980). To monitor this phenomenon in live cells,activated-c-mycER or normal keratinocytes wererecorded for 4 days using time-lapse video microscopy.These ®lms showed stratifying, di�erentiating kerati-nocytes undergoing nuclear divisions in the absence ofcell division (Figure 3b). This phenomenon was alsoobserved in normal cultures (not shown) but was morefrequent in c-mycER cells after activating c-Myc for 5days (Figure 3b).

Induction of terminal differentiation and c-Myc actionboth trigger keratinocyte endoreplication

To explore whether endoreplication is a direct conse-quence of terminal di�erentiation, we quantitatedpolyploid cells after stimulating terminal di�erentiation

Figure 1 Multinucleation in c-Myc expressing or normalprimary human keratinocytes. (a) Top and middle panels: culturesof mycER-expressing keratinocytes in the absence of OHT (toppanel) or 5 days after activation of c-mycER with OHT (middlepanel; some of the frequent binucleate cells are highlighted witharrows). Bottom panel: normal culture of primary keratinocytescontaining some large, terminally di�erentiating, multinucleatecells (arrows). Bar: 200 mm. (b) Con¯uent keratinocytes expressingmycER that had been activated with OHT for 5 days were stainedfor c-Myc and analysed by confocal microscopy. Bottom panelfocuses on the top of the basal layer (B) and ®rst suprabasal layer(S1) upper panels show two suprabasal layers (S2, S3) containingtri-nucleate cells (arrows) and a gigantic nucleus (arrowhead).Sections: 0.3 mm. Bar: 40 mm

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3280

Oncogene

in normal keratinocytes in suspension or upon con-stitutive activation of c-Myc. As shown in Figure 4a ± c,the proportion of polyploid cells increased signi®cantlyin keratinocytes that were suspended for 14 h (8.2 ±16.4%), concomitantly with the increase of involucrinexpression (insets in Figure 4a ± c). While G2 andpolyploidy increased, G1 and S phases decreased, asobserved previously (Table 1), despite the fact that after14 h cells had no longer the capacity to proliferate whenreplated (not shown). After 24 h polyploidy wasreduced (13.4%), possibly as a consequence of DNAdegradation that takes place at late stages of terminaldi�erentiation. BrdU incorporation analyses in sus-pended cells indicated that DNA synthesis continuedafter 6 h, when cells are irreversibly committed toterminal di�erentiation (Adams and Watt, 1989) and up

to 24 h (Figure 4e). After 24 h DNA synthesis is greatlyreduced (not shown) possibly due to the lack ofanchorage. Promotion of terminal di�erentiation after7 days of activating c-Myc did not suppress DNAsynthesis either (Figure 4f), and provoked an increase inpolyploidy that tightly correlated with the degree ofterminal di�erentiation (Figure 4g). It must be notedthat after activation of c-Myc for 7 days a higherproportion of DNA synthesizing cells were in G1 andS phases, as compared to the suspension situation. Thisis due to the fact that suspension-induced di�erentiationis a terminal process, whereas a proportion of c-Myccells in stratifying cultures continue to proliferate andshed into the culture medium as they undergo latestages of terminal di�erentiation (see Gandarillas andWatt, 1997).

Figure 2 Flow-cytometry analyses of cell cycle and di�erentiation in primary human keratinocytes. (a,b) Single cell suspensionswere ®xed, stained for DNA with propidium iodide (PI; FL2) or double stained for DNA and involucrin. Aggregates were excludedon basis of PI Area/Width (a). Light scatter (SSC-H and FSC-H) re¯ects cell size and complexity. Non-di�erentiating ordi�erentiating keratinocytes were gated on basis of the light scatter dot plot and their DNA content determined (R2 or R3 in a,b).Third panel in b plots Side Scatter (SSC) versus involucrin expression (FITC). Numbers show the quantitation of every region withrespect to the number of cells within each plot. Three colours in third panels of a,b: green dots are total cells in G1 (M1 in a), bluedots total cells in S+G2/M (M4 in a) and red dots polyploid cells (M3 in a). M2: cells in G2/M. (c) Brdu incorporation of primarykeratinocytes over 2.5 h to analyse DNA synthesis. Cells in R3 are positive for BrdU as determined with non-BrdU control samples.BrdU incorporation of basal or suprabasal cells was determined on basis of light scatter as in a,b. The far-right panel shows thelight scatter properties of cells that incorporated BrdU. Data are from multiple samples of three di�erent strains of foreskinkeratinocytes. Numbers in brackets are the s.e.m

Oncogene

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3281

Events with more than 4 N DNA content from theDNA stainings in Figure 2a,b were sorted on slidesand visualized to con®rm that they were single,polyploid cells (Figure 5a ± c). Note the presence ofthree nuclei in a cell that begun to lose the involucrinepitope (Figure 5b; involucrin is undetectable when it isincorporated into the corni®ed envelope). Polyploidcells often had a single, very bright gigantic polyploidnucleus (Figure 5b,c). Very large cells and nuclei werefrequently observed when keratinocytes were placed insuspension to induce terminal di�erentiation (Figure5d,e). Figure 5f shows individual polyploid cells asidenti®ed using a Laser Scanning Cytometer to analysenormal keratinocytes stained for DNA and placed andmounted on slides. Note that either big individualnuclei, binucleate cells (arrow-heads) or multinucleatecells (arrows) were found within the polyploid fraction.

Blocking keratinocyte mitosis triggers terminaldifferentiation and endoreplication

The accumulation of di�erentiating keratinocytes in G2/M and the presence of endoreplication in suprabasal cellsindicate that a mitosis blockade associates with terminaldi�erentiation. To further test this association, westudied it in an inverse fashion: we blocked keratinocytemitosis continuously with nocodazole and determinedthe e�ect on di�erentiation and endoreplication. Primary

keratinocytes were hard to synchronise completely, asthey are di�cult to growth arrest in the absence of serumand growth factors (Watt et al., 1991; unpublishedobservations). DNA synthesis continued in the presenceof nocodazole (compare BrdU incorporation in Figure6e with untreated control cells in Figure 6a), consistentwith observations that a nocodazole block still permitsDNA synthesis in some cells (Kung et al., 1990).Nevertheless, a 48 h treatment with nocodazole pro-voked an accumulation in G2/M, an increase inpolyploidy (compare M2 and M1, respectively, in`DNA' Figure 6a,b) and interestingly, a dramaticincrease of cell-size (`scatter' in Figure 6a,b) and terminaldi�erentiation (involucrin expression; Figure 6a,b).After nocozadol treatment peaks beyond 4 N were moreevident, likely due to a higher degree of synchronisation.The morphology of cells treated with nocodazole wasalso indicative of an increased strati®cation and shedding(compare Figure 6d with c showing con¯uent areas), agreat increase of cell size and premature terminaldi�erentiation in the basal layer (compare non-con¯uentareas in Figure 6e and f).

Induction of differentiation, activation of c-Myc or amitosis block, all provoke an increase of cellular size

Terminal di�erentiation associates with an increase ofcell size (Banks-Schlegel and Green, 1981; Watt and

Figure 3 Nuclear division (endomitosis) in endoreplicating, di�erentiating primary keratinocytes. (a) multiple simultaneousanaphases in a di�erentiating keratinocyte as revealed by confocal analyses of double staining for DNA (propidium iodide; leftpanels) and a variant of CD44 that is speci®cally expressed during terminal di�erentiation (CD44V3; right panels). Lower panelsfocus on a suprabasal layer (S); upper panels focus on a rounded-up cell within the same layer; broken lines highlight the edge ofthe cell and straight lines the position of the three endomitotic spindles. (B) fainter, out of focus basal nuclei are visible; (S)suprabasal nuclei; sections of 0.2 mm; bar: 20 mm. (b) six consecutive sequences of time-lapse ®lming of primary culturedkeratinocytes. The di�erentiating cell pointed-out with arrows goes from two to four nuclei during the 26 min of the sequence andits edge has been stressed with black broken lines in sequence 6. Bar: 160 mm

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3282

Oncogene

Green, 1981; see also Figure 2a,b). In addition,polyploidy has been found associated with larger nucleiand cells in endoreplicating systems (Traas et al., 1998;Edgar, 1995; Ho�man, 1989; Jack et al., 1990). Wehave observed that keratinocytes become larger afteractivation of c-Myc in conventional cultures and invitro reconstituted epidermis (Gandarillas and Watt,1997). We determined the relationship between DNAcontent, terminal di�erentiation and cell size inkeratinocytes (Figure 7). Suspension-induced di�eren-tiation, nocodazole treatment or activation of c-Myc,all provoked an accumulation of keratinocytes in G2/M and a signi®cant increase of cell size. This increasealso took place in G1 cells, but was more marked inG2/M and polyploid cells (Figure 7a ± c). Consistently,a 48 h treatment with hydroxy-urea (HU) that blockedkeratinocytes in G1/S and reduced polyploidy, causeda limited increase of cell size compared to thatprovoked by nocodazole. In contrast, a 48 h TGF-btreatment, that inhibits the G1/S transition, prolifera-tion and c-Myc expression in keratinocytes (see e.g.

Pietenpol et al., 1990), did not cause a signi®cantincrease of cell size (not shown). Due to the conditionalER fusion protein we could monitor the kinetics of thee�ect of c-Myc on cell size (Figure 7d,e). This e�ectwas detected even 24 h after activation of c-mycER,when all other di�erentiation markers were stilluna�ected (Figure 7e; Gandarillas and Watt, 1997).

Discussion

Differentiating keratinocytes initiate terminaldifferentiation from any point in the cell cycle andundergo endoreplication

Epidermal keratinocytes have been believed to with-draw from the cell cycle and arrest in G0 beforeinitiating terminal di�erentiation. However, we haveobserved a similar cell cycle distribution in di�erentiat-ing and non-di�erentiated keratinocytes, even shortlyafter inducing terminal di�erentiation in suspension. In

Figure 4 DNA content and DNA synthesis in primary keratinocytes during suspension- or c-Myc-promoted terminaldi�erentiation. (a ± c) involucrin expression and DNA content of primary keratinocytes before (a) or after 14 h (b) or 24 h (c) insuspension was analysed by ¯ow-cytometry. Small insets show involucrin staining and a negative control antibody staining (humanCD8; broken line; a). Percentages show the proportion of cells in G2/M phase (R3) or polyploidy (R4); numbers are means of threeindependent experiments and standard deviations are shown in brackets. (d ± f) keratinocytes placed in suspension for 24 h wereincubated with no BrdU (d) or with BrdU for the last 18 h (e). Note that adherent keratinocytes were incubated with BrdU for 1.5 hafter activation of c-Myc for 7 days (f). (g) show quantitations of involucrin positive cells (left panel) or polyploid cells (right panel)in primary keratinocytes expressing the constructs indicated, after addition of OHT for the times indicated. Data in g arerepresentative of three independent experiments; small bars are the s.e.m

Oncogene

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3283

addition, the proportion of cells in G1 was signi®cantlyreduced and the proportion in G2/M increased, in theonset of di�erentiation. It must thus be concluded thatkeratinocyte terminal di�erentiation initiates from anypoint within the cell cycle, con®rming previoussuggestions in a similar direction (Dover and Watt,1987; Kartasova et al., 1992; Nakatani et al., 1992;Harvat et al., 1998). Moreover, terminally di�erentiat-ing cells continued DNA synthesis and we have founda signi®cant proportion of keratinocytes that arepolyploid and restricted to the suprabasal population.

Keratinocyte cell cycle and terminal di�erentiationare therefore uncoupled and this results in endoreplica-tion: extra rounds of DNA replication in the absenceof cell division. Some polyploid cells undergo nucleardivision (endomitosis), and become multinucleate, asobserved in other endoreplicating systems (e.g. Kirkand Clingan, 1980; Jack et al., 1990). More than afourth of the suprabasal population was polyploid at agiven time and about a third was in G2/M or G1. It istechnically complex to determine whether all di�er-entiating cells eventually abandon G1, but thediminution of the G1 peak as terminal di�erentiationprogresses suggests that they tend to do so. Aproportion of Keratinocytes, however, accumulated inG1 after 24 h in complete absence of cell adhesion, butDNA synthesis was also suppressed in that situation.

Di�erentiating keratinocytes may or may not leave G1depending on time and/or molecular signals, but it isclear that a G1 or G0 arrest is not required forinitiation of terminal di�erentiation. The resultspresented here indicate that terminal di�erentiationtends to associate with mitosis-defective cell cycleprogression, rather than with arrest in G0. That wasthe case during spontaneous or suspension-induceddi�erentiation, upon activation of c-Myc or afterblocking mitosis with nocodazole.

Mechanisms of keratinocyte endoreplication and c-Mycfunction

The molecular mechanisms that suppress cell divisionand trigger endoreplication in the onset of terminaldi�erentiation are to be elucidated. When proliferatingbasal cells detach from the basement membrane theyproduce the di�erentiation cytoskeleton and subse-quently, the corni®ed envelope (Fuchs and Byrne,1994). Our model is that this new cytoskeleton mightbe too rigid and create a physical constraint that wouldprevent the cell from splitting up even when the DNAreplication machinery is still functional. Time-lapsevideo recording showed di�erentiating cells initiatingcell division, but failing to perform cytokinesis. Severallines of evidence support this model: (a) in our

Figure 5 Polyploid cells and nuclei in keratinocyte primary cultures. (a ± c) photographs of polyploid keratinocytes sorted by ¯ow-cytometry from the double staining involucrin (green)/DNA (red) in Figure 2b. Note multinucleate cells in (a arrowheads), a cellwith three nuclei in (b) and a gigantic, very bright single nucleus in (c). (d,e) double staining involucrin (green)/DNA (red) ofprimary keratinocytes that were induced to di�erentiate in suspension for 24 h; note the large size of cell and nucleus and thepresence of two nuclei in some very di�erentiated cells. (f) single polyploid events were identi®ed, after staining DNA as in Figure2a, by a Laser Scanning Cytometer on cell suspensions that were placed on slides. These events were single cells with either a singlebig nucleus, two nuclei (arrowheads) or several nuclei (arrows). Bar: (a) 100 mm; (b ± f) 20 mm

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3284

Oncogene

experiments suppression of mitosis by nocodazolemimicked the cell cycle distribution of spontaneouslydi�erentiating cells, and triggered terminal di�erentia-tion and endoreplication; (b) constitutive expression ofthe di�erentiation cytoskeleton keratins K1 or K10 inkeratinocyte cell lines does not sustain proliferationand yet allows DNA synthesis (Kartasova et al., 1992;Paramio et al., 1999) and (c) interestingly, the skin oftransgenic mice expressing K10 under the K5 promoter(speci®cally expressed in the basal layer) consists of asingle layer of ¯attened keratinocytes (JL Jorcano,personal communication).

Endoreplication thus explains why expression ofdi�erentiation keratins or corni®ed envelope precursorsand factors that promote di�erentiation, either inculture (Kartasova et al., 1992; Gandarillas and Watt,1997) or in vivo (Penneys et al., 1970; Morasso et al.,1996) coexist with DNA synthesis. It may also explainwhy inhibition of the G1/S transition in keratinocytesby anti-c-myc antisense oligonucleotides, TGFb (Reissand Sartorelli, 1987; Pietenpol et al., 1990; Hashiro etal., 1991), overexpression of cell cycle inhibitors P16,P21 or P27 (Harvat et al., 1998; Di Cunto et al., 1998)

or inhibition of S phase kinase cdk2 (Ben-Bassat et al.,1997; Martinez et al., 1999) did not result in anincreased terminal di�erentiation.

Furthermore, in the light of endoreplication, c-Myc-promoted keratinocyte di�erentiation appears compa-tible with its known role in driving cell cycleprogression. Constitutive activation of c-Myc producedan increase of endoreplication that associated withincreased cell size and di�erentiation. In keratinocytes,endoreplication masked the e�ect of c-Myc on cellcycle when the whole population was considered(Gandarillas and Watt, 1997). However, we haveobserved a transient increase of BrdU incorporation24 h after activating c-Myc, when only basal cells wereanalysed (unpublished observations). Continuous sti-mulation of cell cycle may ®rst drive stem cells tobecome transit amplifying cells and eventually, triggerterminal di�erentiation and a G2/M blockade whileDNA replication continues. The correct checkpoints inG2/M would then be lost in the onset of di�erentia-tion. Interestingly, the phenotype caused by a 48 hnocodazole treatment was very similar to that observedafter activation of c-Myc for 5 days (Figure 6d,f;

Figure 6 E�ects of blocking mitosis with nocodazole on keratinocyte cell cycle and di�erentiation. Untreated, controlkeratinocytes (a) or keratinocytes that had been treated with nocodazole for 48 h (b), were analysed for DNA synthesis (BrdU:FL1), DNA content (Propidium Iodide), light Scatter (SSC/FSC) and involucrin expression. Numbers show the proportions ofBrdU positive cells (R2), cells in G2/M (M2) or polyploid (M1) and are means of three independent experiments. S.e.m. are shownin brackets. (c ± f) morphology of untreated cells (c,e) or those treated with nocodazol (d,f). (c,d) show cells in con¯uent areas; (e,f)show individual colonies. Note the massive strati®cation at con¯uence in d and the large size of cells in f compared to controlkeratinocytes. Bar: 200 mm

Oncogene

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3285

Gandarillas and Watt, 1997). Four very recent reports,together with our results, suggest that continuousactivity of c-Myc may uncouple cell growth and celldivision: (a) overexpression of c-Myc promoted en-doreplication in ®broblasts and epithelial cells thatwere arrested in metaphase (Li and Dang, 1999); (b) d-myc, the drosophila homologue of c-Myc, stimulatedcellular growth rather than proliferation, due to mitosisbeing independently controlled (Johnston et al., 1999)and (c) continuous c-Myc activity provoked an increasein protein synthesis and cell size in a B-cell line andduring lymphocyte development of transgenic micewhen stimulating cell growth but not cell division(Iritani and Eisenman, 1999; Schuhmacher et al., 1999).

By driving cellular growth and cell cycle progression,c-Myc may promote proliferation when mitosis is

correctly executed, but di�erentiation and endoreplica-tion when mitosis is impaired (see also Iritani andEissenman, 1999; Schuhmacher et al., 1999). It is worthnoting that c-Myc activity caused both polyploidy andapoptosis in Rat-1 cells in the absence of mitosis (Liand Dang, 1999), and that terminal di�erentiation inkeratinocytes may be the counterpart of apoptosis inother cell types (reviewed in Gandarillas, 2000).Endogenous c-Myc is hardly detectable in epidermisand is down-regulated during in vitro-induced terminaldi�erentiation (see Gandarillas and Watt, 1997). Asustained increase of c-Myc function in individual stemcells may drive them into the actively proliferativecompartment. This may be su�cient to commit thesecells to subsequent di�erentiation, when c-Myc activitywould no longer be required (see Gandarillas and

Figure 7 Analyses of cell size and DNA content in human keratinocytes after suspension-induced di�erentiation, continuousactivation of c-Myc or blocking cell cycle. (a ± c) plot Side Scatter (SSC) versus DNA (PI) of normal primary keratinocytes after24 h in suspension (a), activation of c-mycER upon OHT for 6 days (c-myc+) or a treatment with either nocodazole or hydroxy-urea for 48 h (c). Parallel controls of untreated, adherent, normal primary keratinocytes are also shown. In b the control is c-mycERexpressing cells cultured in the absence of OHT. Numbers are the per cents of cells with a high SSC value (R1); in brackets ares.e.m. of three independent experiments. (d) comparison of Side Scatter levels of primary keratinocytes expressing empty vectorpBabe, inactive c-myc mutant 106ER or c-mycER cultured in the presence of tamoxifen (OHT) for 9 days; (e) similar analyses in c-mycER keratinocytes before or after activating c-Myc with OHT for the number of days indicated

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3286

Oncogene

Watt, 1997). Interestingly, constitutive activity of c-Myc in mouse transgenic epidermis provoked hyper-proliferation and hyperkeratosis (thickening of di�er-entiating layers), occasional nuclear division indi�erentiating cells and angiogenesis, but not apoptosis(Pelengaris et al., 1999; Waikel et al., 1999). Mitogenicsoluble factors produced in vivo during vascularisation,in¯ammation or tumorigenesis may favour myc-induced proliferation by alleviating the mitosis block-ade.

What mechanisms link a G2/M blockade withinitiation of terminal di�erentiation? Successive cyclesof DNA replication in the absence of mitosis are knownto allow cells to become bigger (see e.g. Traas et al., 1998;HuÈ lskamp et al., 1999) and we have found thatkeratinocyte terminal di�erentiation correlates withDNA content. In our experiments, the increase of cellsize caused when blocking G2/M with nocodazole wasgreater than that caused when blocking G1/S withhydroxy-urea. This was also observed during terminaldi�erentiation and after activating c-Myc. Interestingly,keratinocyte size correlates with di�erentiation in cultureand in vivo, so that beyond a certain volume they mayinitiate the expression of di�erentiation markers (Wattand Green, 1981). Moreover, cell shape and the area ofcell adhesion to the substrate tightly control keratinocytedi�erentiation (Watt et al., 1988; Adams and Watt,1989). The e�ect on cell size elicited by activating c-Mycwas evident after 24 h, prior to any stimulation ofdi�erentiation. It is tempting to speculate that anincreased cell volume may ultimately trigger terminaldi�erentiation by reducing positive signalling via celladhesion. Interestingly, we have previously shown that c-Myc activity down-regulates integrin expression (Gan-darillas and Watt, 1997; see also Judware and Culp,1997). A link between cell volumen, integrin signalling,cell cycle and cell fate has been proposed elsewhere (seee.g., Meyerowitz, 1994; Chicurel et al., 1998; Neufeld andEdgar, 1998; Conlon and Ra�, 1999).

Physiological significance

In a stady-state epidermis, uncoupling cell cycle andterminal di�erentiation may be important as a self-defence mechanism against oncogenic mutations and tokeep a tight balance between proliferation and terminaldi�erentiation. In the event of loss of the cell cyclecontrol, the initiation of terminal di�erentiation uponloss of adhesion would block cell division without theneed to suppress DNA replication. Linking di�erentia-tion with cell cycle progression may ensure that wheneverthe latter is stimulated, terminal di�erentiation alsoincreases. Interestingly, this is what occurs in skinhyperproliferative lesions such as psoriasis or wound-healing. Endoreplication may also account for appar-ently paradoxycal reported coexpression of S phase anddi�erentiation markers and suprabasal `mitotic' ®guresin epidermis (see Introduction; Regnier et al., 1986; Bata-Csorgo et al., 1993; Pinkus and Hunter, 1966; Penneys etal., 1970). Consistently, we have found a signi®cantproportion of di�erentiating polyploid keratinocytesisolated from normal human epidermis (Gandarillas etal., unpublished observations).

Endoreplication is known to occur during Drosophiladevelopment (Edgar, 1995) and in human megakario-cytes (Ho�man, 1989), osteoclasts (Solari et al., 1996),

endometrium (Kirk and Clingan, 1980) and liver (Jack etal., 1990). In all these cases it associates with increasedcell size and a requirement for a specialized function.Very interestingly, endoreplication is known to beimportant in plant epidermal di�erentiation to theproduction of large single cell structures (trichomes;Traas et al., 1998; HuÈ lskamp et al., 1999).

We are currently investigating the molecular me-chanisms connecting cell cycle, cell size, endoreplica-tion and terminal di�erentiation. These should be keyissues to better understand how tissue homeostasis isregulated.

Materials and methods

Cell culture

Primary keratinocytes were isolated from neonatal humanforeskin and cultured in the presence of a feeder layer ofmouse J2-3T3 ®broblasts in FAD medium as describedpreviously (Rheinwald, 1989; Gandarillas and Watt, 1997).Early passages (from 1 to 4) of Keratinocytes from fourdi�erent individuals were used (strains ka, kq, kz and kmb).To block keratinocytes at di�erent phases of the cell cycle,

stratifying cultures were treated with 2 mM hydroxy-urea,10 ng/ml TGFb or 20 mM nocodazole for the length of timeindicated.

Expression of myc conditional forms

Primary keratinocytes expressing empty retroviral vectorpBabe, or pBabe containing conditional c-mycER, D106-143ER (Littlewood et al., 1995) were obtained by retroviralinfection as previously described (Gandarillas and Watt,1997) and cultured as normal keratinocytes. Activation of c-Myc conditional forms was achieved by adding 100 nM 4-hydroxytamoxifen (variant Z; Research Biochemicals Inter-national) to the culture medium every 48 h. Primarykeratinocytes from two di�erent individuals (strains ka andkq) expressing conditional myc forms were used in this study.

Confocal analyses of stratifying keratinocytes

Keratinocytes expressing mycER were immunostained withanti-myc mAb 9E10 (Evan et al., 1985) as previouslydescribed (Gandarillas and Watt, 1997). Normal cultureswere ®xed in 3.8% formaldehyde and permeabilised in7208C-cold methanol and double stained for CD44 withmAb CD44V3 (kind gift of F Watt; Hudson et al., 1996) asfor 9E10 and for DNA with 40 mg/ml propidium iodide for5 min at room temperature. Stainings were then analysedwith a Microphot FX microscope (Nikon UK Ltd) equippedwith an MRC-600 laser scanning confocal microscopeattachment (Bio Rad Microscience, Hemel Hempstead,UK). To focus on di�erent layers of keratinocyte culturesoptical sections were 0.2 or 0.3 mm thick.

Quantitation of DNA content, differentiation and DNAsynthesis by flow cytometry

Trypsinised keratinocytes were washed once with PBS and®xed. For involucrin staining, cells were ®xed in 1%paraformaldehyde for 5 min, washed in PBS and stainedfor involucrin as previously described (Gandarillas and Watt,1997). Monoclonal anti-involucrin antibody SY5 was a kindgift of F Watt. For DNA content, un®xed keratinocytes orkeratinocytes that had been stained for involucrin were ®xedand permeabilized, or just permeabilized, respectively, in 70%cold ethanol for at least 30 min at 48C, with vortexing for the®rst minute. Cells were then washed twice with PBS and

Oncogene

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3287

resuspended in PBS containing 20 mg/ml ribonuclease and40 mg/ml propidium iodide. Flow cytometry analysed werethen performed as before.

For DNA synthesis analyses, keratinocytes that had beencultured in the presence of 10 mm BrdU were trypsinized,washed once in PBS and ®xed in ethanol as above. BrdUstaining was then performed as described (Gandarillas andWatt, 1997) followed by RNAse treatment and propidiumiodide for DNA content as above.

Keratinocytes stained for involucrin, or those doublestained for involucrin or BrdU and DNA, were ®rmlyresuspended and ®ltered through a 70 mM mesh to minimizethe presence of aggregates and then analysed by ¯owcytometry on a Becton Dickinson FACScan. 10 000 ± 50 000events were gated on the basis of PI area/width to excludecell agregates as in (Ormerod, 1990), and adquired in listmode for every sample.

Time-lapse videomicroscopy

Normal or c-mycER expressing primary keratinocytes wererecorded for up to 4 days using an inverted microscopecontaining a CO2 chamber (Olympus IMTI or IMT2).Frames were taken every 2 min utilizing video equipment asdescribed previously (Gandarillas and Watt, 1997).

Cell sorting and laser scanning cytometer

Cultured or freshly isolated keratinocytes were ®xed andstained for involucrin and DNA as above and cells having

more than 4 N DNA content were sorted using a Becton-Dickinson FACStar Plus and visualized on a Zeiss Axiophot¯uorescence microscope.To quantitate DNA of single nuclei, two methods were

used: (a) 15 ml of single cell suspensions of cultured or freshlyisolated keratinocytes after trypsin treatment were air-driedon glass slides at 378C, ®xed in 3.8% formaldehyde andwashed in PBS. Cells were then stained for DNA for 5 min asbefore; (b) single cell suspensions were stained for DNA asbefore and then 20 ml were placed on slides. Cells from eithermethod were covered with coverslips, sealed with nail varnishand analysed with a Laser Scanning Cytometer (CompuCyte,Cambridge, MA, USA).

AcknowledgementsThe initial observations of this work were made in thelaboratory of F Watt in ICRF of London, and AGandarillas is especially grateful to her generosity andhelpful suggestions. We thank C Brooks for some technicalassistance, D Fisher for constructive suggestions and MGomez and H Land for helpful comments at an early stageof the work. Thanks to JP Moles the data obtained fromEnglish skin could be displayed on French computers. AGandarillas was funded by Bristol-Myers Squibb, ARCand EMBO. This research received ®nancial support fromCNRS and Ligue Nationale Contre le Cancer.

References

Adams JC and Watt FM. (1989). Nature, 340, 307 ± 309.Amati B, Alevizopoulos K and Vlach J. (1998). Front.

Biosci., 3, D250 ±D268.Banks-Schlegel S and Green H. (1981). J. Cell. Biol., 90,

732 ± 737.Bata-Csorgo Z, Hammerberg C, Voorhees JJ and Cooper

KD. (1993). J. Exp. Med., 178, 1271 ± 1281.Ben-Bassat H, Rosenbaum-Mitrani S, Hartzstark Z, Shlo-

mai Z, Kleinberger-Doron N, Gazit A, Plowman G,Levitzki R, Tsvieli R and Levitzki A. (1997). Cancer Res.,57: 3741 ± 3750.

Chicurel ME, Chen CS and Ingber DE. (1998). Curr. Opin.Cell. Biol., 10: 232 ± 239.

Conlon I and Ra� M. (1999). Cell, 96: 235 ± 244.Dang CV. (1999). Mol. Cell. Biol., 19: 1 ± 11.Di Cunto F, Topley G, Calautti E, Hsiao J, Ong L, Seth PK

and Dotto GP. (1998). Science, 280, 1069 ± 1072.Dover R and Watt FM. (1987). J. Invest. Dermatol., 89,

349 ± 352.Edgar B. (1995). Curr. Opin. Cell. Biol., 7, 815 ± 824.Evan GI, Lewis GK, Ramsay G and Bishop JM. (1985). Mol.

Cell. Biol., 5, 3610 ± 3616.Frisch SM and Francis H. (1994). J. Cell. Biol., 124, 619 ±

626.Fuchs E. (1993). J. Cell. Sci. (Suppl.), 17, 197 ± 208.Fuchs E and Byrne C. (1994). Curr. Opin. Genet. Dev., 4,

725 ± 736.Gandarillas A. (2000). Exp. Gerontol., 35, 53 ± 62.Gandarillas A and Watt FM. (1995). Oncogene, 11, 1403 ±

1407.Gandarillas A and Watt FM. (1997). Genes Dev., 11, 2869 ±

2882.Gandarillas A, Goldsmith LA, Gschmeissner S, Leigh IM

and Watt FM. (1999). Exp. Dermatol., 8, 71 ± 79.Gra® G. (1998). Exp. Cell. Res., 244, 372 ± 378.Grandori C and Eisenman RN. (1997). Trends Biochem. Sci.,

22, 177 ± 181.Hall PA and Watt FM. (1989). Development, 106, 619 ± 633.

Harvat BL, Wang A, Seth P and Jetten AM. (1998). J. Cell.Sci., 111, 1185 ± 1196.

Hashiro M, Matsumoto K, Okumura H, Hashimoto K andYoshikawa K. (1991). Biochem. Biophys. Res. Commun.,174, 287 ± 292.

Hauser PJ, Agrawal D, Flanagan M and Pledger WJ. (1997).Cell Growth Di�er., 8, 203 ± 211.

Hawkins NC, Thorpe J and Schupbach T. (1996). Develop-ment, 122, 281 ± 290.

Henriksson M and LuÈ scher B. (1996). Adv. Cancer Res., 68,109 ± 182.

Ho�man R. (1989). Blood, 74, 1196 ± 1212.Hudson DL, Speight PM and Watt FM. (1996). Int. J.

Cancer, 66, 457 ± 463.Hulskamp M, Schnittger A and Folkers U. (1999). Int. Rev.

Cytol., 186, 147 ± 178.Hurlin PJ, Foley KP, Ayer DE, Eisenman RN, Hanahan D

and Arbeit JM. (1995). Oncogene, 11, 2487 ± 2501.Iritani BM and Eisenman RN. (1999). Proc. Natl. Acad. Sci.

USA, 96, 13180 ± 13185.Jack EM, Bentley P, Bieri F, Muakkassah-Kelly SF, Staubli

W, Suter J, Waechter F and Cruz-Orive LM. (1990).Hepatology, 11, 286 ± 297.

Johnston LA, Prober DA, Edgar BA, Eisenman RN andGallant P. (1999). Cell, 98, 779 ± 790.

Jones PH and Watt FM. (1993). Cell, 73, 713 ± 724.Judware R and Culp LA. (1997). Oncogene, 14, 1341 ± 1350.Kartasova T, Roop DR and Yuspa SH. (1992). Mol.

Carcinog., 6, 18 ± 25.Kirk D and Clingan DA. (1980). Cell. Biol. Int. Rep., 4, 83 ±

92.Kung AL, Sherwood SW and Schimke RT. (1990). Proc.

Natl. Acad. Sci. USA, 87, 9553 ± 9557.Lavker RM and Sun TT. (1983). J. Invest. Dermatol., 81,

121S ± 127S.Li Q and Dang CV. (1999). Mol. Cell. Biol., 19, 5339 ± 5351.Littlewood TD, Hancock DC, Danielian PS, Parker MG and

Evan GI. (1995). Nucleic Acids Res., 23, 1686 ± 1690.

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3288

Oncogene

Martinez LA, Chen Y, Fischer SM and Conti CJ. (1999).Oncogene, 18, 397 ± 406.

Meyerowitz EM. (1994). Curr. Opin. Genet. Dev., 4, 602 ±608.

Missero C, Di Cunto F, Kiyokawa H, Ko� A and Dotto GP.(1996). Genes Dev., 10, 3065 ± 3075.

Morasso MI, Markova NG and Sargent TD. (1996). J. Cell.Biol., 135, 1879 ± 1887.

Nakatani S, Okada N, Okumura H and Yoshikawa K.(1992). Arch. Dermatol. Res., 284, 349 ± 352.

Neufeld TP and Edgar BA. (1998). Curr. Opin. Cell. Biol., 10,784 ± 790.

Ormerod EM. (1990). Flow Cytometry: A practical Ap-proach. Omerod EM (ed). IRL Press: Oxford, pp 69 ± 87.

Paramio JM, Casanova ML, Segrelles C, Mittnacht S, LaneEB and Jorcano JL. (1999). Mol. Cell. Biol., 19, 3086 ±3094.

Pelengaris S, Littlewood T, Khan M, Elia G and Evan G.(1999). Mol. Cell., 3, 565 ± 577.

Penneys NS, Fulton Jr JE, Weinstein GD and Frost P.(1970). Arch. Dermatol., 101, 323 ± 327.

Pietenpol JA, Stein RW, Moran E, Yaciuk P, Schlegel R,Lyons RM, Pittelkow MR, Munger K, Howley PM andMoses HL. (1990). Cell, 61, 777 ± 785.

Pinkus H and Hunter R. (1966). Arch. Dermatol., 94, 351 ±354.

Regnier M, Vaigot P, DarmonM and Prunieras M. (1986). J.Invest. Dermatol., 87, 472 ± 476.

Reiss M and Sartorelli AC. (1987). Cancer Res., 47, 6705 ±6709.

Rheinwald JG. (1989). Cell Growth and Division: A practicalapproach. Baserga R (ed). IRL Press: Oxford, pp 81 ± 94.

Schuhmacher M, Staege MS, Pajic A, Polack A, Weidle UH,Bornkamm GW, Eick D and Kohlhuber F. (1999). Curr.Biol., 9, 1255 ± 1258.

Solari F, Flamant F, Cherel Y, Wyers M and Jurdic P.(1996). J. Cell. Sci., 109, 1203 ± 1213.

Traas J, Hulskamp M, Gendreau E and Hofte H. (1998).Curr. Opin. Plant. Biol., 1, 498 ± 503.

Waikel RL, Wang XJ and Roop DR. (1999). Oncogene, 18,4870 ± 4878.

Watt FM. (1989). Curr. Opin. Cell. Biol., 1, 1107 ± 1115.Watt FM and Green H. (1981). J. Cell. Biol., 90, 738 ± 742.Watt FM and Jones PH. (1993). Dev. (Suppl), 185 ± 192.Watt FM, Hudson DL, Lamb AG, Bolsover SR, Silver RA,

Aitchison MJ and Whitaker M. (1991). J. Cell. Sci., 99,397 ± 405.

Watt FM, Jordan PW and O'Neill CH. (1988). Proc. Natl.Acad. Sci. USA, 85, 5576 ± 5580.

Oncogene

c-Myc and endoreplication in keratinocyte differentiationA Gandarillas et al

3289