Opposing activities of c-Fos and Fra-2 on AP-1 regulated transcriptionalactivity in mouse keratinocytes induced to di�erentiate by calcium andphorbol esters

Susan E Rutberg, Enrique Saez2, Sam Lo, Shyh-Ing Jang3, Nedialka Markova3,Bruce M Spiegelman2 and Stuart H Yuspa1

1Laboratory of Cellular Carcinogenesis and Tumor Promotion, National Cancer Institute, Bethesda, Maryland 20892;2Dana-Farber Cancer Institute and the Department of Cell Biology, Harvard Medical School, 44 Binney Street, Boston,Massachusetts 02115; 3Laboratory of Skin Biology, National Institute of Arthritis and Musculoskeletal and Skin Diseases,Bethesda, Maryland, USA

The major di�erentiation products of maturing kerati-nocytes contain AP-1 regulatory motifs, and AP-1 DNAbinding activity increases in cultured keratinocytesinduced to di�erentiate by calcium. Here, we haveanalysed AP-1 transcriptional activity in mouse kerati-nocytes treated with calcium and 12-O-tetradecanoylphorbol-13-acetate (TPA), two agents that induceterminal di�erentiation of keratinocytes with di�erentphenotypic consequences. Reporter constructs represent-ing multimers of AP-1 sequences found in keratinocytemarker genes demonstrated that the calcium-inducedAP-1 DNA binding activity does not correlate withtranscriptional activation. Moreover, expression fromactive subunits of the pro®laggrin and spr 1 promotersincreased in calcium-treated keratinocytes when the AP-1 sites were disrupted, indicating that AP-1 maynegatively regulate certain promoters in these cells. Incontrast, AP-1 reporter activity was increased inkeratinocytes treated with TPA. This induction wasdependent upon the expression of c-Fos since AP-1transcriptional activity was not increased in TPA-treatedkeratinocytes derived from c-fos null mice. Analysis ofAP-1 protein expression in calcium- and TPA-treatedkeratinocytes demonstrated that only TPA increased theexpression of c-Jun, while Jun B and Jun D wereinduced by both of these agents. c-Fos was expressedonly in TPA treated keratinocytes, Fra-2 was expressedonly in calcium-treated cells, and Fra-1 was expressed inboth. Exogenous expression of Fra-2 repressed AP-1transcriptional activity in TPA-treated keratinocytes,while c-Fos expression activated the AP-1 sequence incalcium-treated keratinocytes. These data indicate thatFra-2 and c-Fos play opposing roles in regulating AP-1activity in keratinocytes and that multiple inducer-dependent regulatory pathways may exist for theexpression of keratinocyte di�erentiation markers.

Keywords: c-Fos; Fra-2; mouse keratinocytes; differ-entiation; calcium; phorbol ester

Introduction

Epidermal di�erentiation is characterized by theinduction and repression of marker genes that de®ne

speci®c stages of keratinocyte maturation (reviewed inYuspa, 1994). In mouse and human skin, theproliferative basal layer consists of cells expressingtranscripts encoding keratins 5 and 14. As keratino-cytes withdraw from the cell cycle, they undergospinous di�erentiation, characterized by the down-regulation of keratins 5 and 14, and the induction oftranscripts encoding keratins 1 and 10 (K1 and K10)(Roop et al., 1988; Stoler et al., 1988; Fuchs, 1990).Transcripts encoding K1 and K10 are subsequentlydown-regulated in the granular layer, where the latemarker genes loricrin and ®laggrin are induced (Roopet al., 1988; Fuchs, 1990). In addition, keratinocytetransglutaminases are activated in the granular layer tocatalyze the formation of corni®ed envelopes (reviewedin Polakowska and Goldsmith, 1991), insolublestructures composed of precursor proteins involucrin(Rice and Green, 1979), small proline rich (spr)proteins (Kartasova et al., 1987; Marvin et al., 1992),loricrin (Mehrel et al., 1990), and other components(Steinert and Merekov, 1995).Many of the changes in marker gene expression that

occur in the di�erentiating epidermis can be recapitu-lated in mouse keratinocytes maintained in vitro byincreasing the level of calcium in the medium from 0.05to 0.12 mM (Yuspa et al., 1989). For example, underthese conditions transcripts encoding keratins 1 and 10are induced within 24 h after raising the calciumconcentration. By 48 h, these transcripts are down-regulated while those encoding the late markers loricrin,®laggrin, spr 1 and transglutaminase are induced.Terminal di�erentiation of mouse keratinocytes canalso be induced by 12-O-tetradecanoyl phorbol-13-acetate (TPA) (Yuspa et al., 1982; Parkinson et al.,1984). TPA-treated keratinocytes in vitro express highlevels of the late markers loricrin and ®laggrin (Dlugoszand Yuspa, 1993), produce activated transglutaminase(Lichti and Yuspa, 1988), and undergo an e�cientprocess of corni®cation (Yuspa et al., 1982). Yet, thesecells do not express the early marker genes K1 and K10seen in calcium-treated cells (Roop et al., 1987;Toftgard et al., 1985). Analysis of cellular proteinsproduced by calcium- and TPA-treated keratinocytesindicated similarities and di�erences in protein expres-sion and phosphorylation patterns in response to theseagents, indicating overlap and divergence in thepathways by which calcium and TPA induce terminaldi�erentiation (Wirth et al., 1987).The changes in marker gene expression that occur in

di�erentiating keratinocytes in vitro are dependentCorrespondence: SH YuspaReceived 25 March 1997; revised 19 May 1997; accepted 19 May 1997

Oncogene (1997) 15, 1337 ± 1346 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00

upon the activation of protein kinase C (PKC)(Dlugosz and Yuspa, 1993, 1994). Since the AP-1transcription factor complex is a well-de®ned nucleartarget of PKC activation (reviewed in Hill andTreisman, 1995), our current studies are focused uponunderstanding the role of AP-1 in the regulation ofevents related to keratinocyte di�erentiation. The AP-1transcription factor complex is composed of hetero-dimers formed between members of the Jun and Fosfamilies that bind to DNA in a sequence-speci®cmanner (reviewed in Curran and Franza, 1988; Angeland Karin, 1991). The Jun family consists of threeproteins (c-Jun, Jun B and Jun D) while the Fos familyis composed of four proteins (c-Fos, Fos B, Fra-1 andFra-2) that require heterodimerization with Junproteins or other members of the bZIP superfamilyof DNA binding proteins (Johnson and McKnight,1989) to bind to DNA. The Fos proteins contributedistinct functions towards the activity of the AP-1heterodimer. For example, c-Fos can both activate andrepress transcription (Gius et al., 1990), the full-lengthFos B is a transcriptional activator (Dobrazanski et al.,1991) and a naturally occurring short form of Fos Binhibits AP-1 transactivation (Yen et al., 1991;Nakabeppu and Nathans, 1991). The Fos-relatedantigens Fra-1 and Fra-2 lack functional transactiva-tion domains and are poor transcriptional activators(Gius et al., 1990; Yoshioka et al., 1995; Suzuki et al.,1991).The consensus AP-1 DNA binding site, also known

as the TPA-responsive element (TRE) (Angel et al.,1987), is prevalent in regulatory positions of manygenes expressed during keratinocyte di�erentiation,suggesting that AP-1 may play an important role incontrolling gene expression during this process. AP-1DNA binding sites have been shown to play animportant role in cell- and tissue-speci®c expressionof the loricrin (DiSepio et al., 1995), involucrin (Crishet al., 1993; Welter et al., 1995), keratin 5 (Casatorreset al., 1994) and keratin 1 (Rothnagel et al., 1993; Luet al., 1994) genes in vitro and in vivo. Additional invitro studies indicate that AP-1 proteins largely play apositive role in regulating the expression of keratino-cyte genes that harbor AP-1 sites (Navarro et al., 1995;Yamada et al., 1994; Takahashi and Iizuka, 1993;DiSepio et al., 1995; Bernerd et al., 1993; Oshima etal., 1990), although in vivo studies will ultimately berequired to provide a de®nitive demonstration of therole of AP-1 in regulating gene expression in theepidermis. Since epidermal genes containing AP-1binding sites are both induced and suppressed duringthe di�erentiation program, the possibility exists thatthe AP-1 transcription factor complex may contributeboth positive and negative regulatory e�ects during thisprocess.We have shown that PKC-dependent AP-1 DNA

binding activity increases in keratinocytes induced todi�erentiate by the addition of calcium to the medium(Rutberg et al., 1996). The levels of AP-1 DNAbinding activity peak late in the di�erentiationprogram, and correlate with a PKC-dependentincrease in the expression of Fra-2, Jun B and JunD. Immunostaining with antibodies against AP-1proteins indicated that Fra-2 and Jun B are localizedin the granular layer of mouse epidermis, supportingthat these transcription factors are important during

the late stages of keratinocyte di�erentiation. Thepresent study demonstrates that the calcium-inducedincrease in AP-1 DNA binding activity fails to result intranscriptional activation and may negatively regulateAP-1 dependent gene expression in keratinocytes. TPAalso causes an increase in AP-1 DNA binding activityin keratinocytes, but in contrast to the e�ects ofcalcium, TPA enhances the expression of AP-1regulated reporter genes. The functional di�erencesbetween the AP-1 complexes induced by calcium orTPA can be attributed to di�erences in the composi-tion of the AP-1 DNA binding complex that is formedin the presence of each stimulus.

Results

TPA and calcium induce AP-1 DNA binding activity

Since PKC activation is a necessary event duringkeratinocyte di�erentiation and AP-1 is a knownnuclear target of protein kinase C activity, wecompared the levels of AP-1 DNA binding activity inkeratinocytes treated with calcium and TPA, two PKCactivators that induce keratinocyte di�erentiation invitro (Denning et al., 1995a; Yuspa et al., 1982, 1989;Parkinson et al., 1984). A comparison of the AP-1DNA binding complexes formed in TPA and calciumtreated keratinocytes indicated that while AP-1 DNAbinding activity was induced under both conditions,the binding complexes di�ered in migration patternand intensity (Figure 1). Nuclear extracts preparedfrom keratinocytes maintained for 48 h in 0.12 mM

calcium contained a large and intense, slowly-migratingAP-1 DNA binding complex. In contrast, the TPA-inducible AP-1 DNA binding complex was moredi�use, and appeared to consist of multiple complexesthat formed very close to one another, producing asmear on the gel. These complexes contrast with thevery weak AP-1 DNA binding activity observed inbasal keratinocytes (maintained in 0.05 mM calcium).The complexes formed under each of these conditionswere speci®c for the AP-1 site, as determined bycompetition with 100-fold excess speci®c (AP-1) andnon-speci®c (SP-1) DNA sequences. Previous dataindicated that the calcium-inducible AP-1 DNAbinding complex consists of c-Jun, Jun B, Jun D,Fra-1 and Fra-2 (Rutberg et al., 1996); supershiftanalysis indicated that the AP-1 DNA binding complexin TPA-treated keratinocytes consisted of c-Jun, Jun Band Jun D, c-Fos and Fra-1 (Figure 2).

Calcium-induced keratinocyte di�erentiation correlateswith decreased AP-1 transcriptional activity

To assess the functional role of the calcium-inducedincrease in AP-1 DNA binding, we evaluated thetranscriptional activity of a multimer of the consensusAP-1 target binding squence (56TRE-CAT) (Angel etal., 1987) in basal and di�erentiating keratinocytes. Inaddition, AP-1 sequences from the calcium-responsiveelement of the human keratin 1 gene (gaTGATTCAct)(Hu� et al., 1993; Rothnagel et al., 1993), fromposition 777 of the 5' region of the humanpro®laggrin promoter (gaTGAATCAct) (Markova etal., 1993), and from one of the AP-1 sites

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1338

(gtTGACTAAcg) at position 7288 of the 5' non-coding region of the human involucrin gene (Lopez-Bayghen et al., 1996) were also cloned as 56multimers into the same pBLCAT2 reporter construct(Figure 3a). Supershift analysis indicated that thecomposition of the AP-1 complexes bound to thesesequences in 0.12 mM calcium was the same (notshown). The AP-1 site in the 5' region of the loricringene is identical to the consensus AP-1 sequence in the56 TRE vector, and is referred to as the loricrin AP-1site in this ®gure.A modest time-dependent decrease in AP-1 tran-

scriptional activity was seen with three of the four AP-1 reporter vectors in keratinocytes induced todi�erentiate by the addition of calcium to the medium(loricrin, HK1 and involucrin), indicating that calcium-

induced keratinocyte di�erentiation in vitro correlateswith a decrease in AP-1 transcriptional activity (Figure3b). While the pro®laggrin AP-1 site remaineduna�ected by the addition of calcium to the medium,none of the AP-1 sequences tested demonstratedincreased activity in response to calcium. Transfectedcells harvested 54 ± 60 h after the addition of calciumexhibited similar levels of AP-1 activity as thosecollected at 48 h (data not shown).

AP-1 negatively regulates the pro®laggrin and spr 1promoters in di�erentiating keratinocytes

Because the analysis of isolated sequence elementscloned in front of a reporter gene can sometimes o�eran incomplete picture of transcription factor activity,we examined whether the apparent lack of activationpotential of the AP-1 transcription complex incalcium-treated keratinocytes was relevant to theexpression of the pro®laggrin and spr 1 genes indi�erentiating keratinocytes. Promoter constructscontaining the ®rst 116 base pairs of the pro®laggrinpromoter were generated in which the AP-1 site atposition 777 was maintained in the wild-typecon®guration (7116 wild-type) or deleted (7116mutant, Figure 4a, upper panel). The activity ofthese constructs in basal and di�erentiating keratino-cytes was compared with that of the core promotercontaining the ®rst 59 base pairs of the promoteralone (759 CAT). As shown in Figure 4b, the activityof the mutant construct was similar to that of thewild-type in basal keratinocytes. Upon addition of

Pro

be

No

ne

AP

-1

SP

-1

No

ne

AP

-1

SP

-1

No

ne

AP

-1

SP

-1

Competitor DNA

0.05 mM calcium

162nM TPA 4 hours

0.12 mM calcium

Figure 1 AP-1 DNA binding activity increases in keratinocytestreated with calcium and TPA. Nuclear extracts prepared fromprimary mouse keratinocytes maintained in 0.05 mM calcium,with or without 4 h of TPA-treatment, and from keratinocytesmaintained in 0.12 mM calcium for 48 h were incubated with a32P-labeled probe bearing the AP-1 target binding sequence inelectrophoretic mobility shift assay. To establish the speci®city ofDNA binding reaction, the extracts were also incubated with a100-fold excess competitor DNA representing wild-type AP-1 orSP-1 target sequences

No

An

tib

od

y

c-Fo

s

Fos

B

Fra-

1

Fra-

2

c-Ju

n

Jun

B

Jun

D

super-shifted complexes

Figure 2 Proteins that bind to the AP-1 target sequence in TPA-treated keratinocytes. Supershift analysis was performed withextracts from TPA-treated keratinocytes. The speci®c antibodiesused in this analysis are indicated, as well as the position of thesuper-shifted complexes

a

b

Figure 3 Calcium negatively regulates various AP-1 sequences indi�erentiating keratinocytes. (a) Variations of the AP-1 bindingsite found in keratinocyte genes cloned as 56 multimers into thepBLCAT2 plasmid. The loricrin construct is identical to the56TRE-CAT vector described by (Angel et al., 1987). (b)Primary mouse keratinocytes were transfected with the vectorsdescribed in Figure 3a. Cells were maintained in 0.05 mM calciumfor 24 h, or in 0.12 mM calcium for 24 and 48 h. Values +s.e.m.are expressed relative to the activity of the parent pBLCAT vectoralone. These data were replicated in three separate experiments

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1339

calcium to the medium, the activity of the wild-typeconstruct increased ®vefold within 24 h while theabsence of the AP-1 site at position 777 resulted in atenfold increase in CAT activity over that seen inbasal cells. These results suggest that the factorsbinding to the AP-1 site on the pro®laggrin promoterinterfere with positive regulation or negatively regulatethis portion of the pro®laggrin promoter in keratino-cytes induced to di�erentiate by calcium. The activityof the mutant construct remained approximatelyfourfold higher than that of the wild-type reporter60 h after the addition of calcium, a time when theactivity of the wild-type and 759 constructs haddecreased.Promoter constructs from a second keratinocyte

gene were used to support the ®ndings obtained withthe pro®laggrin reporter constructs. A 66 base pairregion of the spr 1 promoter, which contains two AP-1 sites in close proximity, was used to generate wild-type and mutant constructs in which each of the AP-1sequences was mutated alone or in combination, asdescribed in Reddy et al., 1996 (see also Figure 4a,lower panel). The 5' AP-1 DNA binding site consistsof a consensus AP-1 binding sequence 5'-gtTGAGT-CAac-3' while the 3' site bears the sequence 5'-ggTGAATCAca-3'. As shown in Figure 4c, theactivity of the 5' mutant (TtcGTCA) was 2.5-foldgreater than the wild-type construct in basalkeratinocytes and remained 2.5-fold greater thanwild-type in di�erentiated keratinocytes 24 and 60 hafter the addition of calcium to the medium. Thus,this site appears to act as a repressor of geneexpression in both basal and di�erentiating cells. Incontrast, the 3' mutant (TaAATCA) demonstrateddecreased activity compared to the wild-type constructin basal keratinocytes, suggesting that this AP-1 siteenhances expression of this region of the spr 1promoter. Moreover, the 3' mutant was calciuminducible, suggesting that the 3' AP-1 site is notrequired for calcium inducibility of this portion of thespr 1 promoter. The activity of the double mutantappeared to be intermediate between that of the singlemutants. These results indicate that (1) negativeregulation by AP-1 occurs in a sequence-speci®cmanner, and (2) that the function (positive ornegative) of a given AP-1 DNA binding sequence islikely to depend upon the context of the promoter onwhich it is found. This is inferred from the fact thatthe 3' AP-1 DNA binding sequence on the spr 1promoter is identical to that on the pro®laggrinconstruct, yet alteration of this sequence yieldeddi�erent results within the two promoters tested.These di�erences may also re¯ect consequences ofmutating (spr 1) or deleting (pro®laggrin) the AP-1site rather than functional di�erences between theidentical AP-1 sites on the two di�erent promoters.

Calcium and TPA have opposing e�ects on AP-1transcriptional activity in keratinocytes

To examine whether the increase in AP-1 DNAbinding activity in response to TPA-treatment ofkeratinocytes serves a similar regulatory function asthat induced by calcium, keratinocytes transfected withthe 56 TRE-CAT reporter (loricrin AP-1 site) weretreated with 162 nM TPA or 0.12 mM calcium and

analysed 24 and 48 h later, respectively. A comparisonof the e�ect of calcium and TPA on AP-1-CAT activityin these cells indicated that, in contrast to the decreasein AP-1 dependent activity in calcium-treated cells,TPA treatment increased AP-1 transcriptional activity

a

b

c

Figure 4 Disruption of AP-1 regulatory sequences increases theactivity of the pro®laggrin and spr 1 promoters in di�erentiatingkeratinocytes. (a) The pro®laggrin and spr 1 promoter constructsused in this study as described in (Jang et al., 1996) and (Reddy etal., 1996). (b) Primary keratinocytes were transfected with the7116 wild-type AP-1, 7116 mutant AP-1, and 759 corepro®laggrin promoter constructs described in a. Cells weremaintained in 0.05 mM calcium for 24 h, or in 0.12 mM calciumfor 24 or 60 h. Values reported represent mean+s.e.m. fortriplicate dishes from a representative experiment. (c) Primarykeratinocytes were transfected with a 66 base pair region of thespr 1 promoter in the wild-type con®guration, or in which the 3'and 5' AP-1 sequences have been mutated alone or together(double mutant). Cells were maintained in 0.05 for 24 h or0.12 mM calcium for 24 or 60 h. Values reported representmean+s.e.m. for triplicate dishes from a representative experi-ment

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1340

threefold (Figure 5a). This observation held with eachof the 56 multimers of the AP-1 sequence described inFigure 3a, all of which were induced 2 ± 4-fold by theaddition of TPA to the medium and similar levels ofinduction were observed in cells maintained in 8% and0.5% serum (data not shown). The simultaneousaddition of calcium and TPA to the cells resulted inlevels of AP-1-CAT activity that were intermediatebetween those in 0.12 mM calcium and 162 nM TPA(data not shown). TPA-inducible AP-1-CAT activityhas been attributed to the induction, and/or modifica-tion of c-Fos and c-Jun (Greenberg and Zi�, 1984;Lamph et al., 1988; Devary et al., 1991). To establishwhether c-Fos plays a role in TPA-mediated AP-1-CAT activity in keratinocytes, AP-1 transcriptionalactivity was evaluated in c-fos null keratinocytestreated with TPA. In contrast to the increase in AP-1-CAT activity caused by TPA in wild-type cells, noTPA-related induction of AP-1-CAT activity was seenin c-fos null keratinocytes, indicating that TPA-mediated activation of AP-1-CAT activity in keratino-cytes is dependent upon the presence of c-Fos (Figure5b).The pro®laggrin promoter constructs used in Figure

4b were evaluated in TPA-treated keratinocytes toestablish whether their activities would be altered underconditions where c-Fos is expressed. As shown inFigure 5c, the wild-type reporter was induced 2.5-foldin the presence of TPA supporting the increasedexpression of the pro®laggrin message seen in TPAtreated keratinocytes (Dlugosz and Yuspa, 1993). Incontrast to the increased activity of the AP-1 mutantconstruct in calcium-treated cells, the activity of themutant promoter was not increased upon TPAtreatment. Thus, AP-1 proteins appear to play acritical role in activating this portion of the profilag-grin promoter in TPA-treated keratinocytes.

c-Fos and Fra-2 are di�erentially regulated by calciumand TPA in keratinocytes

The opposing functional roles of the AP-1 DNAbinding activity induced by calcium and TPA couldbe due to di�erences in the composition of the AP-1complex induced by each stimulus. To evaluate thedi�erences in AP-1 protein expression in response tocalcium and TPA, we performed Western blot analysisusing antibodies speci®c for known members of theAP-1 family (Figure 6). TPA caused an increase in thelevels of c-Jun expression. This is in contrast to theslightly decreased levels of c-Jun in calcium-treatedkeratinocytes as shown here and previously (Rutberg etal., 1996). In contrast, Jun B and Jun D were inducedto a similar extent by calcium and TPA. Additionallower molecular weight Jun B-reactive bands that arepartially sensitive to alkaline phosphatase (data notshown) were detected with antibodies against Jun B inTPA-treated cells, indicating that potential di�erencesin post-translational modi®cation of this protein mayexist between calcium and TPA treated keratinocytes.The appearance of additional lower molecular weightJun B-reactive bands in response to TPA treatment hasbeen described previously (Au et al., 1994). The41 kDa Jun D-reactive band observed in calcium andTPA-treated cells is a naturally occurring truncatedform of the protein (Rezzonico et al., 1995). With

regard to the Fos family proteins, c-Fos was inducedonly in TPA treated keratinocytes: no c-Fos proteinwas detected in keratinocytes maintained in 0.05 mM

or 0.12 mM calcium, consistent with the absence of aphenotype related to c-fos de®ciency in keratinocytes invitro (Rutberg et al., 1996) or in newborn skin from c-fos null mice (Saez et al., 1995). In contrast, Fra-2 wasexpressed exclusively in keratinocytes induced todi�erentiate by the addition of calcium to themedium. The 48 kDa Fra-2 reactive band in Figure 6represents the phosphorylated form of Fra-2 (Gruda etal., 1994). The 43 kDa unphosphorylated form of Fra-1 was expressed at equal levels in basal keratinocytesand in keratinocytes treated with calcium or TPA. TPA

a

b

c

Figure 5 TPA activates AP-1 transcriptional activity inkeratinocytes. (a) Primary mouse keratinocytes transfected witha 56 multimer of the TRE binding sequence were maintained in0.05 or 0.12 mM calcium for 48 h (a) or 0.05 mM calcium with orwithout 162 nM TPA for 24 h (b). TRE-CAT activity relative tothe parent pBLCAT plasmid is shown. The data represent+s.e.m. for three separate experiments. (b) Primary keratinocytesfrom c-fos null and wild-type mice transfected with the 56TRE-CAT were harvested 24 h after the addition of 162 nM TPA.TRE-CAT activity relative to pBLCAT is shown and representsmean+s.e.m. for three separate experiments. (c) Primary mousekeratinocytes transfected with the wild-type and mutated forms ofthe pro®laggrin promoter were maintained in 0.05 mM calciumwith or without the addition of 162 nM TPA to the medium. Thevalues reported represent mean and s.e.m. for triplicate dishesfrom a representative experiment

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1341

caused the phosphorylation of Fra-1, as detected bythe appearance of a 46 kDa Fra-1 protein in thesecells.

c-Fos and Fra-2 have di�erent e�ects on AP-1transcriptional activity in keratinocytes

In order to determine the e�ects of increasing theexpression levels of c-Fos and Fra-2 on the levels ofAP-1-CAT activity in TPA-treated keratinocytes,expression vectors encoding mouse Fra-2, c-Fos andJun B cDNAs were co-transfected (together andseparately) with the 56 TRE-CAT reporter constructinto keratinocytes which were then treated with TPA(Figure 7a). Jun B was chosen as an appropriatedimerization partner for c-Fos and Fra-2 since Jun Bexpression was increased in both calcium and TPA-treated cells (Figure 6). The ability of the expressionvectors to produce high levels of protein within 24 hwas con®rmed through Western blot analysis ofnuclear extracts prepared from transfected cells (notshown). This approach indicated that neither exogen-ous Jun B nor c-Fos alone were able to increase thelevels of AP-1-CAT activity above that seen in TPA-treated control cells transfected with the parent

constructs. However, the combination of Jun B andc-Fos expression vectors increased AP-1-CAT activityabove that of the controls indicating that the c-Fos/JunB heterodimer formed by the forced expression of thesetwo factors is a transcriptionally active complex. Incontrast, Fra-2 alone, or expressed in combination withJun B, negatively regulated AP-1-CAT activity in TPAtreated keratinocytes indicating that Fra-2 is a power-ful negative transcriptional regulator in these cells.These data support that di�erences in the expressionpatterns of c-Fos and Fra-2 may explain the oppositelevels of AP-1-dependent transactivation in keratino-cytes treated with calcium or TPA.To establish whether the reduced levels of AP-1-

dependent transcriptional activity measured in differ-entiating keratinocytes could be elevated by theexpression of c-Fos alone, c-Fos and Fra-2 expressionvectors were co-transfected with the 56TRE-CATreporter into basal keratinocytes which were then

39

4543

4641

62

46

43

48

c-Jun

Jun B

Jun D

c-Fos

Fra-1

Fra-2

– + – 0.05 0.05 0.12

162 nM TPA mM calcium

Figure 6 Expression of AP-1 proteins in TPA- and calcium-treated keratinocytes. Western analysis using speci®c AP-1antibodies was performed with nuclear extracts from keratino-cytes maintained in 0.05 or 0.12 mM calcium for 48 h, or 162 nMTPA for 4 h. The molecular weights of the proteins of interest areindicated by the arrows

a

b

Figure 7 Fra-2 and c-Fos have di�erent e�ects on AP-1transcriptional activity in keratinocytes. (a) Expression vectorsencoding Jun B, Fra-2 and c-Fos were co-transfected with the56TRE-CAT multimer into keratinocytes which were thentreated with TPA and analysed 24 h later. Control cells wereco-transfected with the parental expression vectors pCMV andpSG5. TRE-CAT activity was evaluated relative to the activity ofthe parent pBLCAT2 vector alone. (b) Fra-2 and c-Fos expressionvectors wre co-transfected with the TRE-CAT vector intokeratinocytes which were then treated with 0.12 mM calciumand analysed 24 later. Control cells were co-transfected with theparental expression vectors pCMV and pSG5. The valuespresented in a and b represent mean+s.e.m. for TRE-CATactivity relative to the pBLCAT control

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1342

treated with 0.12 mM calcium to induce di�erentiation(Figure 7b). While exogenous Fra-2 expressiondecreased the levels of AP-1 transcriptional activity inthese cells, the forced expression of c-Fos increasedAP-1 transcriptional activity 3.5-fold in calcium-treatedcells, indicating that aberrant expression of c-Fos indi�erentiating keratinocytes can reverse the negativetranscriptional regulation normally imposed by the AP-1 complexes present in these cells.

Discussion

In this study we demonstrate that the opposing e�ectsof calcium and TPA on the regulation of AP-1transcriptional activity in di�erentiating keratinocytescorrelate with di�erential expression of Fra-2 and c-Fos. The exogenous expression of Fra-2 alone, or inconjunction with Jun B, results in negative regulationof AP-1 reporter constructs indicating that the Fra-2/Jun B heterodimer may be responsible for thedecreased AP-1 transcriptional activity observed incalcium-treated keratinocytes. In contrast, TPAtreatment enhances AP-1 transcriptional activity inkeratinocytes, correlating with the induction of c-Fosexpression and the ability of c-Fos to bind to the AP-1 target sequence. Furthermore, exogenous expressionof c-Fos alone is su�cient to elevate AP-1 transcrip-tional activity in calcium-treated keratinocytes. To-gether, these data indicate that Fra-2 and c-Fos haveopposite e�ects on AP-1 transcriptional activity inkeratinocytes.Multimers of the AP-1 target sequence linked to the

CAT reporter gene demonstrated a time-dependentdecrease in transcriptional activity in keratinocytesinduced to di�erentiate by calcium. This decreaseoccurs in concert with the formation of a strong AP-1 DNA binding complex that peaks in intensity 36 ±48 h after the addition of calcium to the medium(Rutberg et al., 1996) suggesting that this complex mayserve to negatively regulate transcription in differentiat-ing keratinocytes. The use of wild-type and mutatedforms of portions of the pro®laggrin and spr 1promoters support this notion since disruption of AP-1 sites on these constructs lead to increased activity ofthe reporters in calcium-treated keratinocytes. Assaysutilizing the spr 1 promoter constructs indicated thatnegative transcriptional regulation mediated by AP-1occurs in a sequence speci®c manner, since of the twoAP-1 sites present on the construct, only the 5'consensus AP-1 site exhibited this e�ect. Our dataalso suggest that the regulatory e�ects imposed by agiven AP-1 sequence may be context-dependent sincethe 3' AP-1 sequence on the spr 1 promoter and theAP-1 site on the pro®laggrin promoter construct areidentical in sequence, but appear to mediate oppositeregulatory e�ects within the di�erent promoterconstructs. In a similar fashion, AP-1 proteins havebeen shown to bind to a region of the involucrinpromoter that functions as a transcriptional silencer incalcium-treated human keratinocytes maintained in theabsence of EGF and bovine pituitary extract whileother AP-1 sites on the promoter acted as transcrip-tional activators in these cells (Lopez-Bayghen et al.,1996). Together, these data support that location- orsequence-speci®c factors may in¯uence the activity of

AP-1 DNA binding proteins within the context of aparticular promoter region.It is possible that negative regulation exerted by AP-

1 proteins requires interactions with factors binding tonearby sites on the DNA that di�er between theregulatory regions of di�erent genes, or that variationsin the ¯anking sequences surrounding a given AP-1 sitefavor the binding of particular AP-1 heterodimers withdi�erent transcriptional potential. The mechanism bywhich speci®c AP-1 proteins (such as Fra-2) decreasethe transcriptional activity of the pro®laggrin and spr 1promoters in di�erentiating keratinocytes may involvesquelching the activities of positive-acting AP-1 factors,or blocking the function of unrelated transcriptionalactivators through stearic hindrance at the AP-1 targetsite. An Ets binding site overlaps with the AP-1 DNAbinding sequence at position 777 of the pro®laggrinpromoter (Jang et al., 1996). It is possible that theapparent negative regulation of this site in differentiat-ing keratinocytes results from interactions between AP-1 and Ets transcription factors. Alternatively, negativeregulation by AP-1 may re¯ect a combination ofchanges in the relative expression levels, DNA bindingactivity and/or phosphorylation status of particularAP-1 proteins, all factors that contribute to the levelsof AP-1 transcriptional activity. Since our data iscompatible with these and other mechanisms, furtherstudies will be required to unravel how AP-1 proteinsnegatively regulate epidermal-speci®c genes.Our work shows that the induction of the portion of

the pro®laggrin promoter used here occurs throughdi�erent mechanisms in response to calcium and TPA.Although both calcium and TPA induced thepro®laggrin promoter construct to a similar extent,the calcium-mediated induction of the promoteroccurred independently of AP-1, perhaps through theaction of other transcriptional regulators with bindingsites on this region of the promoter (Jang et al., 1996).In contrast, the induction of the pro®laggrin promoterconstruct in TPA-treated cells appeared to be entirelydependent upon an intact AP-1 site. It is interestingthat the silencing e�ects of the AP-1 sequences on theinvolucrin promoter in calcium-treated human kerati-nocytes described above (Lopez-Bayghen et al., 1996)were not apparent in TPA-treated human keratinocytes(Welter et al., 1995). It is likely that c-Fos is inducedunder these conditions, and that the silencing e�ects ofother AP-1 family members are masked by thepresence of c-Fos.In light of these observations, it becomes clear that

the conditions under which keratinocytes are main-tained in culture signi®cantly in¯uence the expressionof speci®c AP-1 family members resulting in divergente�ects on gene transcription. The lack of constitutivec-Fos expression in primary mouse keratinocytes hasallowed us to analyse the function of other AP-1regulators, independent of c-Fos. This appears to beimportant for as shown in Figure 7b, the expression ofc-Fos is su�cient to overcome the down-regulation ofAP-1 activity that accompanies calcium-inducedkeratinocyte di�erentiation. Since many of the studiesinvestigating the role of AP-1 activity in vitro areperformed with human keratinocytes, it is tempting toassume that species di�erences contribute to thedi�erential e�ects of AP-1 in mouse and human cells.However, signi®cant di�erences exist between the

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1343

keratinocyte serum-free medium used in the main-tenance of human keratinocytes and the serum-containing medium used for the routine culture ofmouse keratinocytes. In particular, human cells arecommonly cultured in a cocktail of growth factorsincluding EGF and insulin. It is possible that underidentical culture conditions, the expression of AP-1proteins and the levels of AP-1 transcriptional activityin di�erentiating human and mouse keratinocyteswould be similar, the determining factor being theexpression, or lack of expression, of c-Fos.Our transcriptional assays indicate that the Fra-2/

Jun B heterodimer is a powerful negative transcrip-tional regulator in keratinocytes, ®ndings that areconsistent with previous reports (Sonobe et al., 1995;Suzuki et al., 1991). Down-regulation of transcriptionalactivity by the Fra-2/Jun B heterodimer in granularlayer cells could serve a variety of purposes during theformation of normal skin. In the transition from thespinous to the granular layer, the expression of genesinvolved in the early stages of epidermal di�erentiationmust be silenced and substituted with that of genesrelevant to the late stages of di�erentiation. Moreover,the transition from the granular to the corni®ed layer ischaracterized by the replacement of live cells withenucleated mature squames consisting only of thecomponents necessary for the formation of thecorni®ed envelope. This step involves the down-regulation of genes that are not relevant to cornifica-tion. The Fra-2/Jun B heterodimer may contribute toeither of these, or still other processes related to the®nal stages of epidermal di�erentiation. In vivo studieswill be required to establish the role of the di�erentAP-1 heterodimers expressed in the distinct layers ofthe epidermis.Although PKC mediates the induction of keratino-

cyte di�erentiation by calcium as well as TPA, thepattern of marker genes expressed in response to thesetwo stimuli is di�erent, as is their e�ect on keratinocytePKC isozymes (Denning et al., 1995a, 1995b).Calcium-induced keratinocyte di�erentiation in vitrocorrelates with translocation (consistent with enzymeactivation) of PKCa, d and e, but not PKCZ (Denninget al., 1995a); in contrast, TPA activates all of thesePKC isozymes (Denning et al., 1995b). This raises thepossibility that TPA-induced AP-1 activity requiresstimulation of a PKC isozyme, such as PKCZ, that isnot activated during calcium-induced keratinocytedi�erentiation. Such isozyme-speci®c activation mayin¯uence particular transcriptional regulators, withresultant changes in gene expression, to provide amechanism by which external stimuli could in¯uencethe cellular phenotype. Consistent with this idea, over-expression of PKC isozymes has recently been shownto regulate TPA-responsive promoters in an isozyme-speci®c manner (Reifel-Miller et al., 1996). Theexpression pattern of PKCZ in vivo suggests afunction for this isozyme in the ®nal stages ofepidermal di�erentiation and corni®cation (Koizumiet al., 1993; Osada et al., 1993), aspects of keratinocytedi�erentiation that are also induced by TPA (Lichtiand Yuspa, 1988; Yuspa et al., 1982; Jetten et al.,1989). Whether the di�erences in the regulation of the®laggrin promoter in calcium and TPA treated cellsresults from di�erential activation of PKC isozymesremains to be determined. Nevertheless, our data

clearly indicate that di�erent AP-1 factors areexpressed in keratinocytes in response to these twoagents.

Materials and methods

Cell culture

Primary mouse keratinocytes were prepared and main-tained as described previously (Hennings et al., 1980).Brie¯y, 36106 freshly isolated keratinocytes were platedonto a 60 mm dish for transfection assays, and 56106 cellswere plated per 100 mm dish for the preparation of nuclearextracts for Western blot and gel shift analysis. Cells wereplated in Eagle's Minimal Essential Medium plus 8%chelex-treated fetal bovine serum containing 0.2 mM

calcium, and switched to 0.05 mM calcium medium within16 h. Di�erentiation was induced by the addition of anappropriate amount of 300 mM CaCl2 on day 6 afterplating, and the cells were harvested 24 or 48 h later forthe preparation of nuclear proteins. TPA-treated keratino-cytes were maintained in 0.5% serum for 4 h prior to andduring treatment with 162 nM TPA. Nuclear extracts wereprepared 4 h later (also on day 6 after plating). The 4 htime point was found to be optimal for the induction ofAP-1 DNA binding activity and c-Fos expression. Noincrease in AP-1 DNA binding activity was seen incalcium-treated keratinocytes 6 ± 24 h after raising thecalcium concentration in the medium (Rutberg et al.,1996).

Transfection

Primary keratinocytes were transfected on day 6 afterplating using the Lipofectamine Reagent (GIBCO, GrandIsland, New York). In assays using only reporterconstructs, 4 mg of DNA were transfected per 60 mmdish. When a combination of reporter and expressionplasmids was used, 6 mg of DNA were transfected per60 mm dish in the ratio of 1 : 2 of reporter to expressionvector. In all cases, the e�ects of the AP-1 expressionplasmids were compared to those of the empty expressionvectors (pCMV and/or pSG5) alone. Cells were transfectedin serum-free, 0.05 mM calcium medium. After a 6 htransfection procedure, medium containing 0.05 or0.12 mM calcium, or 162 nM TPA in 0.05 mM calciumwas added, all in the presence of 8% chelex-treated fetalbovine serum. Cells were harvested 24, 48 or 60 h later toobserve the e�ects of the di�erent treatments, and to allowsu�cient time for CAT protein to be produced. Cellstreated with TPA were not analysed beyond 24 h due tothe rapid detachment of cells from the dishes.

CAT assays

CAT assays were performed according to the method ofNeumann et al. (1987). Three to ®ve micrograms ofcellular extract were used in each reaction, and the levelsof CAT activity were evaluated after a 1 h incubation at378C. Results of the AP-1-CAT reporter assays areexpressed relative to the activity of the parent pBLCAT2plasmid to ensure that the observed activity was due onlyto the AP-1 DNA binding sequences on the vector (asdescribed in Angel et al., 1987). These experiments werereplicated at least three times each. The experiments withthe pro®laggrin and spr 1 constructs were also replicatedat least three times, but in these cases, two separatepreparations of plasmid DNA were used to minimizedi�erences in transfection e�ciency and to control forvariations in the quality of the preparations of plasmidDNA.

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1344

Electrophoretic mobility shift assay (EMSA)

EMSA was performed as described previously (Rutberg etal., 1996) using 3 mg of nuclear extract (Schreiber et al.,1989) per reaction. DNA binding complexes werevisualized using a 32P-labeled probe bearing a single copyof the consensus AP-1 target sequence (5'-TGAGTCA-3').In supershift reactions, speci®c antibodies were added tothe DNA binding reaction 10 min after the addition ofradiolabeled probe. Competitive inhibition of DNAbinding was achieved by adding unlabeled competitorDNA to the reactions 10 min prior to the addition of theradiolabeled probe. The AP-1 and SP-1 oligomers werepurchased from Promega (Madison, Wisconsin).

Western blotting

Western blot analysis was performed as describedpreviously (Rutberg et al., 1996). The reactions presentedare representative of results obtained with nuclear extractsisolated from three separate preparations of primarykeratinocytes.

Plasmids and antibodies

The pBLCAT2 and 56TRE-CAT vectors (Angel et al.,1987) were obtained from M Karin (University ofCalifornia at San Diego, La Jolla, California). Additionalreporters containing AP-1 like sequences were generated bycloning oligomers bearing 56multimers of variations of

the epidermal AP-1 sequences into the pBLCAT2 construct(see Figure 3a). The pro®laggrin reporter constructs havebeen described (Jang et al., 1996). The spr 1 reporterconstructs (Reddy et al., 1996) were a gift from Reen Wu(University of California at Davis). Antibodies against c-Jun, Jun B, Jun D and Fra-2 were obtained from SantaCruz Biotechnology (Santa Cruz, California). c-Fos andFra-1 antibodies were obtained from Rodrigo Bravo(Bristol-Myers Squibb, Princeton, New Jersey), and the c-Fos antibody (P3) used in the supershift reactions waspurchased from Caltag Laboratories (San Francisco,California). The Jun family expression vectors (pSG5c-Jun, -JunB and -JunD) were obtained from Lester Lau(University of Illinois, Chicago, Illinois), the CMV-cFosvector was a gift from Tom Curran (St Jude Children'sResearch Hospital, Memphis, Tennesse), and the CMV-Fra-2 vector was provided by Donna Cohen (TheAustralian University, Canberra, Australia).

AcknowledgementsWe would like to thank Michael Karin and Reen Wu forsharing reporter constructs, Lester Lau, Tom Curran andDonna Cohen for expression constructs, and RodrigoBravo for AP-1 speci®c antibodies. We would also like tothank Peter Steinert (Laboratory of Skin Biology, NationalInstitute of Arthritis and Musculoskeletal and SkinDiseases, Bethesda, Maryland) for his help with ourstudies related to the pro®laggrin promoter.

References

Angel P, Imagawa M, Chiu R, Stein B, Imbra RJ,Rahmsdord, HJ, Jonat C, Herrlich P and Karin M.(1987). Cell, 49, 729 ± 739.

Angel P and Karin M. (1991). Biochim. Biophys. Acta, 1072,129 ± 157.

Au YP, Dobrowolska G, Morris DR and Clowes AW.(1994). Circ. Res., 75, 15 ± 22.

Bernerd F, Magnaldo T, Freedberg IM and Blumenberg M.(1993). Gene Expr., 3, 187 ± 199.

Casatorres J, Navarro JM, Blessing M and Jorcano JL.(1994). J. Biol. Chem., 269, 20489 ± 20496.

Crish JF, Howard JM, Zaim TM, Murthy S and Eckert RL.(1993). Di�erentiation, 53, 191 ± 200.

Curran T and Franza BR. (1988). Cell, 55, 395 ± 397.Denning MF, Dlugosz AA, Williams EK, Szallasi Z,Blumberg PM and Yuspa SH. (1995a). Cell GrowthDi�er., 6, 149 ± 157.

Denning MF, Kazanietz MG, Blumberg PM and Yuspa SH.(1995b). Cell Growth Di�er., 6, 1619 ± 1626.

Devary Y, Gottlieb RA, Lau LF and Karin M. (1991). Mol.Cell Biol., 11, 2804 ± 2811.

DiSepio D, Jones A, Longley MA, Bundman D, RothnagelJA and Roop DR. (1995). J. Biol. Chem., 270, 10792 ±10799.

Dlugosz AA and Yuspa SH. (1993). J. Cell Biol., 120, 217 ±225.

Dlugosz AA and Yuspa SH. (1994). J. Invest. Dermatol., 102,409 ± 414.

Dobrazanski P, Noguchi T, Kovary K, Rizzo CA, Lazo PSand Bravo R. (1991). Mol. Cell Biol., 11, 5470 ± 5478.

Fuchs E. (1990). J. Cell Biol., 111, 2807 ± 2814.Gius D, Cao XM, Rauscher FJ 3d, Cohen DR, Curran T andSukhatme VP. (1990). Mol. Cell Biol., 10, 4243 ± 4255.

Greenberg ME and Zi� EB. (1984). Nature, 311, 433 ± 438.Gruda MC, Kovary K, Metz R and Bravo R. (1994).

Oncogene, 9, 2537 ± 2547.Hennings H, Michael D, Cheng C, Steinert P, Holbrook Kand Yuspa SH. (1980). Cell, 19, 245 ± 254.

Hill CS and Treisman R. (1995). Cell, 80, 199 ± 211.

Hu� CA, Yuspa SH and Rosenthal D. (1993). J. Biol. Chem.,268, 377 ± 384.

Jang SI, Steinert PM and Markova NG. (1996). J. Biol.Chem., 271, 24105 ± 24114. (Abstr.)

Jetten AM, George MA, Pettit GR, Herald CL and RearickJI. (1989). J. Invest. Dermatol., 93, 108 ± 115.

Johnson PF and McKnight SL. (1989). Annu. Rev. Biochem.,58, 799 ± 839.

Kartasova T, Cornelissen BJC, Belt P and van de Putte P.(1987). Nucleic Acids Res., 15, 5945 ± 5962.

Koizumi H, Kohno Y, Osada S, Ohno S, Ohkawara A andKuroki T. (1993). J. Invest. Dermatol., 101, 858 ± 863.

Lamph WW, Wamsley P, Sassone-corsi P and Verma IM.(1988). Nature, 334, 629 ± 631.

Lichti U and Yuspa SH. (1988). Cancer Res., 48, 74 ± 81.Lopez-Bayghen E, Vega A, Cadena A, Granados SE, JaveLF, Gariglio P and Alvarez-Salas LM. (1996). J. Biol.Chem., 271, 512 ± 520.

Lu B, Rothnagel JA, Longley MA, Tsai SY and Roop DR.(1994). J. Biol. Chem., 269, 7443 ± 7449.

Markova NG, Marekov LN, Chipev CC, Gan SQ, Idler WWand Steinert PM. (1993). Mol. Cell Biol., 13, 613 ± 625.

Marvin KW, George MD, Fujimoto W, Saunders NA,Bernacki SH and Jetten AM. (1992). Proc. Natl. Acad. Sci.USA, 89, 11026 ± 11030.

Mehrel T, Hohl D, Rothnagel JA, Longley MA, BundmanD, Cheng C, Lichti U, Bisher ME, Steven AC, SteinertPM, Yupsa SH and Roop DR. (1990). Cell, 61, 1103 ±1112.

Nakabeppu Y and Nathans D. (1991). Cell, 64, 751 ± 759.Navarro JM, Casatorres J and Jorcano JL. (1995). J. Biol.

Chem., 270, 21362 ± 21367.Neumann JR, Morency CA and Russian KO. (1987).

Biotechniques, 5, 444 ± 447.Osada S, Hashimoto Y, Nomura S, Kohno Y, Chida K,Tajima O, Kubo K, Akimoto K, Koizumi H andKitamura Y. (1993). Cell Growth Di�er., 4, 167 ± 175.

Oshima RG, Abrams L and Kulesh D. (1990). Genes Dev., 4,835 ± 848.

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1345

Parkinson EK, Pera MF, Emmerson A and Gorman PA.(1984). Carcinogenesis, 5, 1071 ± 1077.

Polakowska RR and Goldsmith LA. (1991). Physiology,Biochemistry, and Molecular Biology of the Skin: The cellenvelope and transglutaminases. Goldsmith, LA (ed.)Oxford University Press: New York, pp 168 ± 201.

Reddy SPM, Chuu YJ, Lao PN, Donn J, Ann DK andWu R.(1996). J. Biol. Chem., 270, 26451 ± 26459.

Reifel-Miller AE, Conarty DM, Valasek KM, Iversen PW,Burns DJ and Birch KA. (1996). J. Biol. Chem., 271,21666 ± 21671.

Rezzonico R, Loubat A, Lallemand D, Pfarr CM, Far DF,Proudfoot A, Rossi B and Ponzio G. (1995). Oncogene, 11,1069 ± 1078.

Rice RH and Green H. (1979). Cell, 18, 681 ± 694.Roop DR, Huitfeldt H, Kilkenny A and Yupsa SH. (1987).

Di�erentiation, 35, 143 ± 150.Roop DR, Krieg TM, Mehrel T, Cheng CK and Yuspa SH.(1988). Cancer Res., 48, 3245 ± 3252.

Rothnagel JA, Greenhalgh DA, Gagne TA, Longley MAand Roop DR. (1993). J. Invest. Dermatol., 101, 506 ± 513.

Rutberg SE, Saez E, Glick A, Dlugosz AA, Spiegelman BMand Yuspa SH. (1996). Oncogene, 13, 167 ± 176.

Saez E, Rutberg SE, Mueller E, Oppenheim H, Smoluk J,Yuspa SH and Spiegelman BM. (1995). Cell, 82, 721 ± 732.

Schreiber E, Matthias P, Muller MM and Scha�ner W.(1989). Nucleic Acids Res., 17, 6419.

Sonobe MH, Yoshida T, Murakami M, Kameda T and IbaH. (1995). Oncogene, 10, 689 ± 696.

Steinert PM and Marekov LN. (1995). J. Biol. Chem., 270,17702 ± 17711.

Stoler A, Kopan R, Duvic M and Fuchs E. (1988). J. CellBiol., 107, 427 ± 446.

Suzuki T, Okuno H, Yoshida T, Endo T, Nishina H and IbaH. (1991). Nucleic Acids Res., 19, 5537 ± 5542.

Takahashi H and Iizuka H. (1993). J. Invest. Dermatol., 100,10 ± 15.

Toftgard R, Yuspa SH and Roop DR. (1985). Cancer Res.,45, 5845 ± 5850.

Welter JF, Crish JF, Agarwal C and Eckert RL. (1995). J.Biol. Chem., 270, 12614 ± 12622.

Wirth PJ, Yuspa SH, Thorgeirsson SS and Hennings H.(1987). Cancer Res., 47, 2831 ± 2838.

Yamada K, Yamanishi K, Kakizuka A, Kibe Y, Doi H andYasuno H. (1994). Biochem. Mol. Biol. Int., 34, 827 ± 836.

Yen J, Wisdom RM, Tratner I and Verma IM. (1991). Proc.Natl. Acad. Sci. USA, 88, 5077 ± 5081.

Yoshioka K, Deng T, Cavigelli M and Karin M. (1995).Proc. Natl. Acad. Sci. USA, 92, 4972 ± 4976.

Yuspa SH, Ben T, Hennings H and Lichti U. (1982). CancerRes., 42, 2344 ± 2349.

Yuspa SH, Kilkenny AE, Steinert PM and Roop DR. (1989).J. Cell Biol., 109, 1207 ± 1217.

Yuspa SH. (1994). Cancer Res., 54, 1178 ± 1189.

Opposing activities of c-Fos and Fra-2 in keratinocytesSE Rutberg et al

1346