Vol. 6, 807-816, july 1995 Cell Growth & Differentiation 807

A Secreted Serine Protease Can Induce Apoptosisin Pam212 Keratinocytes

Jeffrey Marthinuss, Patricia Andrade-Gordon,and Miri Seiberg1Skin Biology Research Center, The R. W. Johnson Pharmaceutical

Research Institute, Raritan, New Jersey 08869 [J. M., M. S.], and DrugDiscovery, The R. W. Johnson Pharmaceutical Research Institute, SpringHouse, Pennsylvania 19477 [P. A-G.E

Abstrad

The epidermal keratinocyte cell line Pam2l 2 undergoesspontaneous apoptosis in culture, providing an in vitromodel for the early steps of epidermal differentiation.Pam2l 2 cells exhibit charaderistics of basalkeratinocytes, committed for the transition to thespinous layer of the epidermis. Bcl-2 can regulate thedifferentiation of these cells by negatively regulatingseveral genes that have been implicated in apoptosis.We show evidence that a serine protease adivity,secreted by the Pam2l 2 cells, could induce apoptosis inPam2l 2 and several other cell lines. This adivity mightbe regulated via the bcl-2 pathway. We suggest that thisserine protease could either diredly, via binding and/orcleavage of a serine protease-adivated receptor, orindiredly, via the cleavage of an unknown protein,adivate the signaling for apoptosis in Pam2l 2 cells.Alternatively, this secreted serine protease could reenterthe cell and start a proteolytic cascade readion thatleads to cell death. This is based on the indudion ofapoptosis in several cell lines by the partially purifiedserine protease adivity, and the minimal effed ofprotein synthesis inhibition on Pam2l 2 apoptosis. Wepropose that in vivo, a two-step mechanism controlskeratinocyte apoptosis and differentiation. The basalcells of the epidermis contain all of the necessaryproteins required for apoptosis, as well as the repressorprotein BcI-2. As Bcl-2 levels go down, the cells committo terminal differentiation. A serine protease, secretedfrom these cells, then induces the death process. Thissecond step enables the cells to undergo apoptosis andcontinue the process of terminal differentiation.

Introdudion

PCD2 is a fundamental aspect of development, morphogen-esis, and tissue homeostasis (1-3). Many PCD pathways

Received 2!9!95; revised 4!i 0/95; accepted 4!27195.

1 To whom requests for reprints should be addressed, at Skin Biology

Research Center, Johnson & Johnson CPWW, Administration Building, 199Grandview Road, Skillman, NJ 08558.

2 The abbreviations used are: PCD, programmed cell death; TUNEL, terminaldeoxynucleotidyl transferase-mediated nick-end labeling; PAR-2, protease-

activated receptor 2; CM, conditioned medium or media; STI, soybeantrypsin inhibitor; PMSF, phenylmethylsulfonyl fluoride; DCI, 3,4-dichloro-

isocoumarin; TPCK, L-1 -chloro-3-14-tosylamido]-4-phenyl-2-butanone;cHx, cycloheximide; RT-PCR, reverse transcription-PCR; TR, thrombinreceptor; uPA, urokinase plasminogen activator; ICE, interleukin convertingenzyme; SCCE, stratum corneum chymotryptic enzyme.

lead to apoptosis, a mode of cell death involving cytoplas-mic condensation and specific DNA fragmentation (4-6).Genomic DNA is initially degraded into high molecularweight fragments (300 and 50 kb), followed by the releaseof oligonucleosomal DNA fragments that produce a char-actenistic DNA ladder (7, 8). This nuclear DNA fragmenta-tion was found to correlate with apoptosis in many systems,but examples of PCD in the absence of the DNA ladderhave also been reported (1 ). The controls and signals initi-ating apoptosis are only partially shared between different

biological systems (reviewed in Refs. 6, 9, and 1 0). In manysystems, apoptosis requires new RNA and protein synthesis.In others, where no new synthesis is required, it is suggestedthat either the removal of an inhibitor (with a shorter half-life than that of the apoptotic machinery) or the transduc-tion of a signal results in PCD (reviewed in Refs. 1 and 1 1).A number of gene products have been identified that affectthe apoptotic process. One of these genes, bcl-2, was foundto negatively regulate PCD and to reduce or eliminateapoptosis (reviewed in Ref. 1 2).

Cytoplasmic proteases play a functional role in PCD. Cellgranule proteases (granzymes) were found to induce apop-tosis in permeabilized cells (1 3-20). The Caenorhabditiselegans protein Ced-3 is essential for cell death (21). Itsmammalian homologues are the cysteine proteases ICE,Nedd-2, Ich-1, prICE, and CPP32, which act as vertebratePCD genes (22-27). Inhibitors of the cysteine protease cal-pain inhibit apoptotic cell death in both the “induction” andthe “release of inhibition” models of apoptosis (28). More-over, senine protease inhibitors can block apoptotic celldeath (29-31). These inhibitors act early in the PCD path-way, affecting the cell shrinkage and the nuclear collapse

but not the high molecular weight DNA cleavage (31). Theactivity of the serine proteases is not dependent on RNA orprotein synthesis (31).

PCD is central to skin biology because epidermal differ-entiation involves PCD (reviewed in Ref. 32). The epidermisis composed of four layers, starting with the mitotically

active basal keratinocytes and ending with terminally dif-ferentiated cornified cells (reviewed in Refs. 33-36). Duringepidermal differentiation, keratinocytes migrate outwardand undergo apoptosis to become cornified cells. Apoptoticcells have been identified in the epidermis (37), mainly inthe granular layer (38). We had shown previously thatPam2i 2 cells (a spontaneously transformed, newbornBALB/c-derived epidermal keratinocyte cell line; Ref. 39)undergo spontaneous apoptosis in culture. Apoptosis wasdemonstrated by morphology, apoptotic stain (TUNEL), andDNA fragmentation (40). We also demonstrated thatPam2l 2 cells exhibit characteristics of basal keratinocytes.The spontaneous apoptosis ofthese cells represents a modelof the transition from the basal to the spinous layer of theepidermis. Upon characterization of apoptosis in Pam2i 2cells, we demonstrated that bcl-2 can control the differen-tiation of these cells by negatively regulating several genesthat have been implicated in apoptosis (40).

a -

.� _-f-S

.t�e - --

‘I

em

C �

30

25

� 20

4’ 15

3T3C PamCM 3T3CM

b �#{176}

35

30

w- ‘

c�

JJ.0

�0.�-I

C.)

3T3

IC,)

(3

(3

�0�

808 Serine Protease Induces Apoptosis in Keratinocytes

d16

14

12

� 10

a

2

PamC PamcM 353 CM

�. �,e�

��2 �

� C.�r% � ‘�� �

: �

eC.) C.)

E

o�E

on�(0)

�(�)

��;;

3T3 Pam 3T3 Pam

I I i

Fig. 1. Pam212 CM can induce

apoptosis. Subconfluent culturesof Pam212 and 3T3 cells weretreated with 3T3 and Pam2l 2 CM,

supplemented with an equal vol-ume of DMEM, without serum, for

3 days. Cells were TUNEL stained,

and apoptotic cells were detected

with peroxidase. 3T3 cells treatedwith 3T3 CM (a) and Pam212 CM

(b) are shown (Bars, 3.125 pm).Note that partial detachment ofapoptotic cells puts them “out of

focus,” relative to attached cells.The brown staining, therefore,

represents apoptotic, but yet fully

attached cells; therefore, it is Un-derrepresentative of the true rela-

tive level of apoptosis. Quantita-tion ofTUNEL stained 3T3 (c( and

Pam212 (d) cells treated with the

different CM (counted with con-tinuous changing of focus) andstatistical analysis reveals that

only Pam212 CM can induce ap-

optosis. 3T3 Cand Pam Care thecontrol, untreated cells. Bars, SD.Size analysis of 3T3 and Pam212

DNAs, following 3 days incuba-tion in the presence of the differ-

ent CM (ladder assay, e), showsincreased fragmentation only withPam212 CM.

�a�:l�I ;;‘. (ra,:��:L �

:‘.. � �,g’ #{149} 4��-�!r �

� . � . #{149}_‘:-�‘- .14 #{149}“� �. ,,\ .1��:

� �,� -

- .‘ �‘,“.) _.� .:*A. --

�

Fig. 2. CM from ASbcI-2 cells (Pam2i 2 cells expressing reduced levels ofthe Bcl-2 protein) induces more apoptosis than Pam212 CM. a, 3T3 cells treated withASbcl-2 CM. Compare to Fig. 1, a and b. Bar, 3.1 25 pm. b, quantitation and statistical analysis ofTUNEL-stained 3T3 cells treated with 3T3, Pam2i2, and ASbcl-2CMs indicates that bc!-2affectsthe level ofthe CM apoptosis-inducing activity. Comparing Pam2i2 and ASbcl-2 CMs in this experiment, the Student ttest shows

P = 0.21 ; bars, SD. c, DNA size analysis shows inverse correlation between bcl-2 levels and DNA fragmentation. Details as in Fig. 1.

Cell Growth & Differentiation 809

Here we show evidence that a senine protease activity,secreted by the Pam2l 2 cells, can induce apoptosis inPam2i 2 and several other cell lines. This activity might beregulated via the bcl-2 pathway. The possible mechanismsfor the serine protease-induced apoptosis were investigated.We propose that either the protease itself, or the cleavage ofan unknown protein by this protease, results in a ligand thatactivates the apoptosis receptor. Alternatively, the cleavageof a senine protease-activated receptor is involved in thesignaling of apoptosis. This receptor is probably differentfrom the TR (41) and the PAR-2 (42). We cannot rule out thepossibility that the secreted senine protease could reenterthe cell and initiate a cascade of proteolytic reactions thatlead to apoptosis and cell death.

Results

Pam2l 2-conditioned Media Can Induce Apoptosis. Wehad shown previously that the epidermal keratinocyte cellline Pam212 undergoes spontaneous apoptosis in culture(40). We found that the degree of apoptosis in this cell linecorrelates with cell density. This might result from eithercell-cell contact or changes in media composition. Dailychanges of the culture media of confluent Pam2i 2 cellsresults in reduction or elimination of DNA fragmentationwithin the attached cell population (data not shown). Thissuggests that a cumulative change in the media and not thecell-cell contact is important for the DNA fragmentation.Such a change could be either the consumption and elim-ination of an essential media component or the accumula-tion of a “factor” involved in apoptosis. Therefore, we com-pared CM from Pam2i 2 and 3T3 (control) cells. CM werecollected after growing cells for 3 days at 1 06 cells/i 0-cmplate. CM were centrifuged to remove apoptotic cells anddebris and applied to 3 x i0� or 106 cells/10-cm plate of3T3 and Pam212 cells for 3 days. As shown in Fig. 1, onlythe Pam2l 2 CM, and not the 3T3 CM, could induce apop-tosis in these two cell lines. The Pam2l 2 CM could alsoinduce apoptosis in T-47D and HepG2 cells (data notshown). Apoptosis was demonstrated by morphology andTUNEL staining (Fig. 1 , a and b). Statistical analysis of theTUNEL staining data (Fig. 1 , c and d) revealed a significantincrease in apoptosis only in cultures treated with thePam2i2 CM. DNA analysis of both attached (data notshown) and detached (Fig. 1 e) cells showed an increase inDNA fragmentation in the cells treated with the Pam2i 2CM. The increase in DNA laddering was more pronouncedin the detached cell fraction (Fig. 1 e), since the Pam2l 2 CMdramatically increases the number of apoptotic cells de-tached from the culture plate. The relative effect of thePam2i 2 CM was more pronounced on the 3T3 cells (com-pare Fig. 1 c to Pam2i 2 in Fig. 1 d), because they do notexhibit as high a level of spontaneous apoptotic death as thePam2i 2 cells (40). These data indicate that: (a) the generalconsumption of media components by cells does not in-duce 3T3 or Pam2i 2 cells to die via apoptosis (see 3T3CM); and (b) a “factor,” which exists in the Pam2l 2 CM but

not in the 3T3 CM, can induce these cells and others to

undergo apoptosis.The “CM apoptotic activity” could result from a combi-

nation of the “factor” and reduced concentration of nutri-ents or accumulation of toxic breakdown products In themedium. Therefore, we supplemented the CM with freshDMEM. We found that the Pam2i 2 CM could induce ap-optosis in Pam2l 2, 3T3, T-47D, and HepG2 cell lines, evenwhen supplemented with large volumes of fresh media(data not shown). This observation led us to believe that theCM apoptotic activity does not require the depleted mediato induce apoptosis.

To facilitate biochemical analysis of the CM apoptoticactivity, we assayed the growth and apoptosis of Pam2l 2and 3T3 cells in serum-free medium. Minimal changeswere observed in cell morphology and growth, and nochanges were observed in the ability of the Pam2l 2 CM toinduce apoptosis when the Pam212 cells were grown inplain DMEM for 3-6 days (data not shown). This suggeststhat no serum component is essential for the apoptoticactivity of the Pam212 CM. A low level of apoptosis wasdemonstrated in the subconfluent 3T3 control culture whengrown in serum-free medium. No apoptosis of 3T3 cells isdetectable in the presence of 10% serum (40). The 3T3 CMfrom both serum-free and serum-supplemented media

could not induce apoptosis in any cell line tested.CM Apoptotic Adivity Might Be Regulated via the bcl-2

Pathway. Pam2l 2 cells transfected with an antisense bcl-2expression vector (ASbcl-2-cells) exhibit an increased levelof apoptosis that correlates with the down-regulation of thebcl-2 gene (40). Analysis of ASbcl-2 CM revealed that thisCM is more potent than the Pam2l 2 CM in inducing ap-optosis. As shown in Fig. 2, ASbcl-2 CM, relative to Pam2l 2CM, could induce higher levels ofapoptosis and more DNAfragmentation in 3T3 cells (compare to Fig. 1). While therelative effect was not as strong in Pam2l 2 cells, and there-fore did not reach statistical significance, the ASbcl-2 CMalways induced more apoptosis than the Pam2l 2 CM. Theapoptotic activity of CM from control transfected Pam2i 2cells was indistinguishable from that of Pam2i 2 CM (datanot shown). It is possible, therefore, that the production ofthe CM apoptotic activity is regulated via the bcl-2pathway.

Pam2l 2 Apoptosis Is Induced by Trypsin and Blocked bySoybean Trypsin Inhibitor. To study the possibility that theCM apoptotic activityisa protein, we treated the Pam212CM with trypsin, followed by STI inactivation. Contrary toour expectations, we found that Pam2i 2 CM, treated withtrypsin for 30 mm, followed by STI inactivation, results inhigh levels of apoptosis. Furthermore, trypsin-treatedPam2l 2 CM (with no STI inactivation) induced higher 1ev-els of DNA fragmentation, and the STl-treated CM alone(with no trypsin) eliminated the DNA laddering of Pam2i 2cells. To further investigate this effect, we treatedPam212 cultures (which were not exposed to any CM)with trypsin and STI. Treatment of the Pam2l 2 cells with

a

d 7#{176}60

50

0. 400

� 30

20

10

0PamC Tryp STI

C

C.>1

IFig. 3. Trypsin induces and STI inhibits Pam2l 2 apoptosis. TUNEL staining and fluorescence detection of Pam2l 2 cells treated with DMEM (spontaneous

apoptosis; a), trypsin (10 mgjml; b(, and STI (100 pg/mI; c(. Bars, 3.125 pm. Note the increased number of apoptotic bodies in the trypsin-treated cells (b).Apoptotic, STI-treated cells were rarely detected (c(. Quantitation and statistical analysis of TUNEL-stained Pam2l 2 cells (d; bars, SD) and DNA fragmentationanalysis (e( shows that trypsin is a strong inducer and STI is a strong inhibitor of Pam2l 2 apoptosis. These effects were dose dependent (data not shown); maximaleffect is shown. Details as in Fig. 1.

810 Serine Protease Induces Apoptosis in Keratinocytes

trypsin and STI results in the enhancement and inhibition

of apoptosis, respectively (Fig. 3). TUNEL staining (Fig. 3,

a-d) and ladder assays (Fig. 3e) demonstrated that trypsin

enhances, and trypsin inhibitor eliminates, the spontane-

ous apoptosis and the DNA fragmentation of these cells.This suggests that trypsin, or a similar serine protease

activity, might be involved in keratinocyte apoptosis.

e.cC

C.>1

a,�0#{149}0

-I

Serine Protease Inhibitors Block Apoptosis and DNAFragmentation. The blocking of PCD and apoptosis by

serine protease inhibitors has been demonstrated before(29-31). To investigate the possibility that serine protease

activity might be involved in Pam2l 2 apoptosis, we studiedthe effect of several protease inhibitors on DNA fragmen-

tation. The serine protease inhibitors PMSF (inhibits chy-

0. -U. � � -Cl) � (3 0.� C) #{149} �

C) a. 0 .j I- C) C)

30

25

20

15

10

5

Fig.4. Some serine protease inhibitors can inhibit Pam212 apoptosis.

Pam212 cells were treated for 3 days with PMSF (10, 3, and 0.3 mss), DCI(100, 30, and 10 pM), leupeptin (10, 1, and 0.1 mM), TPCK (10, 1, and 0.1

mM), and calpain I inhibitor (1 msi and 1 00 and 25 pM). DNAs were extractedfrom the media (representing apoptotic cells detached from the plate) and

analyzed by the ladder assay. The wide ends of the triangles represent thehighest concentration tested of each inhibitor. A dose-response is demon-

strated for DCI and PMSF. When no effect was observed, concentrations of

up to 1 0 mM were tested.

0(3 � ,- c,J C’) � 03 CDC,)

I-C’)

0.

Fig. 5. Partial purification of the Pam212 apoptosis-inducing activity.Pam212 CM was fractionated over an STI column. Fractions were eluted atlow pH, neutralized, and dialyzed against DMEM (M, 10,000 cutoff). Frac-(ions were applied to 3T3 cells for 3 days, supplemented with an equal

volume of DMEM, followed by TUNEL staining. Quantitation and statistical

analysis indicates that fractions 3 and 4 have a significant increase in

apoptosis-inducing activity, compared to the Pam212 CM. Similar resultswere obtained with Pam212 cells (data not shown(. Silver stain analysis of

SDS-PAGE revealed no bands in these active fractions, while many bandswere detected in the Pam2l 2 CM lane (data not shown). This demonstrates

the low level ofthis protein in the Pam212 CM.

Cell Growth & Differentiation 811

motrypsin, trypsin, and thrombin), and DCI (inhibits a largenumber of serine proteases, such as elastase, cathepsin G,and endoproteinase Glu-C) could inhibit oligonucleosomaldegradation of Pam2i 2 DNA in a dose-dependent manner(Fig. 4). However, while the effect of DCI was observed at10 �M and was very dramatic at 30 �M, only 10 m�i PMSFcould result in a similar level of inhibition. This could bedue either to the nature of the inhibited protease or to themechanism of inhibition. The serine protease inhibitorsleupeptin (inhibits serine and thiol protease such as trypsin,plasmin, papain, and cathepsin B) and TPCK (inhibits manyserine proteases such as chymotrypsin, bromelanin, ficin,and papain) did not inhibit the DNA fragmentation, even atconcentrations as high as 1 0 mtvi (Fig. 4). These data suggestthat a serine protease, inhibited by DCI (and to a lower levelby PMSF), could be involved in Pam212 oligonucleosomaldegradation. The calpain I inhibitor (inhibits a calcium-dependent neutral cysteine protease, which has been im-plicated in apoptosis) inhibited DNA degradation at 1 mt�’tbut enhanced the fragmentation at 0.1 m�i (Fig. 4). No effectwas detected at 1 0, 25, and 50 �M (data not shown), a rangeof concentrations that inhibits thymocyte apoptosis. Al-though these data could suggest that calpain I might beinvolved in DNA fragmentation of Pam2l 2 apoptosis, itclearly demonstrates that calpain I is not a major player inthis process. Other wide range inhibitors of cysteine pro-teases, metalloproteases, am i nopeptidases, and acidproteases, did not affect the oligonucleosomal degradation(data not shown).

Partial Purification of the Apoptotic Adivity. In an at-tempt to partially purify the putative senine protease-apop-totic activity from the Pam2l 2 CM, we used an affinity resincontaining immobilized STI. Proteins that bound to the STIcolumn were eluted at low pH and immediately neutral-ized. The column flow-through and the various fractionswere dialyzed (Mr 10,000 cutoff) against DMEM and ap-plied to both Pam212 and 3T3 cells for 3 days. Statisticalanalysis of the TUNEL staining data revealed a significantincrease in the apoptosis-inducing activity of two fractionseluted from the STI column (fractions 3 and 4; Fig. 5),compared to the total Pam2i 2 CM. The column flow-through had a reduced apoptotic activity, compared to the

starting material. The control fractionated 3T3 CM had noapoptotic activity in all of the fractions tested (data notshown). These data indicate that a Pam2l 2 protein(s), big-ger than Mr 10,000 and with the ability to bind STI, caninduce apoptosis in Pam212 and 313 cells.

Trying to quantitate the apoptotic activity of the elutedfractions, we found that the STI-punified fraction (or the

amount of CM used) is not the rate-limiting step of the deathresponse. Both the dose-response and the maximal percent-age of apoptotic cells within the culture were found to be

cell-type dependent. A 10-fold increase in the STI-punifiedfraction could not increase that cell type-dependent maxi-mal level of apoptosis in the culture. Therefore, our data aresemiquantitative only. We believe this saturation responsecould provide a clue for the pathway/mechanism of thePam212 CM-induced PCD. The CM apoptotic activityeluted from the STI column is at least 1000 times moreconcentrated than the original CM activity, based on dilu-tion experiments (data not shown). Yet, we are unable toidentify a band(s) on protein gels, demonstrating the limitingquantities of this activity.

Mechanism of Apoptosis in Pam2l 2 Cells. The mode ofactivation ofthe PCD pathway can be analyzed by blockingprotein synthesis. In the “induction” mechanism, when newgene expression is required, apoptosis is blocked whenmRNA or protein synthesis is inhibited. In the “release”mechanism, the death genes are constitutively expressedbut inhibited by a repressor. Blocking protein synthesisresults in the induction of apoptosis, if the repressor has ashorter half-life than that of the death proteins. In the “trans-duction” mechanism, all of the necessary molecules forapoptosis are ready and are activated by the transduction ofa signal. In this case, the inhibition of macromolecularsynthesis has no effect on apoptosis.

No changes were observed in the level of apoptosis (Fig.6) or DNA fragmentation (not shown) when Pam212 cells

16

14

12

0.

6

4

2

0

Fig. 6. Protein synthesis inhibition by CH�< does not affect Pam2i 2 apop-tosis. Pam2i2 cells were treated with CHx (10 pg/mI) for 5 and 18 h (o/n).After 5 h CHX treatment, cells were also allowed to recover for 2 h in DMEM

(Wash), to analyze the effect of the accumulation of death mRNAs. Cells

were TUNEL stained and counted. Statistical analysis indicates a nonsignif-icant effect of CH� on Pam2l 2 apoptosis. Similar results were obtained bythe ladder assay (data not shown). Bars, SD.

812 Serine Protease Induces Apoptosis in Keratinocytes

4 M. Seiberg, unpublished data.

were treated with CHX (1 0 �jg/ml for 5 h). Washing the CHXand allowing the cells to recover for 2 h led to a verylimited, nonsignificant increase in the percentage of apop-totic cells (2-5%; Fig. 6). A longer (18-h) treatment withCHX results in the same low increase in apoptosis, probablydue to the long inhibition of protein synthesis. Therefore,we conclude that induction of apoptosis in Pam212 cellsdoes not require new gene expression and is not mediatedvia the release mechanism. The senine protease-inducedPam212 apoptosis could involve, therefore, either a signaltransduction event or an intracellular proteolytic cleavage,since both these mechanisms are not affected by CHX.

A Serine Protease-adivated Receptor Might Be Involvedin the Signaling of Apoptosis. A senine protease-activatedsignal transduction mechanism could activate a receptordirectly by proteolytic cleavage. Alternatively, a cleavageproduct of an unknown protein could activate the receptor.We looked for candidate receptors that might be directlyinvolved in the induction of apoptosis. The only well-char-actenized, protease-activated, C-protein-coupled receptoris the TR. When cleaved by the senine protease thrombin,the new NH2-terminal of this receptor acts as a self-activat-ing tethered ligand (41). PAR-2 has been cloned recentlyand shown to be activated through a similar tethered ligandmechanism (42). Its physiological activator has not yet beendescribed, although it is activated by trypsin (42). Humankeratinocytes possess, in addition to TR, functional PAR-2.3RT-PCR analysis revealed that both TR and PAR-2 are ex-pressed in Pam2i 2 cells, and their mRNA levels do not

3 R. J. Santulli, C. K. Derian, A. L. Darrow, K. A. Tomko, A. J. Eckardt, M.Seiberg, R. M. Scarborough, and P. Andrade-Gordon. Evidence for thepresence of a protease-activated receptor distinct from thrombin receptor inhuman keratinocytes, submitted for publication.

correlate with the cellular level of the Bcl-2 protein (Fig.7a). While TR and PAR-2 mRNA levels are not regulated bybcl-2, their expression in keratinocytes makes them goodcand idates for a sen ne protease-activated, receptor-med i-ated signaling of apoptosis. Therefore, we studied thepossible role of these receptors in Pam2l 2 apoptosis.

Treatment of Pam2l 2 cultures with the TR peptideSFLLRN, which mimics its tethered ligand (41) as well aswith the thrombin inhibitorhirudin, both resultin an insig-nificant decrease in apoptosis (Fig. 7b). The Xenopus IaevisTR agonist peptide TFRIFD (43) and the negative controlpeptide FSLLRN slightly induced apoptosis in these cells(Fig. 7b). These data suggest that the TR of Pam2i 2 kerati-nocytes is not involved in apoptosis. SLIGRL, the agonistpeptide for PAR-2 (42), had no effect on the level of apop-tosis in Pam2l 2 cells (Fig. 7b), excluding this receptor as acandidate for the signaling mechanism. We suggest that if asenine protease activates the signaling of apoptosis in kera-tinocytes, this process involves either a novel senine pro-tease-activated receptor or the cleavage of an unknownprotein.

The protease could also serve as a ligand that activatesthe “apoptosis receptor” by binding and not by proteolyticcleavage. There is growing evidence that the occupancy ofthe uPA receptor, which is known to be expressed in kera-tinocytes, might activate a signal transduction pathway (re-viewed in Ref. 44). Both uPA, plasminogen activator inhib-itor-1 (a natural inhibitor of uPA), and antibodies to themouse uPA had only marginal effect on the level of Pam2i 2apoptosis (Fig. 7b), excluding the uPA receptor as a majorcandidate for apoptosis signaling.

Another possibility is a granule-exocytosis mechanism,like the one used by cytotoxic lymphocytes. This mecha-nism combines perform, a pore-forming protein, and gran-zymes, a family of granule proteins, many of which aresenine proteases, that together induce target cell DNAbreakdown and apoptotic cell death (1 3-20). At least onemember of the granzyme family, TIA-1 , is expressed inPam2i 2 cells.4 Our data cannot distinguish between such amechanism and a signal transduction event.

Discussion

We show evidence for a senine protease activity secreted bythe Pam2i 2 keratinocyte cell line that might be regulatedvia the bcl-2 pathway. This activity can induce apoptosis inseveral cell lines.

Cytoplasmic proteases are involved in many PCD andapoptosis pathways. In addition, proteases are highly con-served in the evolution of apoptosis. The C. elegans deathgene Ced-3 is a protease, which shares similarity with theprotein family of the mammalian cysteine protease ICE.Both Ced-3 and ICE cause cell death when expressed inrodent fibroblasts (23). Interestingly, the active form of ICE,which cleaves pro interleukin 1�3 to active interleukin 1 f3, isproduced by a cleavage of a precursor protein (45). It issuggested that ICE cleaves another, yet unknown substrate,to trigger cell death. However, mice deficient in ICE showno phenotype of dysregulated apoptosis (46).

The endonuclease step is not essential for cell death. TheC. elegans death genes, Ced-3 and Ced-4, act prior to theendonuclease nuc-1 . In thymocyte-apoptosis, proteolysis is

b3#{176}

25

a G3PDH PAR-2 Th-R 20

��1I -iF 1

m+.PAm+.PA 15

110

5

00 �. 0. �: � � ‘- < ‘C� a �- � � � �I ‘ 0. �.

0. C�J .? �< 0.�

0.

Cell Growth & Differentiation 813

Fig. 7. Excluding candidates for a protease-activated receptor involved in Pam2l 2 apoptosis. a, RT-PCR demonstrates that both Th-R and PAR-2 are expressedin Pam212 keratinocytes. +, positive control, using total mouse embryo (13.5 days) RNA for the RT-PCR. -, negative control, using the DNased embryo RNA

with no RT for the PCR reaction. Pam2l 2 RNA (P) and ASbcI-2 RNA (A) were used for the RT-PCR. b, level of apoptosis in Pam212 cells treated with the PAR-2

agonist peptide SLIGRL (Par-2pep; 100 pM), the TR agonist peptide SFLLRN (Thrpep; 100 pM), the Xenopus laevisTR agonist peptide TFRIFD (XL pep; 300 p�i(.the negative control peptide FSLLRN (negpep; 300 pM), hirudin (Hir 300 nsi(, plasminogen activator inhibitor-i (PAl-i; 300 nM(, uPAiO units/mI), and mouse

anti-uPA antibodies (Ab uPA; 50 pg/mI). Peptides and inhibitors were added daily for 3 days, followed by TUNEL staining, quantitation, and statistical analysis.

Bars, SD. No significant change in the level of apoptosis was demonstrated. Same results were also obtained using the ladder assays (data not shown).

required for the activation of the endonuclease step (31).DNase I, a likely candidate for the mammalian nucleaseinvolved in DNA fragmentation, is kept inactive by com-plexing with G-actin (47). It is possible that a proteasecleaves this complex, releasing the DNase I from actinand allowing it to translocate into the nucleus. Such amechanism could explain the demonstrated direct effectof proteases on DNA fragmentation (31).

The C. elegans inhibitor of apoptosis, the Ced-9 gene, hasbeen conserved in evolution as well. Its human homologue,bcl-2, can function in worm, insect, and mammalian cellsto antagonize cell death induced by divergent mechanisms(reviewed in Ref. 48). We had shown previously that bcl-2regulates apoptosis in Pam2l 2 cells and suggested a role forbcl-2 in the control of keratinocyte terminal differentiation(40). Here we show that down-regulation of bcl-2 can resultin an increase in the serine protease apoptosis-inducingactivity (see ASbcl-CM). The level of Bcl-2 protein in thesecells is inversely correlated with the level of the apoptosis-inducing activity. CM from sense bc/-2-transfected Pam2i 2cells induces similar levels of apoptosis as Pam212 CM.Sense bcl-2 transfected Pam2l 2 cells, treated with Pam2i 2CM, exhibit similar levels of apoptosis as their parental,nontransfected cell line.4 This suggests that while reducingBcl-2 levels and increasing the Bax:Bax dimer ratio in-creases the production of the apoptotic activity, the addi-tion of Bcl-2 (and the possible formation of Bcl-2:Bcl-2dimers) does not slow or stop this process. This agrees withthe current notion that Bcl-2 homodimers may not be ac-tive, and the ratio of Bcl-2:Bax to Bax:Bax dimers is thedetermining factor in the death decision (reviewed in Ref.49). However, mechanisms of bc/-2-independent apoptosishave been demonstrated, suggesting redundant inhibitors

for the death process. Therefore, the bcl-2 pathway mightnot be the only inhibitory pathway of apoptosis in thesecells.

For cells to die, a series of events is required. Threemechanisms are proposed for PCD and apoptosis, involvingdifferent cascades. The induction mechanism depends onnew gene expression after exposure to the apoptotic stim-ulus. In the release mechanism, the suicide program isconstitutively expressed but inhibited by a factor(s) withshort half-life. It is suggested that the release mechanism isused by cells that under physiological conditions can giverise to cells with short life expectancies, like blood cells andkeratinocytes (9). In the transduction mechanism cells, mayhave all the necessary apoptotic molecules that await acti-vation by a transmembrane signal. The prototype of thetransduction mechanism is the rapid death of tumor targetcells upon interaction with cytotoxic (killer) cells.

Using CHX, it was demonstrated that: (a) keratinocytepreparations enriched for basal cells undergo apoptosis viathe release mechanism; and (b) keratinocyte preparationsenriched for granular cells do not respond to CHX with achange in apoptosis levels (38). We had shown thatPam212 cells have properties of a subpopulation of thebasal keratinocytes, already committed for terminal differ-entiation and ready for the transition to the suprabasallayers (40). Here we show that CHX did not change thelevel of apoptosis in Pam2i2 cells, suggesting that theyundergo apoptosis via the transduction mechanism. Alter-natively, we suggest a different mechanism, in which theinternalization of a secreted serine protease induces apop-tosis by a proteolytic reaction. This mechanism is indistin-guishable from the transduction mechanism with respect tothe CHX response. We propose that in vivo, the basalkeratinocytes continuously express the death machinery,which is inhibited by a repressor (“release” mechanism;Ref. 38). At this stage, the basal cells express high levels ofBcl-2 (Ref. 40 and references therein). At the time of com-mitment for terminal differentiation (basal-spinous transi-tion), the level of Bcl-2 is dramatically reduced (40), sug-

814 Serine Protease Induces Apoptosis in Keratinocytes

gesting that Bcl-2 might be one ofthe repressors ofthe deathpathway in basal keratinocytes. Following the down-regulation of bcl-2, the death machinery could be in-duced by the serine protease activity. Once induced, thetransition from basal to spinous cells is completed, andthe cells can continue their terminal differentiation pro-cess. This two-step process, the elimination of a repressorand the later activation of the death pathway, ensuresthat the timing of the cell death is highly regulated.

Pam2l 2 CM can induce apoptosis in many cell lines thatvary in their levels of Bcl-2 protein. We see no correlationbetween Bcl-2 levels and the ability of the Pam2l 2 CM toinduce apoptosis. Moreover, 3T3 cells do not express theBcl-2 protein (Refs. 50-52), demonstrating that the activityof the CM is bcl-2 independent. Pam2i 2 cells represent amixed population of attached, nonapoptotic cells, express-ing bd-2 and apoptotic cells, which are de-repressed andare undergoing autocnine PCD induction. Although the pro-duction and/or processing of the senine protease might becontrolled via the bcl-2 pathway, the senine protease induc-tion of PCD in vivo is expected to be bcl-2 independent.

During evolution, the death program has been modifiedto accomplish a number of tissue-specific physiologicalfunctions. Lens epithelial cells, RBC, and terminally differ-entiated keratinocytes represent distinct and specializedforms of cell death, which share many elements with ap-optosis. However, the tissue-specific factors involved insuch “death specializations” are yet to be discovered. Aninteresting candidate for a keratinocyte-secreted senine pro-tease involved in epidermal differentiation is SCCE, whichis detected in high suprabasal keratinocytes of the epider-mis (53). It is proposed that this senine protease, expressedin keratinizing squamous epithelia, might play a role in the

terminal stages of epidermal turnover and desquamation.SCCE is expressed as an inactive pre-protein, and trypticdigestion can remove its pre-peptide to activate this senineprotease (53). Similar to SCCE, our senine protease apop-totic-inducing activity is induced with trypsin, inhibited bySTI, and is partially purified on an STI column. These bio-chemical properties, as well as the localization of SCCEwithin the epidermis (54), make SCCE an interesting can-didate for the activation of keratinocyte apoptosis. Alterna-lively, our senine protease could activate SCCE, leading tofurther progression in epidermal differentiation. Using aprimary keratinocyte culture system and a two-dimensionalgel analysis, at least 70 secreted proteins are detected in theCM (55). Many of these proteins are not yet characterized.Using a similar system and antibodies to SCCE, it might bepossible to identify whether this enzyme is secreted fromPam2i2 cells. Neutralizing mAbs could identify whetherSCCE is involved in the regulation of keratinocyte terminaldifferentiation by signaling apoptosis or whether it is acti-vated by our senine protease and the apoptotic pathway.

A mechanism for the induction of apoptosis by senineproteases that our data cannot rule out is the perform-granzyme mechanism used by cytotoxic lymphocytes. Thecombined effect of perform, a pore-forming protein, andgranzymes, a family of granule-proteins that includes manysenine proteases, leads to apoptosis and DNA fragmentationof the target cell (1 3-20). Although this mechanism is doc-umented in bone marrow-derived cells only, we cannotexclude it in our system. Moreover, at least one granzyme,TIA-1 , is expressed in the Pam2l 2 cells.4

We suggest that a protease-activated, receptor-mediatedsignal is involved in the induction of apoptosis in Pam2i 2

cells. Alternatively, a secreted senine protease that reentersthe cell could induce the keratinocyte cell death. This isbased on the induction of apoptosis in several cell lines bya partially purified activity that is blocked by serine proteaseinhibitors and on the minimal effect of protein synthesis

inhibition on Pam212 apoptosis. We excluded the pro-tease-activated receptors TR and PAR-2 and the uPA recep-tor as possible candidates. We propose that in vivo, atwo-step mechanism might control keratinocyte apoptosisand differentiation. The basal cells of the epidermis containall the necessary proteins required for apoptosis, as well asthe repressor protein Bcl-2. As Bcl-2 levels go down, thecells commit to terminal differentiation, and are ready toleave the basal layer. A senine protease secreted supraba-sally signals the induction of the death machinery. Thissecond step enables the cells to undergo apoptosis andcontinue the process of terminal differentiation.

Materials and Methods

Cells and Tissue Culture. Pam2i 2 cells were a generousgift of Dr. S. Yuspa (National Cancer Institute, Bethesda,MD). T-47D (a breast carcinoma cell line), HepC2 (a hep-atocellular carcinoma cell line), and NIH3T3 (a contactinhibited mouse embryonal cell line) cells were from Amen-can Type Culture Collection (Rockville, MD). Cell lines

were grown in DMEM (CIBCO-BRL, Caithersburg, MD)supplemented with 10% FCS (CIBCO-BRL) under standardtissue culture conditions. Cells were trypsmnized for 5 mmusing 0.25% trypsin-1 mtvi EDTA (CIBCO-BRL); Pam212cells were rinsed three times with PBS (CIBCO-BRL) beforetrypsinization. This short trypsinization results in no detect-able changes in apoptosis. Cells were plated at 3 x i0� or1 x 1 06 cells/i 0-cm plate or 5 X 10� cells/i -cm2 slidechamber.

Conditioned Media. Pam2i 2 and 3T3 cells, plated at1 06 cells/i 0-cm plate, were grown for 3 or 6 days in eitherserum-free media or supplemented with 1 0% FCS (CIBCO-BRL). Media were spun (600 X gfor 5 mm at 4#{176}C)to removecells and cell debris and were frozen in aliquots. The ac-tivity of the CM was not affected by freezing and thawing(data not shown). Each CM preparation was assayed for itsability to induce apoptosis in both Pam2i 2 and 3T3 cells bythe TUNEL and ladder assays.

TUNEL Assay. Apoptotic staining was performed usingApoptag (Oncor, Caithersburg, MD; manufacturer’s proto-col), a technique based on the labeling of fragmented DNAends (56). Each experiment was repeated at least threetimes. Pictures presented are of a single experiment. Statis-tical analysis included at least three experiments, assayingat least i 000 cells/experiment. Apoptotic cells were definedby both morphology (condensed or fragmented nuclei andcytoplasm) and staining (fragmented DNA within the con-densed nuclei). Statistical analysis (Student’s t test) wasperformed using StatView version 4.01 (Abacus Concepts,Inc., Berkeley, CA).

Ladder Assay. Ladder assays were performed as de-scnibed (40). Briefly, cells were plated at 1 x i0� cells/10-cm plate, and cells and media were collected after 3days in culture. Media were immediately spun (600 X gfor5 mm at 4#{176}C)to concentrate cells and cell debris, and DNAwas extracted. DNA equivalent to one-half of a 10-cm plateor one-half of the cells accumulated in 10 ml media wasloaded in each well. DNA was analyzed on 2% agarose/ethidium bromide gels (CIBCO-BRL).

Cell Growth & Differentiation 815

STI Column. Immobilized STI (STI-agarose; Pierce,Rockford, IL) was equilibrated with 0.1 M Tris-HCI (pH 7.4)and applied to the CM overnight at 4#{176}C,with gentle agita-tion. STI was washed with the same buffer, and fractions of1 ml were eluted with 0.1 M sodium acetate (pH 2.8).Fractions were immediately neutralized with 2.5 volumesof 5 M Tris-HCI (pH 8) and desalted. For biological assaysfractions, were dialyzed against DMEM using a Mr 10,000cutoff (Slide-a-lyzer cassettes; Pierce; manufacturer’sprotocol).

Inhibitors and Peptides. The protease inhibitors PMSF,DCI, TPCK, leupeptin, and calpain I inhibitors were fromBoehninger-Mannheim (Indianapolis, IN). Trypsin wasfrom CIBCO-BRL (Caithersburg, MD), and soybean tryp-sin inhibitor was from Calbiochem (San Diego, CA).Hirudin, uPA, and CHX were from Sigma Chemical Co.(St. Louis, MO). Plasmmnogen activator inhibitor-I andanti-mouse uPA antibody were from American Diagnos-tica, Inc. (Creenwich, CT). The peptides SFLLRN, TFRIFD,SLIGRL, and FSLLRN were synthesized at Robert WoodJohnson Pharmaceutical Research Institute (La Jolla, CA).

RT-PCR. DNased-RNA (RQ1 RNase-free DNase; Pro-mega, Madison, WI; manufacturer’s protocol) from Pam2i 2and ASbcl-2 cells and from a whole mouse embryo (1 3.5days) was reverse transcribed (Superscript reverse tran-scniptase; CIBCO-BRL; manufacturer’s protocol) using ran-dom hexamers (CIBCO-BRL). RT products ofSO ng RNA (20ng for C3PDH) were PCR amplified (Taq polymerase;Perkin Elmer Cetus, Norwalk, CT; manufacturer’s protocol)using the thrombin receptor primers CCTCTCACTCCCA-GAGGTACGTCTACAG (5’) and CCTAACTTAACACCT-TTTTGTATATGCTGTTATTCAGG (3’) at 30 cycles of 30 sat 94#{176}C,60 s at 50#{176}C,and 3 mm at 72#{176}C.The PAR-2 primersGGGAAAGGGG1TCCGGTAGAACCAGGC1TFTCC (5’)and CGCCAACGGCGATG1TTGCCTTCTTCCTCCCC (3’)were used at 30 cycles of 30 s at 94#{176}C,60 s at 55#{176}C,and 3mm at 72#{176}C.C3PDH primers and PCR conditions werefrom Clontech (Palo Alto, CA; manufacturer’s protocol).Ten % of the reaction products were analyzed on 2%agarose/ethidium bromide gels (CIBCO-BRL).

AcknowledgmentsWe thank Dr. S. Yuspa for Pam2l 2 cells, Dr. G. Taylor for oligonucleotidesynthesis, and Drs. D. Johnson, L. Lawrence, S. Prouty, C. Scherczinger,K. Stenn, and R. Zivin for fruitful discussions throughout this study and forcritically reading this manuscript.

References1. Cohen, J. J. Apoptosis. Immunol. Today, 14: 126-130, 1993.

2. Schwartzman, R. A., and Cidlowski, J. A. Apoptosis: the biochemistry andmolecular biology of programmed cell death. Endocrinol. Rev., 14:133-151, 1993.

3. Wyllie, A. H. Apoptosis (The 1992 Frank Rose Memorial Lecture). Br. J.

Cancer, 67: 205-208, 1993.

4. Kerr, J. F. R., Wyllie, A. H., and Currie, A. R. Apoptosis: a basic biologicalphenomenon with wide ranging implications in tissue kinetics. Br. J. Cancer,26:239-257, 1972.

5. Kerr, I. F. R., and Harmon, B. V. Definition and incidence of apoptosis: anhistorical perspective. In: Apoptosis: The Molecular Basis for Cell Death, pp.5-30. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory, 1991.

6. Fesus, L. Biochemical events in naturally occurring forms of cell death.Fed. Eur. Biochem. Soc., 328: 1-5, 1993.

7. Oberhammer, F., Wilson, J. W., Dive, C., Morris, I. D., Hickman, J. A.,Wakeling, A. E., Walker, P. R., and Sikorska, M. Apoptotic death in epithelial

cells: cleavage of DNA to 300 and!or 50 kb fragments prior to or in theabsence of internucleosomal fragmentation. EMBO J, 12: 3679-3684, 1993.

8. Walker, P. R., Weaver, V. M., Lach, B., LeBlanc, J., and Sikorska, M.Endonuclease activities associated with high molecular weight and internu-cleosomal DNA fragmentation in apoptosis. Exp. Cell Res., 213: 100-106,1994.

9. Owens, G. P., and Cohen, J. J. Identification of genes involved inprogrammed cell death. Cancer Metastasis Rev., 11: 149-156, 1992.

10. Williams, G. T., Smith, C. A., McCarthy, N. J., and Grimes, E. A.Apoptosis: final control point in cell biology. Trends Cell Biol., 2: 263-267,1993.

1 1 . Martin, S. Apoptosis: suicide, execution or murder? Trends Cell Biol., 3:141-144, 1993.

1 2. Reed, I. C. Bcl-2 and the regulation of programmed cell death. J. CellBiol., 142: 1-6, 1994.

1 3. Russell, 1. H. Internal disintegration model of cytotoxic lymphocyte-induced target damage. Immunol. Rev., 72: 97-1 01 , 1983.

14. Hayes, M. P., Berrebi, G. A., and Henkart, P. A. Induction oftarget cellDNA release by the cytotoxic I lymphocyte granule protease granzyme A. J.Exp. Med., 170:933-946, 1989.

15. Tian, Q., Streuli, M., Saito, H., Schlossman, S. F., and Anderson, P. Apolyadenylate binding protein localized to the granules of cytolytic lympho-cytes induces DNA fragmentation in target cells. Cell, 67: 629-639, 1991.

16. Shi, L., Kraut, R. P., Aebersold, R., and Greenberg, A. H. A natural killer

cell granule protein that induces DNA fragmentation and apoptosis. J. Exp.Med., 175:553-566, 1992.

1 7. Shi, L., Kam, C-M., Powers, J. C., Aebersold, R., and Greenberg, A. H.Purification of three cytotoxic lymphocyte granule serine proteases thatinduce apoptosis through distinct substrate and target cell interactions. I. Exp.Med., 176:1521-1529, 1992.

18. Kawakami, A., Tian, Q., Duan, x., Streuli, M., Schlossman, S. F., andAnderson, P. Identification and functional characterization ofa TIA-1 -related

nucleolysin. Proc. NatI. Acad. Sci. USA, 89: 8681-8685, 1992.

19. Shiver, J. W., Lishan, S., and Henkart, P. A. Cytotoxicity with target DNAbreakdown by rat basophilic leukemia cells expressing both cytolysin andgranzyme A. Cell, 171: 315-322, 1992.

20. Sarin, A., Adams, D. H., and Henkart, P. A. Protease inhibitors selec-tively block T cell receptor triggered programmed cell death in a murine T

cell hybridoma and activated peripheral T cells. J. Exp. Med., 178:1639-1700, 1993.

21 . Yuan, j., Shaham, S., Ledoux, S., Ellis, H. M., and Horvit.z, H. R. The C.elegans cell death gene ced-3 encodes a protein similar to mammalianinterleukin-1�-converting enzyme. Cell, 75: 641-652, 1993.

22. Hogquist, K. A., Nett, M. A., Unanue, E. R., and Chaplin, D. D. Inter-leukin 1 is processed and released during apoptosis. Proc. NatI. Acad. Sci.

USA, 88: 8485-8489, 1991.

23. Miura, M., Zhu, H., Rotello, R., Hartwieg, E. A., and Yuan, J. Inductionof apoptosis in uibroblasts by IL-i � converting enzyme, a mammalianhomolog of the C. elegans cell death gene ced-3. Cell, 75: 653-660, 1993.24. Kumar, S., Kinoshita, M., Noda, M., Copeland, N. G., and Jenkins, A.Induction of apoptosis by the mouse Nedd2 gene, which encodes a proteinsimilar to the product of the C. elegans cell death gene ced-3 and themammalian IL-i f3 converting enzyme. Genes Dev., 8: 1613-1626, 1994.

25. Wang, L., Miura, M., Bergeron, L., Zhu, H., and Yuan, J. lch-1, anIce/ced-3-related gene, encodes both positive and negative regulators ofprogrammed cell death. Cell, 78: 739-750, 1994.

26. Lazebnik, Y., Kaufmann, S., Desnoyers, S., Poirier, G., and Earnshaw,W. Cleavage of poly (ADP-ribose) polymerase by a proteinase with proper-ties like ICE. Nature (Lond.), 371: 346-347, 1994.

27. Fernandes-Alnemri, T., Litwack, G., and Alnemri, E. S. CPP32, a novelhuman apoptotic protein with homology to C. elegans cell death proteinCed-3 and mammalian interleukin-i �3 converting enzyme. J. Biol. Chem.,269: 30761-30764, 1994.

28. Squier, M. K., Miller, A. C., Malkinson, A. M., and Cohen, J. J. Calpainactivation in apoptosis. J. Cell Physiol., 159: 229-237, 1994.

29. Ruggiero, V., Johnson, S. E., and Baglioni, C. Protection from tumornecrosis factor cytotoxicity by protease inhibitors. Cell Immunol., 107:317-325, 1987.

30. Suffys, P., Beyaert, R., Van Roy, F., and Fiers, W. Involvement of a serineprotease in tumor necrosis factor mediated cytotoxicity. Eur. J. Biochem.,178: 257-265, 1988.

31 . Weaver, V. M., Lach, B., Walker, P. R., and Sikorska, M. Role ofproteolysis in apoptosis: involvement of serine proteases in internucleosomalDNA fragmentation in immature thymocytes. Biochem. Cell Biol., 71:488-500, 1993.

32. Budtz, P. E. Epidermal homeostasis: a new model that includes apop-tosis. In: L. D. Tomei and F. 0. Cope (eds.), Apoptosis II: The Molecular Basis

816 Serine Protease Induces Apoptosis in Keratinocytes

of Apoptosis in Disease, pp. 165-i 84. Cold Spring Harbor, NY: Cold SpringHarbor Laboratory, 1994.

33. Watt, F. M. Terminal differentiation of epidermal keratinocytes. Curr.

Opin.Cell Biol., 1:1107-1115, 1989.

34. Fuchs, E. Epidermal differentiation: the bare essentials. J. Cell Biol., 111:2807-2814, 1990.

35. Fuchs, E. Epidermal differentiation. Current Opin. Cell Biol., 2:

1028-1035, 1990.

36. Fuchs, E. Epidermal differentiation and keratin gene expression., J. CellSci. Suppl., 17: 197-208, 1993.

37. Lovas, I. G. L. Apoptosis in human epidermis: a postmortem study bylight and electron microscopy. Australas J. Dermatol., 27: 1-5, 1986.38. McCall, C. A., and Cohen, j. J. Programmed cell death in terminallydifferentiating keratinocytes: role of endogenous endonuclease. j. Invest.

Dermatol., 97:111-114, 1991.

39. Yuspa, S., Hawley-Nelson, P., Loehler, B., and Stanley, R. A survey oftransformation markers in differentiation of epidermal cell lines in culture.CancerRes., 40:4694-4703, 1980.

40. Marthinuss, J., Lawrence, L., and Seiberg, M. Apoptosis in Pam2i 2, anepidermal keratinocyte cell line: a possible role for bcl-2 in epidermaldifferentiation. Cell Growth & Differ., 6: 239-250, 1995.

41 . Vu, T-K. H., Hung, D. T., Wheaton, V. I., and Coughlin, S. R. Molecularcloning of a functional thrombin receptor reveals a novel proteolytic mech-anism of receptor activation. Cell, 64: 1057-i 068, 1 99i.

42. Nystedt, S., Emilsson, K., Wahlestedt, C., and Sundelin, J. Molecularcloning of a potential proteinase activated receptor. Proc. NatI. Acad. Sci.USA, 91:9208-9212, 1994.

43. Gerszten, R. E., Chen, J., lshii, M., Ishii, K., Wang, L., Nanevich, T.,Turck, C. W., Vu, T-K H., and Coughlin, S. R. Specificity of the thrombinreceptor for agonist peptide is defined by its extracellular surface. Nature(Lond.), 368: 648-651, 1994.

44. Vassalli, J-D. The urokinase receptor. Fibrinolysis, 8(Suppl. 1): 1 72-i 81,1994.

45. Thornberry, N. A., Bull, H. G., Calaycay, J. R., et a!. A novel het-erodimeric cysteine protease is required for interleukin-1 �3 processing inmonocytes. Nature (Lond.), 356: 768-774, 1992.

46. Li, P., Allen, H., Banerjee, S., et al. Mice deficient in IL-i )3-convertingenzyme are defective in production of mature IL-i j3 and resistant to endo-toxic shock. Cell, 80: 40i-4i i, i995.

47. Peitsch, M. C., Polzar, B., Stephan, H., Crompton, T., MacDonald, H. R.,Mannherz, H. G., and Tschopp, J. Characterization of the endogenousdeoxyribonuclease involved in nuclear DNA degradation during apoptosis

(programmed cell death). EMBO J., 12: 371-377, i993.

48. Vaux, D. L., Haecker, G., and Strasser, A. An evolutionary perspective

of apoptosis. Cell, 76: 777-779, 1994.

49. Korsmeyer, S. J. Regulators of cell death. Trends Genet., 11: ioi-i05,1995.

50. Reed, J. C., Talwar, H. S., Cuddy, M., Baffy, G., Williamson, J., Rapp, U.R., and Fisher, G. J. Mitochondrial protein p26 Bcl-2 reduces growth factorrequirements of NIH3T3 fibroblasts. Exp. Cell Res., 195:277-283, 1991.

51 . Kitada, S., Miyashita, T., Tanaka, S., and Reed, J. C. Investigations ofantisense oligonucleotides targeted against bcl-2 RNAs. Antisense Res. Dev.,3: i57-i69, i993.

52. Martin, J. M., Veis, D., Korsmeyer, S. J., and Sugden, B. Latent membraneprotein of Epstein-Barr virus induces cellular phenotypes independently ofexpression of bcl-2. J. Virol., 67: 5269-5278, 1993.

53. Hansson, L., Stromqvist, M., Backman, A., Wallbrandt, P., Carlstein, A.,

and Egelrud, T. Cloning, expression and characterization of stratum corneumchymotryptic enzyme: a skin-specific human serine proteinase. J. Biol.Chem., 269: i9420-i9426, 1994.

54. Sondell, B., Thornell, L-E., Stigbrand, T., and Egelrud, T. Immunolocal-ization of stratum corneum chymotryptic enzyme in human skin and oral

epithelium with monoclonal antibodies: evidence of a protease specificallyexpressed in keratinizing squamous epithelia. J. Histochem. Cytochem., 42:459-465, 1994.

55. Katz, B. A., and Taichman, L. B. Epidermis as a secretory tissue: an invitro tissue model to study keratinocyte secretion. J. Invest. Dermatol., 102:55-60, 1994.

56. Gavrieli, Y., Sherman, Y., and Ben-Sasson, S. A. Identification of pro-grammed cell death in situ via specific labeling of nuclear DNA fragmenta-tion. J. Cell Biol., 119:493-501, 1992.