keratinocyte growth inhibition by high-dose epidermal growth factor is mediated by transforming...

ORIGINAL ARTICLE

Keratinocyte Growth Inhibition by High-Dose EpidermalGrowth Factor Is Mediated byTransforming Growth Factor bAutoinduction: A Negative Feedback Mechanism forKeratinocyte Growth

Kenshi Yamasaki,n NobukoToriu,n Yasushi Hanakawa,n Yuji Shirakata,n Koji Sayama,n Atsushi Takayanagi,wMasafumi Ohtsubo,w Shinobu Gamou,z Nobuyoshi Shimizu,w Makiko Fujii,y Kohei Miyazono,yzand Koji HashimotonnDepartment of Dermatology, Ehime University School of Medicine, Ehime, Japan; wDepartment of Molecular Biology, Keio University School ofMedicine,Tokyo, Japan; zDepartment of Environment and Life Sciences, Kyorin University School of Health Sciences,Tokyo, Japan; yDepartment ofBiochemistry,The Cancer Institute of the Japanese Foundation for Cancer Research,Tokyo, Japan; zDepartment of Molecular Pathology, Graduate Schoolof Medicine,The University of Tokyo,Tokyo, Japan

The epidermal growth factor receptor and its ligandsinitiate a major signaling pathway that regulates kera-tinocyte growth in an autocrine manner. It is wellknown that high doses of epidermal growth factor re-ceptor ligands inhibit keratinocyte growth. Recently,signal transducers and activators of transcription 1-de-pendent p21Waf1/Cip1 induction were reported to be in-volved in high-dose epidermal growth factor-dependent cell growth arrest in the A431 squamous cellcarcinoma cell line; however, transfection of dominant-negative signal transducers and activators of transcrip-tion 1 adenovirus vector did not block epidermalgrowth factor-induced growth inhibition in normalhuman keratinocytes. As transforming growth factor bis a potent inhibitor of keratinocyte proliferation, wehypothesized that transforming growth factor b contri-butes to epidermal growth factor-mediated keratino-cyte growth inhibition. Epidermal growth factorconcentrations of 10 ng per ml enhanced transforminggrowth factor b1 mRNA expression from 3 to 6 h post-stimulation. Enzyme-linked immunosorbent assay ana-lysis detected 150 pg per ml of transforming growthfactor b1 in the culture medium of keratinocytes incu-bated with 10 and 100 ng per ml epidermal growth fac-

tor, whereas 0.1 and 1.0 ng per ml epidermal growthfactor slightly enhance transforming growth factor b1production. Epidermal growth factor (100 ng per ml)upregulated luciferase activity of p3TP-lux, which con-tains three tandem transforming growth factor b-Smadsignaling responsive elements, 6-fold compared withunstimulated cells. The epidermal growth factor-de-pendent induction of p3TP-lux luciferase activity wasdisrupted by transfection of the dominant negativeform of transforming growth factor b type I receptoradenovirus vector (AxdnALK5), which suggests thatepidermal growth factor-induced transforming growthfactor b acts in an autocrine manner in keratinocytes.Moreover, transfection of AxdnALK5 completelyblocked the growth inhibition induced by 100 ng perml of epidermal growth factor in normal keratinocytes.These data demonstrate that an autocrine transforminggrowth factor b1-ALK5 pathway is a negative feedbackmechanism for epidermal growth factor-induced nor-mal human keratinocyte growth. Key words: keratino-cyte/epidermal growth factor/transforming growth factor b/activin receptor-like kinase/signal transducers/activators oftranscription. J Invest Dermatol 120:1030 ^1037, 2003

The binding of epidermal growth factor (EGF) familyligands to EGF receptors (EGFR) initiates a majorsignaling pathway that regulates epithelial cellgrowth (Gusterson et al, 1984; Johnson et al, 1993;Alison and Sarraf, 1994; Filipe et al, 1995; Rajagopal

et al, 1995). Some types of breast cancer cells have abundantEGFR, and EGFR plays a crucial part in breast cancer tumorprogression (Battaglia et al, 1988; Bevilacqua et al, 1990; Toi et al,1990). Although the A431 squamous carcinoma cell line has abun-dant EGFR, high doses of EGF inhibit A431 cell growth (Gilland Lazar, 1981; Barnes, 1982).The EGF family, which includes heparin-binding epidermal

growth factor-like growth factor, transforming growth factor(TGF)-a, and amphiregulin, is well known as an auto-crine growth factor for normal human keratinocytes (Co¡eyet al, 1987; Cook et al, 1991; Hashimoto et al, 1994; Shirakataet al, 2000). Similar to A431 cells, low doses of EGF, heparin-binding-EGF, and epiregulin stimulation promote keratinocyte

Reprint requests to: Kenshi Yamasaki, Department of Dermatology,Ehime University School of Medicine, Shigenobu-cho, Onsen-gun,Ehime 791-0295, Japan. Email: [email protected]: EGFR, epidermal growth factor receptor; STAT, signal

transducers and activators of transcription; ALK, activin receptor-likekinase.

Manuscript received July 3, 2002; revised December 3, 2002; accepted forpublication January 21, 2003

0022-202X/03/$15.00 . Copyright r 2003 by The Society for Investigative Dermatology, Inc.

1030

growth, whereas high doses of 10^100 ng per ml inhibit prolif-eration (Hashimoto et al, 1994; Chen et al, 1995; Shirakata et al,2000).EGFR internalization occurs after the binding of EGFR li-

gands (Gusterson et al, 1984; Lehtola et al, 1989; Bevilacqua et al,1990). This downregulation is one of the EGFR signal regulatorymechanisms (Masui et al, 1993; Skarpen et al, 1998); however, themechanism of EGF-dependent growth inhibition has yet to befully assessed in keratinocytes. Recently, some groups reportedthat in A431 cells, the cyclin-dependent kinase (CDK) inhibitorp21 is involved in EGF-dependent cell growth arrest (Jakus andYeudall, 1996; Ohtsubo et al, 1998; Skarpen et al, 1998), and thatEGF-dependent STAT (signal transducers and activators of tran-scription) 1 activation induces p21 (Bromberg et al, 1998; Ohtsuboet al, 2000); however, our preliminary experiment shows thattransfection of dominant-negative STAT1 adenovirus vector didnot block the growth inhibition dependent on high-dose EGFin human primary keratinocytes. These data suggest that anothermolecule is involved in EGF-induced growth inhibition in thesecells.Interferon (IFN)-g and TGF-b are potent inhibitors of kerati-

nocyte proliferation (Nickolo¡ et al, 1984; Matsumoto et al, 1990).In keratinocytes, IFN-g stimulation activates STAT1 and inducestranscription of the CDK inhibitor p21. TGF-b binds to TGF-btype II receptor, which then phosphorylates the TGF-b type I re-ceptor (activin receptor-like kinase 5; ALK5). PhosphorylatedALK5 recruits and activates the signal transduction moleculesSmad2 and Smad3 (Attisano andWrana, 2000; Datto andWang,2000; Massague andWotton, 2000; Miyazono, 2000). TGF-b in-duced G0/G1 phase cell cycle arrest of various cells (Longstreetet al, 1992; Reddy et al, 1994), and TGF-b induces the CDK inhi-bitors p27 and p16 in keratinocytes and inhibits cell proliferation(Hannon and Beach, 1994; Ewen, 1996). Keratinocyte IFN-g pro-duction is species dependent; there are reports that human kerati-nocytes produce IFN-g, whereas murine keratinocytes do not(Howie et al, 1996; Akiba et al, 2001). On the other hand, both hu-man and murine keratinocytes produce TGF-b1 and TGF-b2(Bascom et al, 1989; Glick et al, 1990; Keski-Oja and Koli, 1992).The activator protein-1 site in the TGF-b promoter region is sti-mulated by growth factors through mitogen-activated protein ki-nase activation (Salbert et al, 1993; Rak et al, 1995). The EGFfamily phosphorylates EGFR, and EGFR activates mitogen-acti-vated protein kinases, phosphatidylinositol 3 kinase, STAT, and Gproteins (Nakamura et al, 1996; Grandis et al, 1998; Toyoda et al,1998; Hackel et al, 1999; Liebmann, 2001). Therefore, we hypothe-size the involvement of TGF-b in high-dose EGF-dependentkeratinocyte growth inhibition. This study demonstrates that in-hibiting TGF-b receptor signaling results in the suppression ofhigh-dose EGF-induced growth inhibition in keratinocytes.

MATERIALS AND METHODS

Cell culture Normal human skin obtained from plastic surgery was cutinto 3^5 mm strips and incubated with 250 U per ml of dispase inDulbecco’s modi¢ed Eagle’s medium overnight at 41C. The epidermis wasseparated from the dermis by forceps, and the epidermal sheets were rinsedwith phosphate-bu¡ered saline (^), incubated in a 0.25% trypsin solutionfor 10 min at 371C, and teased with forceps. Epidermal cells were rinsed,collected by centrifugation, and resuspended in MCDB153 mediumsupplemented with insulin (1 mg per ml), hydrocortisone (0.5 mg per ml),ethanolamine (0.1 mM), phosphoethanolamine (0.1 mM), CaCl2 (0.03 mM),and bovine hypothalamic extract (BHE) (50 mg per ml). Second or thirdpassage cells were used for all experiments. All procedure involvinghuman subjects received prior approval from the ethical committee of theEhime University School of Medicine, and all subjects provided writteninformed consent.

Reagents Recombinant IFN-g and EGF were generous gifts fromOhtsuka Pharmaceutical Co. Ltd (Tokyo, Japan). Recombinant TGF-b1,anti-TGF-b1 monoclonal neutralizing antibody, and normal mouse IgGwere obtained from R&D Systems (Minneapolis, MN).

Adenovirus vector construction and infection The cosmid cassettepAxCAw (Miyake et al, 1996), control adenovirus AxCALacZ, and theparent virus Ad5-dLX (Miyake et al, 1996) were all kind gifts from DrIzumi Saito (Tokyo University, Japan). Fragments of Stat1F weresubcloned into the adenovirus cosmid cassette pAxCAw. STAT1F codes adominant-negative mutant, which has the tyrosine phosphorylation sitesubstituted with phenylalanine. Adenovirus containing the CA promoterand STAT1F (AxSTAT1F) was generated by the COS-TPC method(Miyake et al, 1996). The cosmid DNA was mixed with the EcoT22I-digested DNA-terminal protein complex of Ad5-dLX, and used tocotransfect 293 cells. Recombinant viruses were generated by homologousrecombination in 293 cells. Virus stocks were prepared by a standardprocedure (Miyake et al, 1996). Concentrated, puri¢ed virus stocks wereprepared by CsCl gradient, and the virus titer was veri¢ed using aplaque-formation assay.Construction of the adenovirus vector expressing dominant-negative

TGF-b type I receptor, ALK5 (AxdnALK5) has been described previously(Fujii et al, 1999).Normal human keratinocytes were infected with adenovirus vectors at a

multiplicity of infection (m.o.i.) of 5. AxCALacZ was transfected as acontrol vector.

Cell counts and reduction rate calibration Keratinocytes were seededat 4.0�104 cells per well in six-well collagen-coated dishes (Iwaki Glass,Tokyo, Japan) and cultured in 3 ml of MCDB153 medium with BHE.Twenty-four hours later, the medium was replaced with fresh MCDB153without BHE, and adenovirus vectors (m.o.i.¼5) were added. The cellswere incubated for 24 h for infection and gene expression. Then, variousconcentrations of cytokines were added and the cells were incubated for afurther 96 h. For experiments with anti-TGF-b1 monoclonal neutralizingantibody, antibody, or normal mouse IgG (1 ng per ml) was added at thetime of EGF stimulation. Cells were trypsinized, harvested, and the totalcell number was counted using a Coultert Z1 (Coulter,Tokyo, Japan). Theproliferation rates were calibrated by comparing the total number ofstimulated and unstimulated cells. The mean and SD of relative valuesobtained from three independent experiments were plotted on graphs.

3-[4,5-dimethylthiazol-2-thiazolyl]-2,5-diphenyltetrazulium bro-mide (MTT) assay Keratinocytes were seeded at 5.0�103 cells per wellin 96-well collagen-coated dishes (Iwaki Glass) and cultured in 0.1 ml ofMCDB153 medium with BHE. Adenovirus vector and cytokines wereadded as described for the cell counts. Ninety-six hours after stimulation,cell proliferation was quantitated using a MTT-based assay (CellTiter96tNon-Radioactive Cell Proliferation Assay, Promega, Madison, WI). Theoptical density at 570 nm was measured using a plate reader(Immunomini 2100, Nalge Nunc International K.K,Tokyo, Japan) and thevalue obtained was subtracted from the optical density at 630 nm. Thevalues indicate proliferation relative to the respective unstimulated control.The mean and SD obtained from three independent experiments wereplotted on graphs.

Ribonuclease protection assay Total cellular RNA was isolated fromcultured human keratinocytes with Isogens (Nippon Gene, Toyama,Japan).Single-stranded anti-sense riboprobes were prepared by in vitro

transcription of human cDNA fragments using a RiboQuants In VitroTranscription kit (Pharmingen, San Diego, CA) in the presence of[a-32P]uridine triphosphate. An hCK-3 probe set (Pharmingen) was usedas a template for in vitro transcription. Samples of total RNA (10 mg each)were hybridized with 32P-labeled riboprobe and digested with RNaseusing a RiboQuants RPA kit (Pharmingen) according to themanufacturer’s instructions. Hybridization products were separated on a5% polyacrylamide^8 M urea gel. The gel was absorbed on to ¢lter paper,dried, and then exposed to Kodak Biomax MSs ¢lm (Kodak, Japan) at ^701C. Band density was analyzed using the Diversity Databaset (PDI Inc.,New York). The relative quantities of TGF-b1 and TGF-b2 mRNA wereestimated using GAPDH as an internal reference, and normalized againstthe respective relative values at 0 h post-TGF-b2 stimulation. The relativevalues of TGF-b1 and TGF-b2 mRNAwere plotted on graphs.

Enzyme-linked immunosorbent assay Keratinocytes were seeded at5.0�105 cells per dish in 10 cm diameter collagen-coated dishes (IwakiGlass), and cultured in 10 ml of MCDB153 medium with BHE. Themedium was replaced after 48 h with fresh MCDB153 without BHE(10 ml). Various concentrations of EGF were added and the cells werecultured for a further 3 d. Culture media were collected and stored at ^201C for analysis.TGF-b1 and TGF-b2 were measured in the culture media using

QuantikineTM (R&D Systems) according to the manufacturer’s

NEGATIVE FEEDBACKOF EGF-INDUCED KERATINOCYTE PROLIFERATION BY TGF-b 1031VOL. 120, NO. 6 JUNE 2003

instructions. The optical density at 450 nm was determined using a platereader (Immunomini 2100, Nalge Nunc International K.K) and subtractedfrom the optical density at 540 nm.The mean and SD obtained from threeindependent experiments were plotted on graphs.

Luciferase assay Keratinocytes were seeded at 4.0�104 cells per well in12-well collagen-coated dishes (Iwaki Glass) and cultured in 1 ml ofMCDB153 medium with BHE. The medium was replaced 48 h later with1 ml fresh MCDB153 without BHE, and AxdnALK5 or AxCALacZ weretransfected (m.o.i.¼5). Twenty-four hours after adenovirus transfection, 0.5mg of p3TP-Lux (provided by J.L. Wrana), and 10 ng of pRL-CMV(Promega) were transfected using DOTAPt (Roche Diagnostic, Tokyo,Japan) according to the manufacturer’s instructions. p3TP-Lux containsthree tandem repeats of a TGF-b-signal responsive element, which isactivated by TGF-b stimulation (Hayashi et al, 1997). After a 10 htransfection of the reporter gene, the medium was replaced with 1 ml freshMCDB153 without BHE, various concentrations of EGF, or 0.1 ng per mlof TGF-b1, was added. The cells were harvested 72 h later, and luciferaseactivity was measured using a Dual-Luciferaset Reporter Assay System(Promega) according to the manufacturer’s instructions. Relative valuesof p3TP-Lux luciferase activity were estimated using pRL-CMVluciferase activity as an internal reference, and normalized against therespective relative values at nonstimulation. The mean and SD of relativevalues obtained from three independent experiments were plotted ongraphs.

RESULTS

STAT1 is involved in IFN-c signaling, but not in EGF-dependent growth inhibition in keratinocytes STAT1 isinvolved in EGF-dependent growth inhibition in A431squamous carcinoma cells (Bromberg et al, 1998; Ohtsubo et al,2000). We generated adenovirus vector expressing dominant-negative STAT1 (AxSTAT1F), and examined the involvement ofSTAT1 in primary keratinocyte proliferation. IFN-g is wellknown as a potent inhibitor of primary keratinocyte growth(Nickolo¡ et al, 1984), and it phosphorylates and activates STAT1in keratinocytes (Jiang et al, 1994; Aragane et al, 1997; Naik et al,1997). AxSTAT1F transfection suppressed IFN-g-dependentkeratinocyte growth inhibition, as con¢rmed by cell counts(Fig 1b) and MTT assay (Fig 1d). Because the cell count resultsrepresent the increase in cell number and the MTT assay resultsrepresent total cell number, the values from these methods areslightly di¡erent. Nevertheless, the results from both methodsshow a similar pattern, suggesting that STAT1 is involved inIFN-g-dependent keratinocyte growth inhibition, and thatAxSTAT1F works well as a dominant negative form of STAT1;however, AxSTAT1F transfection did not a¡ect high-dose EGF-dependent keratinocyte growth inhibition (Fig 1a,c), suggestingthe involvement of another mechanism or molecule.

Figure1. STAT1 is involved in IFN-c-dependent, but not in high doses of EGF-dependent, growth inhibition of human keratinocytes. Pri-mary keratinocytes were seeded at 4.0�104 cells per well in six-well plates and cultured in MCDB153 medium with BHE. Twenty-four hours later, themedium was replaced with fresh MCDB153 without BHE. Adenovirus vectors (AxSTAT1F and AxCALacZ, m.o.i.¼5) were added and the cells wereincubated for 24 h. Then, the indicated concentrations of EGF (a) and IFN-g (b) were added to the keratinocyte culture medium and the cells were incu-bated for 96 h. Cells were trypsinized, harvested, and counted, and the proliferation rates were calibrated by comparing the total numbers of stimulated andunstimulated cells. For the MTT assay, primary keratinocytes were seeded at 5.0�103 cells per well in 96-well plates. After transfection with adenovirusvectors, the indicated concentrations of EGF (c) and IFN-g (d) were added. Ninety-six hours after stimulation, proliferation was determined by MTTassay.The values indicate proliferation relative to the respective unstimulated control. The bars and error bars represent mean and SD of three independentexperiments. npo0.05, nnpo0.01, nnnpo0.001.

1032 YAMASAKI ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

EGFR stimulation induced TGF-b1 in keratinocytes LikeIFN-g, TGF-b suppresses keratinocyte proliferation, and TGF-bis produced by keratinocytes (Cook et al, 1990; Matsumotoet al, 1990). We hypothesized that TGF-b is involved in EGF-dependent keratinocyte growth inhibition. We examined theinduction of TGF-b mRNA by EGF stimulation (Fig 2). Underour culture conditions, TGF-b1 was dominantly expressedwithout stimulation. Following 3^6 h of stimulation, 10 ng EGFper ml enhanced TGF-b1 mRNA expression approximately 2-fold over baseline (0 h). TGF-b2 mRNAwas also detected after36^48 h, but TGF-b3 mRNA was not detectable. Culturedkeratinocytes typically secrete some TGF-b1, but not TGF-b2, inculture medium without EGF stimulation (Fig 3). Although lowdoses of EGF (0.1 and 1 ng per ml) slightly increased TGF-b1secretion, higher doses of 10 and 100 ng per ml signi¢cantlyincreased TGF-b1 secretion to approximately 150 pg per ml.Enzyme-linked immunosorbent assay analysis detected no TGF-b2 secretion in the culture medium of primary keratinocytes.These results suggest that TGF-b1, not TGF-b2, may regulatekeratinocyte growth.

ALK5 is involved in EGF-dependent keratinocyte growthinhibition Next we examined whether TGF-b1 induction byEGF a¡ects keratinocyte growth by negative feedback. Weexamined EGF-dependent TGF-b signal activation using thep3TP-Lux plasmid, which contains three tandem repeats of theTGF-b-Smad responsive element linked to the luciferase gene(Wrana et al, 1992). After p3TP-Lux transduction variousconcentrations of EGF were added. The cells were cultured for

48 h and luciferase activity was measured. Low doses of EGF(0.1 and 1 ng per ml) did not enhance luciferase activity (Fig 4);however, 10 and 100 ng per ml of EGF signi¢cantly enhanced thetranscription of p3TP-Lux in the same manner as TGF-b1. Thissuggests that EGF induces the TGF-b-Smad signal inkeratinocytes. Furthermore, this EGF-induced luciferase activitywas suppressed by the transfection of adenovirus vectorexpressing the dominant negative form of TGF-b type Ireceptor, ALK5 (AxdnALK5). These data suggest that EGFstimulation induced TGF-b signaling through TGF-b type Ireceptors, and that TGF-b1 autoinduction is involved in EGF-

Figure 2. EGF upregulates TGF-b mRNA. (a) EGF (10 ng per ml) wasadded to keratinocyte culture medium, and total RNAwas extracted at 0,1, 3, 6, 12, 24, 36, and 48 h after stimulation. Total RNA was hybridizedwith RNA probes, and separated on a polyacrylamide gel. Arrows on theleft side of the panel indicate predicted TGF-b1, TGF-b2, and TGF-b3mRNA bands. L32 and GAPDH are indicated as control mRNA. (b) Re-lative values of TGF-b1 and TGF-b2 mRNA were estimated usingGAPDH as an internal reference, and normalized against the respective re-lative value at 0 h post-TGF-b2 stimulation.

Figure 3. EGF enhances the production of TGF-b. Keratinocytes wereseeded at 5.0�105 per dish in 10 cm diameter collagen-coated dishes andcultured in 10 ml of MCDB153 medium with BHE. Forty-eight hours la-ter, the medium was replaced with 10 ml fresh MCDB153 without BHE.Various concentrations of EGF (n¼ 3) were added, and the cells were cul-tured for more than 3 d. Culture media were collected and theTGF-b1 andTGF-b2 concentration were measured by enzyme-linked immunosorbentassay. Error bars indicate 7SD. npo0.05, nnpo0.01.

Figure 4. EGF-dependent TGF-b signal activation. Keratinocyteswere seeded at 4.0�104 per well in 12-well collagen-coated dishes and cul-tured in 1 ml of MCDB153 medium with BHE. Forty-eight hours later,the medium was replaced with 1 ml fresh MCDB153 without BHE, andAxdnALK5 or AxCALacZ were transfected (m.o.i.¼5). Twenty-fourhours after adenovirus transfection, 0.5 mg of p3TP-Lux (Wrana et al,1992) and 10 ng of pRL-CMV were transfected. After a 10 h transfectionof the reporter gene, media were replaced with 1ml fresh MCDB153 with-out BHE.Various concentrations of EGF or 0.1 ng per ml of TGF-b1wereadded to the medium and the cells were incubated for 72 h. Cells wereharvested and luciferase activity was measured. Error bars indicate 7SD.


dependent p3TP-Lux transcription. Speci¢cally, EGF-inducedTGF-b1 worked in an autocrine fashion. Finally, we examinedthe e¡ect of AxdnALK5 on EGF-dependent keratinocytegrowth inhibition. After AxdnALK5 transfection, EGF wasadded to the cells and the increase in cell number wasdetermined. AxdnALK5 transfection signi¢cantly restored the 10and 100 ng per ml EGF-dependent keratinocyte growthinhibition to normal levels (Fig 5a,c). TGF-b1 neutralizingantibody produced the same e¡ect as AxdnALK5 (Fig 5b,d).These data show that autoinduction of the TGF-b1 and ALK5signal is essential for high-dose EGF-dependent keratinocytegrowth inhibition.

DISCUSSION

The regulation of keratinocyte proliferation is self-limiting. At awound site, the defective epidermal lesion is re-epithelialized bythe proliferation and migration of keratinocytes. Owing tocontact inhibition, keratinocyte growth and migration stops

following re-epithelialization (Adams and Watt, 1988; Pittelkowet al, 1993). At the wound edge, EGFR ligands are induced andcontribute to keratinocyte growth as autocrine growth factors.The function and regulation of the EGFR signal have been dis-cussed in wound healing and keratinocyte growth in vitro (Co¡eyet al, 1987; Cook et al, 1991; Hashimoto et al, 1994; Shirakata et al,2000).The EGFR signal has biphasic e¡ects on epidermal cell

growth: low doses of EGF stimulate keratinocyte growth,whereas high doses of EGF suppress keratinocyte proliferation.Several reports have focused on this biphasic e¡ect. It is wellknown that EGFR is internalized after ligand binding, and thismay contribute to the suppression of a continuous EGFR signal(Gusterson et al, 1984; Lehtola et al, 1989; Bevilacqua et al, 1990;Skarpen et al, 1998). Recent reports, however, show that EGFRinternalization modulates signal transduction in cells, and thatsome signals are enhanced by EGFR internalization (Burke et al,2001; Sorkin, 2001). Consequently, the contribution of receptorinternalization to the biphasic e¡ect of the EGF^EGFR signal re-mains controversial.

Figure 5. TGF-b and ALK5 are involved in EGF-dependent growth inhibition of keratinocytes. Keratinocytes were seeded at 4.0�104 cells perwell in six-well collagen-coated dishes and cultured in 3 ml of MCDB153 medium with BHE. Twenty-four hours later, the medium was replaced withfresh MCDB153 without BHE. Adenovirus vectors (a) or anti-TGF-b1 monoclonal neutralizing antibody (1 ng per ml) (b) were added and the cells wereincubated for 24 h for infection and gene expression. The indicated concentrations of EGF were added and the cells were incubated for 96 h. Cells weretrypsinized, harvested, and counted, and the proliferation rates were calibrated by comparing the total numbers of stimulated and unstimulated cells. Forthe MTTassay, primary keratinocytes were seeded at 5.0�103 cells per well in 96-well plates. After transfection with adenovirus vectors (c), or addition ofanti-TGF-b1monoclonal neutralizing antibody (d), the indicated concentrations of EGF were added and the cells were cultured for 96 h. Proliferation wasdetermined by MTT assay, and the values indicate proliferation relative to the respective unstimulated control. The bars and error bars represent the meanand SD of three independent experiments. npo0.05, nnpo0.01, nnnpo0.001.


Some reports suggest that the EGFR signal induces the CDKinhibitor p21 in A431 cells, and that p21 induction is an e¡ector ofEGF-dependent growth arrest in A431 cells (Jakus and Yeudall,1996; Ohtsubo et al, 1998; Skarpen et al, 1998). Furthermore, p21induction by the EGFR signal depends on STAT1 activation(Bromberg et al, 1998; Ohtsubo et al, 2000). In this study, domi-nant-negative STAT1 had no e¡ect on EGF-induced growth inhi-bition in primary keratinocytes, although it suppressed IFN-g-dependent keratinocyte growth inhibition (Fig 1). This indi-cates that STAT1 is not essential for EGF-dependent growthinhibition, but is essential for the IFN-g signal in primarykeratinocytes.In this study, the TGF-b1 signal contributed to the regulation

of EGF-dependent primary keratinocyte proliferation. Althoughprimary keratinocytes produced both TGF-b1 and TGF-b2mRNA (Fig 2),TGF-b1was dominantly secreted in the mediumunder our culture conditions (Fig 3). TGF-b1 suppresses primarykeratinocyte growth (Hannon and Beach, 1994; Ewen, 1996), andEGF enhanced the expression of TGF-b1 and TGF-b2 mRNAand the secretion of TGF-b1. Interestingly, low doses of EGF(0.1 ng per ml), which stimulated keratinocyte growth, did notincrease TGF-b1 secretion; however, 10 and 100 ng per ml ofEGF, which suppressed keratinocyte proliferation, signi¢cantlyincreased TGF-b1 production in culture medium (Fig 3). This in-crease inTGF-b1 secretion from about 50 pg per ml to 150 pg perml might be enough to suppress primary keratinocyte prolifera-tion to some degree. Under our culture conditions, 50 pg per mlof exogenous TGF-b1 produced only faint growth inhibition,whereas 300 pg per ml of TGF-b1 completely suppressed kerati-nocyte proliferation (data not shown). Over 100 pg TGF-b1 perml reduced keratinocyte growth by half. At 50 pg per ml,TGF-b1 reduced keratinocyte growth by about 10%. Therefore, 10 and100 ng per ml of EGF stimulate the production of enough TGF-b1 to suppress keratinocyte growth. Moreover, the expression ofthe dominant negative form of TGF-b type I receptor, ALK5(AxdnALK5), recovered the growth inhibition using high-doseEGF (Fig 5). This suggests the autocrine-induced TGF-b is a ma-jor cause of the self-limitation of keratinocyte growth. Axd-nALK5, however, did not enhance the keratinocyte proliferation.Some other molecules and mechanisms might be involved in thecomplicated self-limitation of keratinocytes.TGF-b is translated as a latent form precursor and secreted in

latent form (Gleizes et al, 1997; Koli et al, 2001). Secreted latentTGF-b is activated extracellularly by proteolytic cleavage. A po-tential mechanism of TGF-b activation includes plasmin (Lyonset al, 1988, 1990), thrombospondin-1 (Crawford et al, 1998; Ribeiroet al, 1999), integrin avb1 and avb6 (Munger et al, 1998, 1999), andmatrix metalloproteinase-9 (Yu and Stamenkovic, 2000). EGFand EGFR signaling enhance the transcription of plasminogenactivator (Jensen and Rodeck, 1993), thrombospondin (Nickolo¡et al, 1988), and matrix metalloproteinase-9 (Zeigler et al, 1999; O-charoenrat et al, 2000) in keratinocytes. These molecules may ac-tivate latent TGF-b1 secreted from keratinocytes, and activatedTGF-b1 act in an autocrine manner.Constant production of low doses of TGF-b1, which is se-

creted by keratinocytes without stimulation, may contribute tothe regulation of keratinocyte autocrine growth. TGF-b familysignals are transduced by Smad (Attisano and Wrana, 2000;Miyazono, 2000), and TGF-b and activin phosphorylate Smad2and Smad3. These Smad activations are regulated by several me-chanisms. The TGF-b family signal induces transcription ofSmad6 and Smad7, which then suppress Smad signal activationas antagonistic regulators. EGF upregulates Smad6 and Smad7in a lung cancer cell line (Afrakhte et al, 1998), and IFN-g activatesSTAT1 and upregulates Smad7 (Ulloa et al, 1999). EGF and ¢bro-blast growth factor phosphorylate mitogen-activated protein ki-nase, and mitogen-activated protein kinase phosphorylates thelinker domain of Smad, which suppresses Smad activity(Kretzschmar et al, 1997). These reports suggest that basal activa-tion of the TGF-b signal may be suppressed by a growth factorsignal. Our data indicate that high doses of EGF induced TGF-

b1 and suppressed keratinocyte growth. During the keratinocytegrowing phase, inhibitory signals from the TGF-b family mightbe suppressed in normal cells; however, excessive EGFR stimula-tion induces suppressive molecules and/or transduces suppressivesignals for keratinocyte proliferation. Consequently, the TGF-bsignal might have a pivotal role in cell growth regulation. TGF-b1 induced by high doses of EGF suppressed unregulated kerati-nocyte growth. Disruption and deregulation of the TGF-b signalresults in oncogenesis (Hartsough and Mulder, 1997; Klee¡ et al,1999; Wake¢eld and Roberts, 2002); mutations in the TGF-b re-ceptor gene are found in colon cancer, and mutations in theSmad4 gene, a component of the TGF-b signaling molecule,cause colon cancer (Brattain et al, 1996; Oft et al, 1998; Orimoet al, 1998; Grady et al, 1999; Markowitz, 2000;Taketo andTakaku,2000; Imai et al, 2001; Saha et al, 2001; Woodford-Richens et al,2001). TGF-b induces the CDK inhibitors p16 and p27, and p16inhibits CDK4 function. Mutations in the p16 and CDK4 genesare found in squamous cell carcinoma, and constitutive activationof CDK4 induces carcinoma in the epidermis (Matsumoto et al,1999; Sou¢r et al, 1999). In addition, overexpression of EGFR in-duces tumor formation in the epidermis (Xian et al, 1997; Xieet al, 1999; Kiguchi et al, 2000).Therefore, EGFR signal regulationby negative feedback fromTGF-b may be important for prevent-ing epidermal carcinogenesis.

We thankTerukoTsuda and ErikoTan for their technical assistance and Nobuko Nagaifor her secretarial assistance.This work was supported by a Grant-in-Aid for Scienti¢cResearch from the Ministry of Education, Culture, Sports, Science, andTechnology, ofJapan (to K.Y.).

REFERENCES

Adams JC,Watt FM: An unusual strain of human keratinocytes which do not stratifyor undergo terminal di¡erentiation in culture. J Cell Biol 107:1927^1938, 1988

Afrakhte M, Moren A, Jossan S, et al: Induction of inhibitory Smad6 and Smad7mRNA by TGF-beta family members. Biochem Biophys Res Commun 249:505^511, 1998

Akiba H, Ducluzeau MT, Nicolas JF: Interferon-gamma production in skin duringcontact hypersensitivity. No contribution from keratinocytes. J Invest Dermatol117:163, 2001

Alison MR, Sarraf CE: The role of growth factors in gastrointestinal cell prolifera-tion. Cell Biol Int 18:1^10, 1994

AraganeY, Kulms D, Luger TA, Schwarz T: Down-regulation of interferon gamma-activated STAT1 by UV light. Proc Natl Acad Sci USA 94:11490^11495, 1997

Attisano L, Wrana JL: Smads as transcriptional co-modulators. Curr Opin Cell Biol12:235^243, 2000

Barnes DW: Epidermal growth factor inhibits growth of A431 human epidermoidcarcinoma in serum-free cell culture. J Cell Biol 93:1^4, 1982

Bascom CC,Wolfshohl JR, Co¡ey RJ Jr, et al: Complex regulation of transforminggrowth factor beta 1, beta 2, and beta 3 mRNA expression in mouse ¢bro-blasts and keratinocytes by transforming growth factors beta 1 and beta 2.Mol Cell Biol 9:5508^5515, 1989

Battaglia F, Polizzi G, Scambia G, et al: Receptors for epidermal growth factor andsteroid hormones in human breast cancer. Oncology 45:424^427, 1988

Bevilacqua P, Gasparini G, Dal Fior S, Corradi G: Immunocytochemical determina-tion of epidermal growth factor receptor with monoclonal EGFR1 antibodyin primary breast cancer patients. Oncology 47:313^317, 1990

Brattain MG, Markowitz SD,Willson JK: The type II transforming growth factor-beta receptor as a tumor-suppressor gene. Curr Opin Oncol 8:49^53, 1996

Bromberg JF, Fan Z, Brown C, Mendelsohn J, Darnell JE Jr: Epidermal growthfactor-induced growth inhibition requires Stat1 activation. Cell Growth Di¡er9:505^512, 1998

Burke P, Schooler K,Wiley HS: Regulation of epidermal growth factor receptor sig-naling by endocytosis and intracellular tra⁄cking. Mol Biol Cell 12:1897^1910,2001

Chen TC, Persons K, LiuWW, Chen ML, Holick MF: The antiproliferative and dif-ferentiative activities of 1,25-dihydroxyvitamin D3 are potentiated by epider-mal growth factor and attenuated by insulin in cultured human keratinocytes.J Invest Dermatol 104:113^117, 1995

Co¡ey RJJ, Derynck R, Wilcox JN, Bringman TS, Goustin AS, Moses HL,Pittelkow MR: Production and autoinduction of transforming growth fac-tor-a in human keratinocytes. Nature 328:817^820, 1987


Cook PW, Co¡ey RJ Jr, Magun BE, Pittelkow MR, Shipley GD: Expression andregulation of mRNA coding for acidic and basic ¢broblast growth factor andtransforming growth factor alpha in cells derived from human skin. MolEndocrinol 4:1377^1385, 1990

Cook PW, Mattox PA, KeebleWW, et al: A heparin sulfate-regulated human kerati-nocytes autocrine factor is similar or identical to amphiregulin. Mol Cell Biol11:2547^2557, 1991

Crawford SE, Stellmach V, Murphy-Ullrich JE, et al: Thrombospondin-1 is a majoractivator of TGF-beta1 in vivo. Cell 93:1159^1170, 1998

Datto M,Wang XF: The Smads: transcriptional regulation and mouse models. Cyto-kine Growth Factor Rev 11:37^48, 2000

Ewen ME: p53-dependent repression of cdk4 synthesis in transforming growthfactor-beta-induced G1 cell cycle arrest. J Lab Clin Med 128:355^360, 1996

Filipe MI, Osborn M, Linehan J, Sanidas E, Brito MJ, Jankowski J: Expression oftransforming growth factor alpha, epidermal growth factor receptor and epi-dermal growth factor in precursor lesions to gastric carcinoma. Br J Cancer71:30^36, 1995

Fujii M, Takeda K, Imamura T, et al: Roles of bone morphogenetic protein type Ireceptors and Smad proteins in osteoblast and chondroblast di¡erentiation.Mol Biol Cell 10:3801^3813, 1999

Gill GN, Lazar CS: Increased phosphotyrosine content and inhibition of prolifera-tion in EGF-treated A431 cells. Nature 293:305^307, 1981

Gleizes PE, Munger JS, Nunes I, Harpel JG, Mazzieri R, Noguera I, Rifkin DB:TGF-beta latency: biological signi¢cance and mechanisms of activation. StemCells 15:190^197, 1997

Glick AB, Danielpour D, Morgan D, Sporn MB,Yuspa SH: Induction and autocrinereceptor binding of transforming growth factor-beta 2 during terminal di¡er-entiation of primary mouse keratinocytes. Mol Endocrinol 4:46^52, 1990

GradyWM, Myero¡ LL, Swinler SE, et al: Mutational inactivation of transforminggrowth factor beta receptor type II in microsatellite stable colon cancers. CancerRes 59:320^324, 1999

Grandis JR, Drenning SD, ChakrabortyA, Zhou MY, Zeng Q, Pitt AS,Tweardy DJ:Requirement of Stat3 but not Stat1 activation for epidermal growth factorreceptor-mediated cell growth In vitro. J Clin Invest 102:1385^1392, 1998

Gusterson B, Cowley G, Smith JA, Ozanne B: Cellular localisation of humanepidermal growth factor receptor. Cell Biol Int Rep 8:649^658, 1984

Hackel PO, Zwick E, Prenzel N, Ullrich A: Epidermal growth factor receptors:critical mediators of multiple receptor pathways. Curr Opin Cell Biol 11:184^189, 1999

Hannon GJ, Beach D: p15INK4B is a potential e¡ector of TGF-beta-induced cellcycle arrest. Nature 371:257^261, 1994

Hartsough MT, Mulder KM:Transforming growth factor-beta signaling in epithelialcells. PharmacolTher 75:21^41, 1997

Hashimoto K, Higashiyama S, Asada H, et al: Heparin-binding epidermal growthfactor-like growth factor is an autocrine growth factor of human keratino-cytes. J Biol Chem 269:20060^20066, 1994

Hayashi H, Abdollah S, QiuY, et al:The MAD-related protein Smad7 associates withthe TGF-beta receptor and functions as an antagonist of TGF-beta signaling.Cell 89:1165^1173, 1997

Howie SE, Aldridge RD, McVittie E, Forsey RJ, Sands C, Hunter JA: Epidermalkeratinocyte production of interferon-gamma immunoreactive protein andmRNA is an early event in allergic contact dermatitis. J Invest Dermatol106:1218^1223, 1996

Imai Y, Kurokawa M, Izutsu K, et al: Mutations of the Smad4 gene in acute myelo-geneous leukemia and their functional implications in leukemogenesis. Onco-gene 20:88^96, 2001

Jakus J, Yeudall WA: Growth inhibitory concentrations of EGF induce p21 (WAF1/Cip1) and alter cell cycle control in squamous carcinoma cells. Oncogene12:2369^2376, 1996

Jensen PJ, Rodeck U: Autocrine/paracrine regulation of keratinocyte urokinase plas-minogen activator through the TGF-alpha/EGF receptor. J Cell Physiol 155:333^339, 1993

Jiang CK, Flanagan S, Ohtsuki M, Shuai K, Freedberg IM, Blumenberg M: Disease-activated transcription factor. allergic reactions in human skin cause nucleartranslocation of STAT-91 and induce synthesis of keratin K17. Mol Cell Biol14:4759^4769, 1994

Johnson GR, Kannan B, Shoyab M, Stromberg K: Amphiregulin induces tyrosinephosphorylation of the epidermal growth factor receptor and p185erbB2. Evi-dence that amphiregulin acts exclusively through the epidermal growth factorreceptor at the surface of human epithelial cells. J Biol Chem 268:2924^2931,1993

Keski-Oja J, Koli K: Enhanced production of plasminogen activator activity in hu-man and murine keratinocytes by transforming growth factor-beta 1. J InvestDermatol 99:193^200, 1992

Kiguchi K, Bol D, Carbajal S, et al: Constitutive expression of erbB2 in epidermis oftransgenic mice results in epidermal hyperproliferation and spontaneous skintumor development. Oncogene 19:4243^4254, 2000

Klee¡ J, Ishiwata T, Maruyama H, et al: The TGF-beta signaling inhibitor Smad7 en-hances tumorigenicity in pancreatic cancer. Oncogene 18:5363^5372, 1999

Koli K, Saharinen J, Hyytiainen M, Penttinen C, Keski-Oja J: Latency, activation,and binding proteins of TGF-beta. Microsc ResTech 52:354^362, 2001

Kretzschmar M, Doody J, Massague J: Opposing BMP and EGF signalling pathwaysconverge on the TGF-beta family mediator Smad1. Nature 389:618^622, 1997

Lehtola L, Lehvaslaiho H, Sistonen L, Beguinot L, Alitalo K: Receptor downregula-tion and DNA synthesis are modulated by EGF and TPA in cells expressing anEGFR/neu chimera. Growth Factors 1:323^334, 1989

Liebmann C: Regulation of MAP kinase activity by peptide receptor signallingpathway: paradigms of multiplicity. Cell Signal 13:777^785, 2001

Longstreet M, Miller B, Howe PH: Loss of transforming growth factor beta 1 (TGF-beta 1)-induced growth arrest and p34cdc2 regulation in ras-transfected epithe-lial cells. Oncogene 7:1549^1556, 1992

Lyons RM, Keski-Oja J, Moses HL: Proteolytic activation of latent transforminggrowth factor-beta from ¢broblast-conditioned medium. J Cell Biol 106:1659^1665, 1988

Lyons RM, Gentry LE, Purchio AF, Moses HL: Mechanism of activation of latentrecombinant transforming growth factor beta 1 by plasmin. J Cell Biol110:1361^1367, 1990

Markowitz S: TGF-beta receptors and DNA repair genes, coupled targets in apathway of human colon carcinogenesis. Biochim Biophys Acta 1470:M13^M20,2000

Massague J,Wotton D: Transcriptional control by the TGF-beta/Smad signaling sys-tem. EMBO J 19:1745^1754, 2000

Masui H, Castro L, Mendelsohn J: Consumption of EGF byA431 cells: evidence forreceptor recycling. J Cell Biol 120:85^93, 1993

Matsumoto K, Hashimoto K, Hashiro M, Yoshimasa H, Yoshikawa K: Modulationof growth and di¡erentiation in normal human keratinocytes by transforminggrowth factor-beta. J Cell Physiol 145:95^101, 1990

Matsumoto M, Furihata M, Ishikawa T, Ohtsuki Y, Ogoshi S: Comparison of de-regulated expression of cyclin D1 and cyclin E with that of cyclin-dependentkinase 4 (CDK4) and CDK2 in human oesophageal squamous cell carcinoma.Br J Cancer 80:256^261, 1999

Miyake S, Makimura M, Kanegae Y, et al: E⁄cient generation of recombinant ade-noviruses using adenovirus DNA-terminal protein complex and a cosmidbearing the full-length virus genome. Proc Natl Acad Sci USA 93:1320^1324,1996

Miyazono K: TGF-beta signaling by Smad proteins. Cytokine Growth Factor Rev 11:15^22, 2000

Munger JS, Harpel JG, Giancotti FG, Rifkin DB: Interactions between growth fac-tors and integrins: latent forms of transforming growth factor-beta are ligandsfor the integrin alphavbeta1. Mol Biol Cell 9:2627^2638, 1998

Munger JS, Huang X, Kawakatsu H, et al: The integrin alpha v beta 6 binds andactivates latent TGF-beta 1: a mechanism for regulating pulmonary in£amma-tion and ¢brosis. Cell 96:319^328, 1999

Naik SM, Shibagaki N, Li LJ, Quinlan KL, Paxton LL, Caughman SW: Interferongamma-dependent induction of human intercellular adhesion molecule-1 geneexpression involves activation of a distinct STAT protein complex. J Biol Chem272:1283^1290, 1997

Nakamura N, Chin H, Miyasaka N, Miura O: An epidermal growth factor receptor/Jak2 tyrosine kinase domain chimera induces tyrosine phosphorylation ofStat5 and transduces a growth signal in hematopoietic cells. J Biol Chem271:19483^19488, 1996

Nickolo¡ BJ, BashamTY, Merigan TC, MorhennVB: Antiproliferative e¡ects of re-combinant alpha- and gamma-interferons on cultured human keratinocytes.Lab Invest 51:697^701, 1984

Nickolo¡ BJ, Mitra RS, Riser BL, Dixit VM,Varani J: Modulation of keratinocytemotility. Correlation with production of extracellular matrix molecules in re-sponse to growth promoting and antiproliferative factors. AmJ Pathol 132:543^551, 1988

O-charoenrat P, Modjtahedi H, Rhys-Evans P, Court WJ, Box GM, Eccles SA:Epidermal growth factor-like ligands di¡erentially up-regulate matrixmetalloproteinase 9 in head and neck squamous carcinoma cells. Cancer Res60:1121^1128, 2000

Oft M, Heider KH, Beug H: TGF-beta signaling is necessary for carcinoma cell in-vasiveness and metastasis. Curr Biol 8:1243^1252, 1998

Ohtsubo M, Gamou S, Shimizu N: Anti-sense oligonucleotide of WAF1 geneprevents EGF-induced cell-cycle arrest in A431 cells. Oncogene 16:797^802, 1998

Ohtsubo M,Takayanagi A, Gamou S, Shimizu N: Interruption of NFkappaB-STAT1signaling mediates EGF-induced cell-cycle arrest. J Cell Physiol 184:131^137,2000

Orimo H, Ikejima M, Nakajima E, Emi M, Shimada T: A novel missense mutationand frameshift mutations in the type II receptor of transforming growth fac-tor-beta gene in sporadic colon cancer with microsatellite instability. Mutat Res382:115^120, 1998

Pittelkow MR, Cook PW, Shipley GD, Derynck R, Co¡ey RJ Jr: Autonomousgrowth of human keratinocytes requires epidermal growth factor receptor oc-cupancy. Cell Growth Di¡er 4:513^521, 1993

Rajagopal S, Huang S, Moskal TL, Lee BN, el-Naggar AK, Chakrabarty S: Epider-mal growth factor expression in human colon and colon carcinomas: anti-sense epidermal growth factor receptor RNA down-regulates the proliferationof human colon cancer cells. Int J Cancer 62:661^667, 1995

Rak J, Filmus J, Finkenzeller G, Grugel S, Marme D, Kerbel RS: Oncogenes as in-ducers of tumor angiogenesis. Cancer Metastasis Rev 14:263^277, 1995

Reddy KB, Hocevar BA, Howe PH: Inhibition of G1 phase cyclin dependent kinasesby transforming growth factor beta 1. J Cell Biochem 56:418^425, 1994

Ribeiro SM, Poczatek M, Schultz-Cherry S,Villain M, Murphy-Ullrich JE: The ac-tivation sequence of thrombospondin-1 interacts with the latency-associated


peptide to regulate activation of latent transforming growth factor-beta. J BiolChem 274:13586^13593, 1999

Saha D, Datta PK, Beauchamp RD: Oncogenic ras represses transforming growthfactor-beta/Smad signaling by degrading tumor suppressor Smad4. J Biol Chem276:29531^29537, 2001

Salbert G, Fanjul A, Piedra¢ta FJ, Lu XP, Kim SJ, Tran P, Pfahl M: Retinoic acidreceptors and retinoid X receptor-alpha down-regulate the transforminggrowth factor-beta 1 promoter by antagonizing AP-1 activity. Mol Endocrinol7:1347^1356, 1993

ShirakataY, Komurasaki T, Toyoda H, et al: Epiregulin, a novel member of the epi-dermal growth factor family, is an autocrine growth factor in normal humankeratinocytes. J Biol Chem 275:5748^5753, 2000

Skarpen E, Johannessen LE, Bjerk K, et al: Endocytosed epidermal growth factor(EGF) receptors contribute to the EGF-mediated growth arrest in A431 cellsby inducing a sustained increase in p21/CIP1. Exp Cell Res 243:161^172, 1998

Sorkin A: Internalization of the epidermal growth factor receptor: role in signalling.Biochem SocTrans 29:480^484, 2001

Sou¢r N, Moles JP,Vilmer C, et al: P16 UV mutations in human skin epithelial tu-mors. Oncogene 18:5477^5481, 1999

Yu Q, Stamenkovic I: Cell surface-localized matrix metalloproteinase-9 proteolyti-cally activates TGF-beta and promotes tumor invasion and angiogenesis. GenesDev 14:163^176, 2000

Taketo MM, Takaku K: Gastrointestinal tumorigenesis in Smad4 (Dpc4) mutantmice. Hum Cell 13:85^95, 2000

Toi M, Nakamura T, Mukaida H, et al: Relationship between epidermal growthfactor receptor status and various prognostic factors in human breast cancer.Cancer 65:1980^1984, 1990

Toyoda M, Gotoh N, Handa H, Shibuya M: Involvement of MAP kinase-indepen-dent protein kinase C signaling pathway in the EGF-induced p21 (WAF1/Cip1) expression and growth inhibition of A431 cells. Biochem Biophys ResCommun 250:430^435, 1998

Ulloa L, Doody J, Massague J: Inhibition of transforming growth factor-beta/SMAD signalling by the interferon-gamma/STAT pathway. Nature 397:710^713, 1999

Wake¢eld LM, Roberts AB:TGF-beta signaling. positive and negative e¡ects on tu-morigenesis. Curr Opin Genet Dev 12:22^29, 2002

Woodford-Richens KL, Rowan AJ, Gorman P, et al: SMAD4 mutations in colorectalcancer probably occur before chromosomal instability, but after divergence ofthe microsatellite instability pathway. Proc Natl Acad Sci USA 98:9719^9723,2001

Wrana JL, Attisano L, Carcamo J, et al: TGF-beta signals through a heteromericprotein kinase receptor complex. Cell 71:1003^1014, 1992

Xian W, Rosenberg MP, DiGiovanni J: Activation of erbB2 and c-src in phorbolester-treated mouse epidermis: possible role in mouse skin tumor promotion.Oncogene 14:1435^1444, 1997

Xie W, Chow LT, Paterson AJ, Chin E, Kudlow JE: Conditional expression of theErbB2 oncogene elicits reversible hyperplasia in strati¢ed epithelia and up-regulation of TGF-alpha expression in transgenic mice. Oncogene 18:3593^3607, 1999

Zeigler ME, Chi Y, Schmidt T,Varani J: Role of ERK and JNK pathways in regulat-ing cell motility and matrix metalloproteinase 9 production in growth factor-stimulated human epidermal keratinocytes. J Cell Physiol 180:271^284, 1999


keratinocyte growth inhibition by high-dose epidermal growth factor is mediated by transforming...

Documents