tumor necrosis factor related apoptosis inducing ligand triggers apoptosis in dividing but not in...

ORIGINAL ARTICLE

Tumor Necrosis Factor Related Apoptosis Inducing LigandTriggers Apoptosis in Dividing but not in Di¡erentiatingHuman Epidermal Keratinocytes

Bastiaan J. H. Jansen, Fred van Ruissen,� Stefanie Cerneus,Wendy Cloin, Mieke Bergers, Piet E. J. van Erp,and Joost SchalkwijkDepartment of Dermatology, University Medical Center Nijmegen, Nijmegen,The Netherlands, and �Department of Neurogenetics, Academic MedicalCenter, Amsterdam,The Netherlands

Using serial analysis of gene expression we have pre-viously identi¢ed the expression of several pro-apopto-tic and anti-apoptotic genes in cultured human primaryepidermal keratinocytes, including tumor necrosis fac-tor related apoptosis inducing ligand (TRAIL). TRAILis a potent inducer of apoptosis in transformed andtumor cell lines, but usually not in other cells. Here wepresent a study on the e¡ect of TRAIL on cultured ker-atinocytes. It is shown that di¡erentiated and undi¡er-entiated keratinocytes undergo apoptosis after additionof TRAIL to the medium as determined by morpholo-gic and biochemical criteria, such as cellular shrinkageand activation of caspases. The sensitivity for TRAILdi¡ers greatly between undi¡erentiated and di¡erentiat-ing keratinocytes, however, with undi¡erentiated cells

being much more susceptible to apoptosis. Commit-ment to terminal di¡erentiation in the absence ofTRAIL does not in itself induce apoptosis. In contrastto the promyelocytic cell line HL60, internucleosomalDNA fragmentation is not observed in keratinocytes,as assessed by £ow cytometric analysis and agarose gelelectrophoresis. Interestingly, the prime e¡ector ofDNA fragmentation, DNA fragmentation factor of 40kDa (DFF40), is expressed in keratinocytes, yet internu-cleosomal cleavage fails to occur. Our data indicate thatprogrammed cell death during keratinocyte di¡erentia-tion is distinct from receptor-mediated apoptosis inresponse to a death ligand. Key words: caspase/di¡erentia-tion/DNA fragmentation/keratinocyte/skin. J Invest Derma-tol 121:1433 ^1439, 2003

Human epidermis is a strati¢ed, squamous epithe-lium mostly consisting of keratinocytes, andconstitutes the outermost part of the skin. Basalproliferating keratinocytes supply the outer layersof the epidermis with keratinocytes that are set to

undergo a terminal di¡erentiation program. This tightly coordi-nated, complex program ultimately leads to the the formation ofthe stratum corneum, a layer of deceased cells that are covalentlylinked through a network of structural proteins, thus providing adefensive barrier against biologic, chemical, and physical stimuli(Watt, 1989; Fuchs, 1993; Eckert et al, 1997a, 1997b). There is evi-dence that apoptosis and terminal di¡erentiation are two distinctprocesses: terminal di¡erentiation is not associated with mem-brane blebbing and nuclear condensation, and the processrequires days to complete, as opposed to hours for apoptosis(Gandarillas et al, 1999). Interestingly, it has recently been shownthat the typical pro-apoptotic, short prodomain e¡ector caspases-3, -6, and -7 are not activated during terminal di¡erentiation in

the epidermis. Terminal di¡erentiation does involve the skin-spe-ci¢c caspase-14, however, which is exclusively expressed in thestratum granulosum, a layer just below the stratum corneum ofthe epidermis (Eckhart et al, 2000; Lippens et al, 2000).In addition to the normal pathway of programmed cell death

during terminal di¡erentiation, keratinocytes have the armamentto undergo apoptosis upon genomic and toxic insult. High dosesof ultraviolet (UV) irradiation induce apoptosis directly by DNAdamage (Kulms et al, 1999; Kulms and Schwarz, 2000; Pfundtet al, 2001), or activation of the tumor suppressor p53 (Ziegleret al, 1994; Liu et al, 1996), direct triggering of cell death receptors(Aragane et al, 1998; Sheikh et al, 1998), or autocrine release ofdeath ligands (Leverkus et al, 1997). Furthermore, several clinicalconditions are associated with excessive apoptosis in the epider-mis. Autocrine expression of FasL has been described in toxicepidermal necrolysis, a severe adverse, drug-induced reaction inwhich keratinocytes die and epidermis separates from the dermis(Viard et al, 1998), and a role of T-cell-mediated FasL-inducedapoptosis in keratinocytes has been established for various formsof dermatitis (Traidl et al, 2000;Trautmann et al, 2000).We have previously shown, by means of serial analysis of gene

expression (SAGE), that the tumor necrosis factor (TNF) homo-log TNF-related apoptosis inducing ligand (TRAIL) is moder-ately expressed in keratinocytes at the mRNA level (Jansen et al,2001). TRAIL is known to be a potent inducer of apoptosis invarious tumors and cell lines derived thereof, but it has recentlybeen shown that TRAIL can also induce apoptosis through a cas-pase-dependent pathway in hepatocytes, raising concerns about

Reprint requests to: Bastiaan J.H. Jansen, Department of Tumor Immu-nology, Nijmegen Center for Molecular Life Sciences, University MedicalCenter Nijmegen, PO Box 9101, 6500 HB Nijmegen, The Netherlands.Email: [email protected]: DFF, DNA fragmentation factor; SAGE, serial analysis of

gene expression; TRAIL, tumor necrosis factor related apoptosis inducingligand; TRAIL-R,TRAIL receptor.

Manuscript received July 10, 2002; revised July 14, 2003; accepted forpublication August 12, 2003

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

1433

the applicability of TRAIL in cancer therapy (Jo et al, 2000).TRAIL can induce apoptosis in primary keratinocytes, but theapoptotic response is far less intense than in the transformed ker-atinocyte cell line HaCaT (Leverkus et al, 2000), whereas othershave claimed that primary keratinocytes are resistant to TRAIL-induced apoptosis (Kothny-Wilkes et al, 1998), or that sensitivitytoTRAIL requires the addition of the protein synthesis inhibitorcycloheximide (Basile et al, 2001). TRAIL expression was pre-viously established at both the mRNA and protein levels in cul-tured keratinocytes (Kothny-Wilkes et al, 1998). Furthermore,there is evidence that theTRAIL preparation in£uences the sever-ity of the apoptotic response: TRAIL fused to a leucine zippermotif is able to rapidly induce apoptosis in human keratinocytes,whereas nontagged TRAIL does so to a much lesser extent (Qinet al, 2002).Here we have sought to determine the e¡ect of TRAIL on cul-

tured keratinocytes, using a model system that allows di¡erentia-tion in a submerged culture (Van Ruissen et al, 1996; Pfundt et al,2000).We show that TRAIL is able to induce apoptosis in cul-tured keratinocytes, and that the severity of the apoptotic re-sponse is dependent on both concentration of TRAIL andcommitment to di¡erentiation. We were unable to demonstrateinternucleosomal DNA fragmentation, a biochemical hallmarkof apoptosis (Wyllie et al, 1980; Hengartner, 2000), by means ofagarose gel electrophoresis and £ow cytometry. Although quanti-tative RT-PCR indicated that DNA fragmentation factor of40 kDa (DFF40) is expressed in human primary keratinocytes,the mechanism by which DNA is degraded during apoptosisand di¡erentiation remains unknown. Our data suggest that ker-atinocyte cell death as a result of terminal di¡erentiation is di¡er-ent from apoptosis in response to a death ligand.

MATERIALS AND METHODS

Cells and cell culture Primary keratinocytes were obtained as describedelsewhere (Van Ruissen et al, 1996). Cells were grown to con£uence (106

cells) at 371C in 5% CO2 in six-well tissue culture plates in keratinocytegrowth medium (KGM) consisting of KBM (0.15 mM Ca2þ ;BioWhittaker, Verviers, Belgium) supplemented with ethanolamine (0.1mM; Sigma, St Louis, MO), phosphoethanolamine (0.1 mM, Sigma),bovine pituitary extract (0.4% vol/vol; BioWhittaker), epidermal growthfactor (10 ng per mL; Sigma), insulin (5 mg per mL; Sigma),hydrocortisone (0.5 mg per mL; Collaborative Research, Lexington, MA),penicillin (100 U per mL; Life Technologies, Gaithersburg, MD), andstreptomycin (100 mg per mL; Life Technologies). Cells were not switchedor were switched to KGM depleted of growth factors (bovine pituitaryextract, epidermal growth factor, hydrocortisone, and insulin) for 24 hbefore being stimulated for various periods of time with di¡erentconcentrations of TRAIL (Biomol, Plymouth Meeting, PA). Thiscommercial preparation of TRAIL comprises amino acids 114^281 and hasan estimated molecular weight of 23 kDa. It has been expressed in andisolated from Escherichia coli, and was shipped in phosphate-bu¡ered saline(PBS) (pH 7.0) containing 0.1 mM dithiothreitol.HL60 cells were cultured at 371C in 5% CO2 to a density of 107 cells per

mL in RPMI Complete (10% fetal bovine serum (Harlan SeraLab,Crawley Down, UK) in RPMI (ICN Biomedicals, Costa Mesa, CA), 100U per mL penicillin, and 100 mg per mL streptomycin). Fetal bovine serumwas heated for 30 min at 561C in order to avoid e¡ects of components thatmay interfere with apoptosis. HL60 cells were directly stimulated withvarious concentrations of TRAIL in their medium. Protocols for biopsytaking for cell cultures were approved by the local medical ethics committee.

Immunohistochemistry and hematoxylin/eosin staining of culturedkeratinocytes Keratinocytes were cultured on Thermanox coverslips(Nalge Nunc International, Rochester, NY) as indicated. In the case ofimmunohistochemistry, cells were ¢xed in 1:1 vol/vol methanol/acetonefor 10 min at �201C, air dried, and incubated in PBS pH 7 for 15 min atroom temperature. Cells were then preincubated with 20% horse serumfor 15 min at room temperature, followed by incubation with a primarymouse monoclonal antibody against CK10 (ICN Biomed, Costa Mesa,CA), diluted 1:25 in 1% bovine serum albumin in PBS, 1 h at 371C undergentle agitation. Cells were further stained with the Vectastain Elitemouse IgG kit (Vector Laboratories, Burlingame, CA), following the

manufacturer’s recommendations. Cells were counterstained with Mayer’shematoxylin (Sigma) and mounted in glycerol�gelatin (Sigma). Anegative control without primary antibody was included for bothundi¡erentiated and di¡erentiated keratinocytes.In the case of hematoxylin/eosin staining, 3 h after stimulation of

keratinocytes with 250 ng per mL TRAIL, coverslips were collected andwashed in PBS and cells were ¢xed for 10 min in ice-cold methanol/acetone 1:1 vol/vol, incubated at �201C for 10 min, and stored at �201Cuntil further use. Before staining, cells were incubated in 70% ethanol;they were then stained with hematoxylin (Sigma) for 10 min, washed for10 min with tap water, and counterstained with eosin for 90 s. Cells werewashed three times with 100% ethanol. Coverslips were mounted usingPermount (Fisher Scienti¢c, Fairlawn, NJ). Cells were examined under alight microscope at a magni¢cation of 100 � and photographed.

Protein extraction, polyacrylamide gel electrophoresis, and westernblot analysis Cultured keratinocytes were harvested at various timesafter stimulation with TRAIL, depending on the protein underinvestigation, in 150 mL Laemmli bu¡er (Bio-Rad, Hercules, CA), andb-mercaptoethanol was added to a ¢nal concentration of 5% vol/vol.Samples were soni¢cated and boiled for 3 min, and 15 mL was loadedonto 15% polyacrylamide Tris�HCl Ready Gels (Bio-Rad) or 12%BisTris NuPAGE gels (Invitrogen, Breda, The Netherlands) and run for45 min at 175 V in bu¡er supplied by the manufacturer, along withKaleidoscope prestained size standards (Bio-Rad) and BiotinylatedProtein Marker (Cell Signaling, Beverley, MA). After the run, proteinwas blotted onto polyvinylidene £uoride membrane in a Tris/glycine/methanol bu¡er at 400 mA for 30 min. Membranes were washed withTris-bu¡ered saline/Tween-20 (TBS/T) and blocked for 1^2 h in Blotto/0.1% Tween-20 (Bio-Rad). Membranes were incubated overnight at 41Cwith primary mouse antibodies, diluted 1000 � in Blotto, against eitherhuman caspase-3 (monoclonal; R&D Systems, Minneapolis, MN) orhuman caspase-8 (monoclonal; BioSource, Camarillo, CA), eachrecognizing both their inactive substrates and activated caspases, orprimary rabbit antibodies against TRAIL (polyclonal; Santa CruzBiotechnology, Santa Cruz, CA). After washing with TBS/T, membraneswere incubated with secondary antibodies, biotinylated goat antimouse orgoat antirabbit and horseradish peroxidase conjugated goat antibiotin,1000 � diluted in Blotto. Membranes were washed with TBS/T, andthen incubated with LumiGLO (Cell Signaling, Beverly, MA) andperoxidase. Membranes were then exposed to Kodak X-OMAT UV ¢lmand autoradiographs were developed after 1 min, 2 min, and 7.5 min.Autoradiographs showing caspase-3 activation were analyzed digitallywith ImageMaster software, v1.0 (Pharmacia, Uppsala, Sweden).

Isolation of genomic DNA and agarose gel electrophoresis Atvarious time intervals after TRAIL stimulation of keratinocytes, 1.5 mL2 � lysis bu¡er (200 mM Tris�HCl pH 8, 10 mM ethylenediaminetetraacetic acid, 0.4% sodium dodecyl sulfate, 400 mM NaCl, and 400 mgper mL proteinase K) was added to cultured keratinocytes submerged in1.5 mL medium. Samples were incubated for 2 h at 501C, and extractedtwice with 1 volume phenol/chloroform, followed by one extraction withchloroform. DNA was precipitated with 1 volume isopropanol andcentrifuged at 12,000 � g for 45 min. Pellets were washed twice with70% ethanol, air dried, and resuspended in 10 mM Tris�HCl pH 8,1 mM ethylenediamine tetraacetic acid. DNA concentration wasdetermined spectrophotometrically, and 1 mg of DNAwas run on a 1.5%agarose gel in 1 � TAE (Bio-Rad). DNA was visualized with UV light,using ethidium bromide as a chromophore.

Flow cytometry After stimulation with TRAIL for 3 h, cells wereincubated with 0.25% trypsin in PBS for 7^8 min at 371C. Trypsinizationwas stopped by adding 6 vol KGM supplemented with 5% fetal bovineserum (Harlan SeraLab, Crawley Down, UK). Cells were collected bygentle centrifugation (300 � g) for 7^8 min and washed three times withPBS. After the last wash, cells were resuspended in PBS containing 1% wt/vol paraformaldehyde, incubated on ice for 15 min, pelleted by gentlecentrifugation, and washed several times with PBS containing 1% wt/volbovine serum albumin. 50 mL 10 mg per mL RNase A was added to thepellet, along with 40 mg per mL propidium iodide in PBS, and thesample was incubated in the dark for at least 15 min at room temperature,and washed and resupended in PBS. Cells were analyzed on the Epics Elite£ow cytometer equipped with a 40 mWair-cooled argon laser set at 15 mWand tuned at a wavelength of 488 nm. Red £uorescent propidium iodidesignals were detected through a 630 nm long-pass ¢lter. Usually 104 cellswere measured at a £ow rate of 50 cells per s. Data were collected andanalyzed with the WinsList Software Package, v4.0 (Verity SoftwareHouse,Topsham, ME).

1434 JANSEN ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Isolation of total RNA Total RNAwas isolated from cultured primarykeratinocytes and HL60 cells by lysing cells in RNase All (2.1M guanidinethiocyanate (Research Organics, Cleveland, OH), 8.5 mM N-laurylsarcosine (Sigma), 12.5 mM NaAc pH 5.2, 0.35% vol/vol b-mercaptoethanol (Merck, Darmstadt, Germany), and 50% vol/volTris-saturated phenol pH 8.0 (Biosolve, Amsterdam, The Netherlands))and collected using a cell scraper. Total RNAwas extracted by adding 1/10of a volume of chloroform and centrifugation at 12,000 � g at 41C for 15min. The aqueous phase was recovered and total RNAwas precipitated byadding an equal volume of isopropanol, followed by an incubation on icefor 45 min. RNAwas pelleted by centrifugation at 12,000 � g for 15 min,and the pellet was washed with 70% ethanol and dried at roomtemperature. The pellet was then resuspended in 0.1 mL NSE (50 mMNaAc, 0.2% sodium dodecyl sulfate, 2 mM ethylenediamine tetraaceticacid) and 375 mL 100% ethanol was added. For quantitation, 10 mL of thissuspension was used to pellet RNA, and the pellet was resuspended in 1mL H2O. Purity and concentration of total RNA were determinedspectrophotometrically.

Real-time quantitative RT-PCR 2 mg of total RNA, from HL60 cellsand cultured primary keratinocytes, was converted to single-strandedcDNA using the Superscript kit (Life Technologies), following themanufacturer’s guidelines. Quantitative PCR were performed using thePCR SYBR Green Master Mix (Applied Biosystems, Foster City, CA)following the manufacturer’s guidelines. Optimal dilution for quantitativePCR of the cDNAwas determined at 10,000 � . For the PCR to identifyDFF40 the following primers were used: 50 -GAGGACCCGCCGTGGTTTG-AAG-30 (CAD540-forward) and 50 -ATGGAGCCGAGGACCCGCAG-30(CAD709-reverse). As a reference, the expression of porphobilinogendeaminase was measured using commercially available primers. Reactionswere performed in an ABI PRISM 7700 Sequence Detection System(Applied Biosystems), following the manufacturer’s protocol. Relativeexpression levels of DFF40 in two samples derived from HL60 cellsand two samples derived from human primary keratinocytes weredetermined, and reactions per sample were performed in triplicate. Thedi¡erence in expression levels between HL60 cells and primarykeratinocytes was analyzed, following the manufacturer’s guidelines.

RESULTS

Trail is expressed in human keratinocytes and epidermis andinduces apoptosis in cultured, primary human keratino-cytes We have previously constructed SAGE libraries ofresting and TNF-a-stimulated keratinocytes and found thatvarious genes involved in apoptosis and survival were expressedin cultured keratinocytes, irrespective of TNF-a stimulation(Jansen et al, 2001). There is considerable evidence that mRNAlevels may not re£ect the actual expression of a gene at theprotein level (Gygi et al, 1999).We therefore veri¢ed the expres-sion of TRAIL protein in our culture model. Undi¡erentiatedcells and cells that had committed to di¡erentiation for 24 hwere harvested in reducing protein sample bu¡er and weresubjected to western blot analysis, using rabbit polyclonalantibodies raised against human recombinant TRAIL corres-ponding to amino acids 25^281. As shown in Fig 1(A), wecould indeed con¢rm TRAIL protein expression, as itsmonomers of approximately 35 kDa are readily detectable in ourculture model, irrespective of di¡erentiation state. Expression ofTRAIL-R1 and TRAIL-R2 in cultured keratinocytes has beendemonstrated by others (Leverkus et al, 2000). Apparently theconstitutive expression of TRAIL is not su⁄cient to induceapoptosis in the culture system used, which may be due to lowlevels of TRAIL or the action of the antiapoptotic machinery.Here we investigate the response of keratinocytes to humanrecombinant TRAIL, using a previously described culture model(Van Ruissen et al, 1996). Both undi¡erentiated (cultured tocon£uence) and di¡erentiating keratinocytes (cultured tocon£uence, then cultured in medium without growth factors for24 h) were exposed to TRAIL at concentrations of 0 (negativecontrol, Fig 1B, pictures A, D) and 250 ng per mL (Fig 1B,pictures C, F) and stained with hematoxylin/eosin in order tovisualize morphologic changes. Di¡erentiation was investigatedby immunohistochemical staining of undi¡erentiated and

di¡erentiated keratinocytes with a monoclonal antibody againstcytokeratin 10 (CK10), a marker for di¡erentiation. As shown inFig 1B, picture B, undi¡erentiated keratinocytes do not display anycommitment to di¡erentiation as they do not express CK10,whereas expression was visible in cells that had committed todi¡erentiation for 24 h. Interestingly, the absence of CK10expression, and thus the absence of di¡erentiation, coincideswith increased sensitivity to apoptosis induced by TRAIL. Un-di¡erentiated keratinocytes readily show apoptotic, morphologicfeatures like nuclear condensation and cellular shrinkage within3 h of stimulation with 250 ng per mLTRAIL (Fig 1B, picture C;in detail, picture G); virtually all cells had undergone or wereundergoing apoptosis and most of them had detached from theculture dish. Cells that were allowed to di¡erentiate for 24 hwere far less sensitive to TRAIL-induced apoptosis, as shown inFig 1B, picture F, although an occasional cell did display apoptoticfeatures. It should be noted that apoptosis was still incomplete at3 h after stimulation, and that all undi¡erentiated cells haddetached from the culture dish after 16 h (data not shown).Negative controls do not display apoptotic morphology,

Figure1. TRAIL is expressed in the culture model used throughoutthe study, and induces apoptosis in cultured human epidermal ker-atinocytes. Protein samples derived from undi¡erentiated and di¡eren-tiated keratinocytes were subjected to western blot analysis using a rabbitpolyclonal against recombinant human TRAIL, which recognizes the fulllength human TRAIL of approximately 35 kDa. See Results for details.Undi¡erentiated keratinocytes were stimulated directly with 0 (negativecontrol, picture A) and 250 ng per mL TRAIL (picture C; detail, picture G),or ¢rst di¡erentiated for 24 h before stimulation with 0 (negative control,picture D) and 250 ng per mL TRAIL (picture F). Apoptotic cells werestained with hematoxylin to visualize nuclei, and counterstained with eo-sin. Apoptotic cells are characterized by cellular shrinkage and membraneblebbing. Di¡erentiation was checked immunohistochemically by stainingwith a monoclonal antibody against cytokeratin 10, which is absent in un-di¡erentiated cells (picture B) but present in cells that have committed todi¡erentiation (picture E). Neg, negative control; undi¡, undi¡erentiated;di¡, di¡erentiating; aCK10, antibody against cytokeratin 10.

TRAIL INDUCES APOPTOSIS IN KERATINOCYTES 1435VOL. 121, NO. 6 DECEMBER 2003

indicating that endogenously expressed TRAIL is apparently notable to trigger apoptosis, regardless of di¡erentiation (Fig 1B,pictures A, D).

Induction of apoptosis by trail activates executionercaspase-3 As the morphologic analysis (Fig 1B) suggested thatdi¡erentiating cells were less sensitive toTRAIL, we investigatedwhether keratinocyte di¡erentiation in£uences the apoptoticresponse as measured by caspase-3 activation. It appears thatundi¡erentiated keratinocytes are more sensitive to TRAIL-induced apoptosis than those committed to di¡erentiation for24 h (Fig 2). Furthermore, the band representing the 18 kDasubunit of activated caspase-3 is almost absent in di¡erentiatingkeratinocytes, and there is still a considerable amount ofunprocessed, inactive caspase present even at the highestconcentration of TRAIL used. Indeed, digital densitometricanalysis revealed that TRAIL activated caspase-3 to a muchhigher degree in undi¡erentiated cells as the ratio of activatedcaspase (18 and 20 kDa bands) to procaspase was more than 400times higher in these cells at 250 ng per mL TRAIL. At 125 ngper mL TRAIL, this ratio was still 15 times higher inundi¡erentiated keratinocytes (Fig 2, bottom). Taken together,these data indicate that TRAIL-induced apoptosis is incompletein di¡erentiating keratinocytes, and that undi¡erentiatedkeratinocytes are extremely sensitive to TRAIL-inducedapoptosis.

Apoptosis is induced through proximal caspase-8 It hasrecently been shown that TRAIL-induced apoptosis throughTRAIL-R2 requires the recruitment of Fas-associated deathdomain containing protein (FADD) and caspase-8 to the death-inducing signaling complex (DISC), which is associatedintracellularly with TRAIL-R2 (Bodmer et al, 2000; Kischkelet al, 2000; Sprick et al, 2000). Upon binding of TRAIL toTRAIL-R2, it is believed that conformational changes in theDISC induce proteolytic activation of caspase-8 (Muzio et al,1998; Salvesen and Dixit, 1999; Hengartner, 2000). In order toverify that apoptosis is indeed mediated through a receptor andDISC, keratinocytes were grown to con£uence, and eitherdirectly stimulated with TRAIL or di¡erentiated for 24 h beforestimulation. Cells were harvested 90 min after stimulation(Leverkus et al, 2000) in reducing protein sample bu¡er andsubjected to western blot analysis. Caspase-8 was identi¢ed witha mouse monoclonal antibody that recognizes inactiveprocaspase-8 of 55 kDa, the intermediate cleavage product of41 kDa, and the p18 subunit of activated caspase-8. As can beseen in Fig 3, caspase-8 activation is readily induced byTRAIL inundi¡erentiated keratinocytes at concentrations as low as 63 ngper mL TRAIL, whereas only the highest concentration of250 ng per mL TRAIL was able to activate caspase-8 in cellsthat had di¡erentiated for 24 h (Fig 3), albeit at an almostundetectable level. TRAIL-induced apoptosis in keratinocytes is

therefore mediated through one of its receptors TRAIL-R1 orTRAIL-R2, and this activation is dependent on both TRAILconcentration and the di¡erentiation status of the keratinocyteculture.

DNA fragmentation is not induced during apoptosis inkeratinocytes, but its prime e¡ector DFF40 is expressed Inaddition to the morphologic and biochemical evidence forTRAIL-induced apoptosis, another hallmark, internucleosomalDNA fragmentation (Wyllie et al, 1980; Hengartner, 2000), wasalso investigated. Activation of DFF40 during apoptosis leads toan internucleosomal digestion pattern of genomic DNA thatappears as fragments of (multiples of) 180 bp on an agarose gel(Liu et al, 1997; Sakahira et al, 1998). When analyzed by agarosegel electrophoresis, however, typical DNA fragmentation couldnot be detected in apoptotic keratinocytes (Fig 4D, inset).Rather, a smear instead of distinct fragments of degradedgenomic DNA appeared. As a control, HL60 cells were alsoincluded in the experiment, and these cells readily show DNAfragmentation 3 h after TRAIL stimulation (Fig 4B, inset). Flowcytometric analysis of apoptotic keratinocytes, either bypropidium iodide staining (Fig 4D) or TUNEL (not shown),did not reveal typical DNA fragmentation, whereas once againHL60 cells readily underwent fragmentation (Fig 4B). To oursurprise, the G1 peak did not tend to disappear in keratinocytesthat had been treated with 250 ng per mL of TRAIL,suggesting that the apparent total cellular content of DNAhardly changes in apoptotic keratinocytes. Nonstimulated cellsserved as negative controls and did not show fragmentation ineither HL60 cells (Fig 4A) or keratinocytes (Fig 4C). As DFF40is the prime e¡ector of DNA fragmentation in apoptosis, weinvestigated whether the expression levels of DFF40 inkeratinocytes could explain the lack of DNA fragmentation inthese cells. Real-time quantitative PCR only revealed a smalldi¡erence in DFF40 expression between HL60 and keratinocytes( � 2.5-fold lower in the latter; Fig 5), suggesting that other asyet unidenti¢ed mechanisms are responsible for DNA degra-dation, which may or may not involve DFF40.

Figure 2. Human caspase-3 is activated byTRAIL to a much high-er degree in undi¡erentiated cultured human keratinocytes. Undif-ferentiated (right side) and di¡erentiated (for 24 h, left side) cells wereincubated with 0 (negative control), 15, 31, 63, 125, or 250 ng per mLTRAIL, and harvested 3 h later in Laemmli reducing sample bu¡er. Bot-tom: The ratio of activated (18 kDa and 20 kDa products) caspase-3 to pro-caspase as determined by digital densitometric analysis.

Figure 3. Human caspase-8 is rapidly activated byTRAIL in undif-ferentiated cultured human keratinocytes. Undi¡erentiated (left side)and di¡erentiated (for 24 h, right side) cells were incubated with 0 (negativecontrol), 63, or 250 ng per mL TRAIL, and harvested 90 min later inLaemmli reducing sample bu¡er. Samples were run on a 12% polyacryla-mide gel, and blotted onto a polyvinylidene £uoride membrane. Caspase-8was detected with a mouse monoclonal antibody against the active 18 kDasubunit of caspase-8, which also recognizes the active procaspase-8 and anintermediate of 41 kDa. The 43 kDa band seen in di¡erentiated keratino-cytes probably also represents an intermediate.


DISCUSSION

We have previously shown, by means of SAGE, that culturedkeratinocytes express various mediators of apoptosis, irrespectiveof stimulation with TNF-a. TRAIL, a TNF family member, wasfound to be expressed at moderate levels and a role for this cyto-kine in keratinocyte homeostasis and di¡erentiation, if any, hasyet to be established. Its expression has been con¢rmed at themRNA (data not shown) and protein levels by means of RT-PCR and western analysis, respectively, con¢rming results fromothers (Kothny-Wilkes et al, 1998). In this study, we show bothmorphologically and biochemically that the induction of apopto-sis byTRAIL is dependent onTRAIL concentration and the dif-ferentiation status of keratinocytes, using a previously describedsubmerged culture system that allows terminal di¡erentiation(Van Ruissen et al, 1996). Keratinocytes show nuclear condensa-tion and cellular shrinkage within 3 h after addition of TRAILto the medium, albeit to a much lesser extent when cells havecommitted to di¡erentiation. It is shown that both caspase-8and caspase-3 are activated in keratinocytes upon exposure toTRAIL, indicating that TRAIL acts through one of its two deathreceptors,TRAIL-R1 or TRAIL-R2, and the DISC, and that thee¡ects seen in culture are not due to necrosis. InternucleosomalDNA fragmentation could not be detected in keratinocytes,either by agarose gel electrophoresis or by £ow cytometric analy-

sis. By means of quantitative RT-PCR, however, it is shown thatDFF40, the prime e¡ector of this phenomenon, is expressed inhuman keratinocytes.The culture model that we used provides the opportunity to

compare the responses of undi¡erentiated and di¡erentiating ker-atinocytes. The fact that there is still some apoptosis in di¡eren-tiating cultures probably re£ects the proportion of cells that arestill proliferating after being switched to medium withoutgrowth factors. Although studies have been performed in the pastwhere the e¡ect of TRAIL on primary cultured keratinocyteswas found to be mild or absent, it should be noted that these stu-dies did not take into account the state of di¡erentiation of kera-tinocytes. Leverkus et al (2000) have shown that resistance toapoptosis is mediated through the apotosis inhibitor FADD-likeinhibitory protein (cFLIP), but the same study does not rule outthat cFLIP expression is lower or absent in undi¡erentiated pri-mary keratinocytes. Our SAGE studies indicate that cFLIP is ex-pressed in di¡erentiated keratinocytes albeit at low levels. Inanother study, Kothny-Wilkes et al (1998) claim that primary ker-atinocytes are insensitive to TRAIL, which we found not to bethe case. The fact that undi¡erentiated keratinocytes are far moresensitive to TRAIL and respond by undergoing massive apopto-sis, in contrast to di¡erentiating keratinocytes, suggests that cul-ture conditions have an enormous impact on the degree ofapoptosis in these cells. It has recently become apparent that cul-tured primary keratinocytes are relatively resistant to UV-inducedapoptosis when they have been grown to con£uence or have un-dergone senescence, whereas the transformed keratinocyte cellline HaCaT is much more sensitive (Chaturvedi et al, 1999). Itappears that the susceptibility of HaCaT can at least in part beascribed to the absence of the tumor suppressor p16 (due to pro-moter methylation) and dysfunctional nuclear factor kB (NF-kB), which in its functional state is implicated in cell survivalpathways. Interestingly, the same study showed that subcon£uent,undi¡erentiated keratinocytes are much more prone to apoptosis,but that death resistance does not require early di¡erentiation. Ithas also been shown that NF-kB is responsible for the apoptosis-resistant phenotype of keratinocytes (Qin et al, 1999). TRAIL-in-duced apoptosis in keratinocytes apparently does not depend inthe same way on the state of di¡erentiation as UV-induced apop-tosis, which may also re£ect di¡erent death pathways. First, con-£uent, undi¡erentiated keratinocytes still underwent massiveapoptosis, and second, only a short period of di¡erentiation of 3h already caused considerable resistance toTRAIL-induced death(data not shown). This suggests that the presence of a fully di¡er-entiated phenotype is not required to confer resistance to apopto-sis, but that the commitment to terminal di¡erentiation issu⁄cient. That di¡erent death pathways exist in keratinocytes issuggested by the ¢nding that interleukin-1 can inhibit TRAIL-and Fas-induced apoptosis in the transformed keratinocyte cellline HaCaT, but not UV-induced apoptosis (Kothny-Wilkes et al,1999). The inhibitory e¡ect of interleukin-1 involves the activa-tion of NF-kB, but it even enhances the apoptotic e¡ect of UVirradiation. How these pathways converge or diverge has yet to bedetermined, as also TNF-a is released during UV-induced apop-tosis and may add to massive cell death.It has been shown that TRAIL and other death ligands can also

activate p38 MAP kinase through their cognate receptors, leadingto increased expression of pro-apoptotic genes and eventually toautoampli¢cation of apoptosis in the target cell (Herr et al, 2000;Hou et al, 2002). It is tempting to speculate that proliferating cellsare thus more prone to apoptosis, as p38 MAP kinase has alsobeen implicated in relieving the Rb-mediated repression of thecell-cycle regulator and transcription factor E2F1 (Wang et al,1999). This protein promotes S-phase entry during the cell cycle,but it can also promote apoptosis by transcriptional and post-translational mechanisms, although a recent report implicatesE2F1 in the suppression of UV- and g-irradiation-induced apop-tosis (Wikonkal et al, 2003). Nevertheless, E2F1 is able to inducethe expression of pro-apoptotic genes (Pediconi et al, 2003) and,

Figure 5. Real-time quantitative PCR only revealed a small (2.5-fold) di¡erence in expression levels of DFF40 in cultured keratino-cytes and HL60 cells. See Materials and Methods for details.

Figure 4. DNA fragmentation was investigated in HL60 cells andkeratinocytes by means of agarose gel electrophoresis (insets) andpropidium iodide staining, followed by £ow cytometric analysis.Cells were stimulated with 250 ng per mLTRAIL for 3 h. (A) HL60 cells,unstimulated (negative control); (B) HL60 cells, stimulated; (C) undi¡eren-tiated keratinocytes, unstimulated; (D) undi¡erentiated keratinocytes, sti-mulated. Note the absence of fragmentation in keratinocytes.


when relieved from Rb-mediated repression by p38 MAP kinase,may contribute to the massive cell death seen in proliferating ker-atinocytes in response to TRAIL. Di¡erential and coordinatedexpression of E2F family members and Rb proteins modulateepidermal di¡erentiation and play opposite roles during cell dif-ferentiation (Paramio et al, 2000), which may also in part accountfor the di¡erences seen in apoptotic responses of proliferating anddi¡erentiating keratinocytes.TRAIL is also expressed in human epidermis at the mRNA

level (data not shown), posing questions about its function in epi-dermal homeostasis. As the epidermis is at the interface of theenvironment and the organism, and is constantly exposed to en-vironmental insult, it may serve as an apoptotic stimulus toneighboring cells whenever they are subject to processes that in-£ict irreversible damage. It thus provides an e⁄cient means to killunwanted cells (damage control). Recently, however, it has be-come apparent that TRAIL can be involved in reverse signaling,where signals are not transduced through the receptor butthrough the membrane anchored ligand upon ligation. This hasbeen demonstrated for activated T cells, where signaling thoughTRAIL enhances proliferation and interferon-g productionthrough a p38/MAPK dependent pathway (Chou et al, 2001). Itis tempting to speculate that TRAIL, although expressed in manydi¡erent tissues, also may play a role in skin development orhomeostasis. The established expression of TRAIL-R1, TRAIL-R2, and TRAIL-R4 (Leverkus et al, 2000) also adds fuel to theidea that this cytokine may play a more complex role than justbeing an inducer of apoptosis in the skin.One of the key features of terminal keratinocyte di¡erentiation

is that cells eventually will undergo programmed cell death,which takes days to complete as opposed to hours in response toTRAIL. Interestingly, this normal di¡erentiation process is notassociated with internucleosomal DNA fragmentation in cul-tured primary keratinocytes. Real-time quantitative RT-PCRdid not reveal an appreciable di¡erence in DFF40 expression be-tween primary keratinocytes and HL60 cells, indicating thatDNA degradation is di¡erently regulated in primary human ker-atinocytes. It is known that apoptosis with typical DNA frag-mentation does occur in cells of the keratinocyte lineage in hairfollicles, as has been shown in the past by agarose gel electrophor-esis and TUNEL staining (Lindner et al, 1997; Matsuo et al, 1998).There is additional evidence that epidermal stem cells with highproliferative capacity reside in the hair follicle (Taylor et al, 2000;Oshima et al, 2001). For reasons of safety these may be muchmore sensitive to apoptosis stimuli, and thus may in fact expressDFF40 to expedite apoptosis in the case of toxic or carcinogenicinsult; this may also explain why typical DNA fragmentation isonly seen in the hair follicle. Furthermore, DFF40 expressionseen in cultured keratinocytes may re£ect the presence of cells de-rived from hair follicles. In this study we have demonstrated thatTRAIL can induce apoptosis in human keratinocytes through itsreceptors. Further research is needed to elucidate its role in epi-dermal development and homeostasis. The fact that DFF40 is ex-pressed in keratinocytes, together with diminished sensitivity ofdi¡erentiating keratinocytes to TRAIL, strongly suggests thatsuicidal pathways during terminal di¡erentiation are distinctfrom typical apoptosis seen in response to cytotoxic stimuli, andfurther supports previous ¢ndings that terminal di¡erentiationinvolves other pathways (Eckhart et al, 2000; Lippens et al, 2000).Little is known about the machinery that leads to cellular decayduring keratinocyte di¡erentiation, and future research is neededto elucidate and identify relevant pathways and cellular machi-neries involved in this process.

The research described in this paper was supported by the Dutch Cancer Society(KWF).We gratefully thankVeronique Moulin, Department ofTumor Immunology,Nijmegen Center for Molecular Life Sciences, for her help in analyzing the quantita-tive PCR data.

REFERENCES

AraganeY, Kulms D, Metze D,Wilkes G, Poppelmann B, Luger TA, Schwarz T: Ul-traviolet light induces apoptosis via direct activation of CD95 (Fas/APO-1) in-dependently of its ligand CD95L. J Cell Biol 140:171^182, 1998

Basile JR, Zacny V, Munger K: The cytokines tumor necrosis factor-a (TNF-a) andTNF-related apoptosis-inducing ligand di¡erentially modulate proliferationand apoptotic pathways in human keratinocytes expressing the human papillo-mavirus-16 e7 oncoprotein. J Biol Chem 276:22522^22528, 2001

Bodmer JL, Holler N, Reynard S, et al: TRAIL receptor-2 signals apoptosis throughFADD and caspase-8. Nat Cell Biol 2:241^243, 2000

Chaturvedi V, Qin JZ, Denning MF, Choubey D, Diaz MO, Nickolo¡ BJ: Apoptosisin proliferating, senescent, and immortalized keratinocytes. J Biol Chem274:23358^23367, 1999

Chou AH,Tsai HF, Lin LL, Hsieh SL, Hsu PI, Hsu PN: Enhanced proliferation andincreased IFN-g production in T-cells by signal transduced through TNF-related apoptosis-inducing ligand. J Immunol 167:1347^1352, 2001

Eckert RL, Crish JF, Banks EB, Welter JF: The epidermis: Genes on ^ genes o¡.J Invest Dermatol 109:501^559, 1997a

Eckert RL, Crish JF, Robinson NA: The epidermal keratinocyte as a model forthe study of gene regulation and cell di¡erentiation. Physiol Rev 77:397^424,1997b

Eckhart L, DeclercqW, Ban J, et al: Terminal di¡erentiation of human keratinocytesand stratum corneum formation is associated with caspase-14 activation. J InvestDermatol 115:1148^1151, 2000

Fuchs E: Epidermal di¡erentiation and keratin gene expression. J Cell Sci Suppl17:197^208, 1993

Gandarillas A, Goldsmith LA, Gschmeissner S, Leigh IM,Watt FM: Evidence thatapoptosis and terminal di¡erentiation of epidermal keratinocytes are distinctprocesses. Exp Dermatol 8:71^79, 1999

Gygi SP, Rochon Y, Franza BR, Aebersold R: Correlation between protein andmRNA abundance in yeast. Mol Cell Biol 19:1720^1730, 1999

Hengartner MO:The biochemistry of apoptosis. Nature 407:770^776, 2000Herr I, Di Posovszky CMLD, Cifone MG, Boehler T, Debatin KM: Autoampli¢ca-

tion of apoptosis following ligation of CD95-L,TRAIL and TNF-a. Oncogene19:4255^4262, 2000

Hou ST, Xie X, BaggleyA, Park DS, Chen G,Walker T: Activation of the Rb/E2F1pathway by the nonproliferative p38 MAPK during Fas (APO1/CD95)-mediated neuronal apoptosis. J Biol Chem 277:48764^48770, 2002

Jansen BJH, van Ruissen F, de Jongh G, Zeeuwen PLJM, Schalkwijk J: Serial analysisof gene expression in di¡erentiated cultures of human epidermal keratinocytes.J Invest Dermatol 116:12^22, 2001

Jo M, Kim TH, Seol DW, Esplen JE, Dorko K, Billiar TR, Strom SC: Apoptosisinduced in normal human hepatocytes by tumor necrosis factor-related apop-tosis-inducing ligand. Nat Med 6:564^567, 2000

Kischkel FC, Lawrence DA, Chuntharapai A, Schow P, Kim KJ, Ashkenazi A:Apo2L/TRAIL-dependent recruitment of endogenous FADD and caspase-8to death receptors 4 and 5. Immunity 12:611^620, 2000

Kothny-Wilkes G, Kulms D, Poppelmann B, Luger TA, Kubin M, Schwarz T: Inter-leukin-1 protects transformed keratinocytes from tumor necrosis factor-relatedapoptosis-inducing ligand. J Biol Chem 273:29247^29253, 1998

Kothny-Wilkes G, Kulms D, Luger TA, Kubin M, Schwarz T: Interleukin-1 protectstransformed keratinocytes from tumor necrosis factor-related apoptosis-indu-cing ligand- and CD95-induced apoptosis but not from ultraviolet radiation-induced apoptosis. J Biol Chem 274:28916^28921, 1999

Kulms D, SchwarzT: Molecular mechanisms of UV-induced apoptosis. PhotodermatolPhotoimmunol Photomed 16:195^201, 2000

Kulms D, Poppelmann B,Yarosh D, Luger TA, Krutmann J, Schwarz T: Nuclear andcell membrane e¡ects contribute independently to the induction of apoptosisin human cells exposed to UVB radiation. Proc Natl Acad Sci USA 96:7974^7979, 1999

Leverkus M, Yaar M, Gilchrest BA: Fas/Fas ligand interaction contributes toUV-induced apoptosis in human keratinocytes. Exp Cell Res 232:255^262,1997

Leverkus M, Neumann M, Mengling T, Rauch CT, Brocker EB, Krammer PH,Walczak H: Regulation of tumor necrosis factor-related apoptosis-inducing li-gand sensitivity in primary and transformed human keratinocytes. Cancer Res60:553^559, 2000

Lindner G, BotchkarevVA, Botchkareva NV, Ling G, van der Veen C, Paus R: Ana-lysis of apoptosis during hair follicle regression (catagen). AmJ Pathol 151:1601^1617, 1997

Lippens S, Kockx M, Knaapen M, et al: Epidermal di¡erentiation does not involvethe pro-apoptotic executioner caspases, but is associated with caspase-14 induc-tion and processing. Cell Death Di¡er 7:1218^1224, 2000

Liu X, Zou H, Slaughter C,Wang X: DFF, a heterodimeric protein that functionsdownstream of caspase-3 to trigger DNA fragmentation during apoptosis. Cell89:175^184, 1997

Liu ZG, Baskaran R, Lea-Chou ET,Wood LD, Chen Y, Karin M,Wang JY: Threedistinct signalling responses by murine ¢broblasts to genotoxic stress. Nature384:273^276, 1996

Matsuo K, Mori O, HashimotoT: Apoptosis in murine hair follicles during catagenregression. Arch Dermatol Res 290:133^136, 1998


Muzio M, Stockwell BR, Stennicke HR, Salvesen GS, Dixit VM: An inducedproximity model for caspase-8 activation. J Biol Chem 273:2926^2930, 1998

Oshima H, Rochat A, Kedzia C, Kobayashi K, Barrandon Y: Morphogenesis andrenewal of hair follicles from adult multipotent stem cells. Cell 104:233^245,2001

Paramio JM, Segrelles C, Casanova ML, Jorcano JL: Opposite functions for E2F1 andE2F4 in human epidermal keratinocyte di¡erentiation. J Biol Chem 275:41219^41226, 2000

Pediconi N, Ianari A, Costanzo A, et al: Di¡erential regulation of E2F1 apoptotictarget genes in response to DNA damage. Nat Cell Biol 5:552^558, 2003

Pfundt R,Wingens M, Bergers M, Zweers M, Frenken M, Schalkwijk J:TNF-a andserum induce SKALP/ela¢n gene expression in human keratinocytes by a p38MAP kinase-dependent pathway. Arch Dermatol Res 292:180^187, 2000

Pfundt R, vanVlijmen-Willems I, Bergers M,Wingens M, CloinW, Schalkwijk J: Insitu demonstration of phosphorylated c-jun and p38 MAP kinase in epidermalkeratinocytes following ultraviolet B irradiation of human skin. J Pathol193:248^255, 2001

Qin JZ, Chaturvedi V, Denning MF, Choubey D, Diaz MO, Nickolo¡ BJ: Role ofNF-kB in the apoptotic-resistant phenotype of keratinocytes. J Biol Chem274:37957^37964, 1999

Qin JZ, Bacon PE, Chaturvedi V, Bonish B, Nickolo¡ BJ: Pathways involved in pro-liferating, senescent and immortalized keratinocyte cell death mediated by twodi¡erent TRAIL preparations. Exp Dermatol 11:573^583, 2002

Sakahira H, Enari M, Nagata S: Cleavage of CAD inhibitor in CAD activation andDNA degradation during apoptosis. Nature 391:96^99, 1998

Salvesen GS, Dixit VM: Caspase activation: The induced-proximity model. Proc NatlAcad Sci USA 96:10964^10967, 1999

Sheikh MS, Antinore MJ, Huang Y, Fornace AJ: Ultraviolet-irradiation-inducedapoptosis is mediated via ligand independent activation of tumor necrosis fac-tor receptor 1. Oncogene 17:2555^2563, 1998

Sprick MR, Weigand MA, Rieser E, et al: FADD/MORT1 and caspase-8 are re-cruited to TRAIL receptors 1 and 2 and are essential for apoptosis mediatedbyTRAIL receptor 2. Immunity 12:599^609, 2000

Taylor G, Lehrer MS, Jensen PJ, SunTT, Lavker RM: Involvement of follicular stemcells in forming not only the follicle but also the epidermis. Cell 102:451^461,2000

Traidl C, Sebastiani S, Albanesi C, Merk HF, Puddu P, Girolomoni G, Cavani A:Disparate cytotoxic activity of nickel-speci¢c CD8þ and CD4þ T-cell sub-sets against keratinocytes. J Immunol 165:3058^3064, 2000

Trautmann A, Akdis M, Kleemann D, et al:T cell-mediated Fas-induced keratinocyteapoptosis plays a key pathogenetic role in eczematous dermatitis. J Clin Invest106:25^35, 2000

Van Ruissen F, de Jongh GJ, Zeeuwen PL,Van Erp PE, Madsen P, Schalkwijk J: In-duction of normal and psoriatic phenotypes in submerged keratinocyte cul-tures. J Cell Physiol 168:442^452, 1996

Viard I,Wehrli P, Bullani R, et al: Inhibition of toxic epidermal necrolysis by block-ade of CD95 with human intravenous immunoglobulin. Science 282:490^493,1998

Wang S, Nath N, Minden A, Chellappan S: Regulation of Rb and E2F by signaltransduction cascades: Divergent e¡ects of JNK1 and p38 kinases. EMBO J18:1559^1570, 1999

Watt FM: Terminal di¡erentiation of epidermal keratinocytes. Curr Opin Cell Biol1:1107^1115, 1989

Wikonkal NM, Remenyik E, Knezevic D, et al: Inactivating E2f1 reverts apoptosisresistance and cancer sensitivity in Trp53-de¢cient mice. Nat Cell Biol 5:655^660, 2003

Wyllie AH, Kerr JF, Currie AR: Cell death: The signi¢cance of apoptosis. Interna-tional Rev Cytol 68:251^306, 1980

Ziegler A, Jonason AS, Le¡ell DJ, et al: Sunburn and p53 in the onset of skin cancer.Nature 372:773^776, 1994


tumor necrosis factor related apoptosis inducing ligand triggers apoptosis in dividing but not in...

Documents